s ICSPCC


KEYNOTE SPEAKERS
Keynote Speaker 1
Time: 09:30~10:30, Saturday, August 22, 2020
Session Chair:  Jianguo Huang, Northwestern Polytechnical University
 
Wireless AI:

A New Sixth Sense to Deciphering our World

K. J. Ray Liu
Electrical and Computer Engineering Department
University of Maryland, College Park; and Origin Wireless
Abstract:
What smart impact will future 5G and IoT bring to our lives? Many may wonder, and even speculate, but do we really know? With more and more bandwidth readily available for the next generation of wireless applications, many more smart applications/services unimaginable today may be possible. In this talk, we will show that with more bandwidth, one can see many multi-paths, which can serve as hundreds of virtual antennas that can be leveraged as new degrees of freedom for smart life. Together with the fundamental physical principle of time reversal to focus energy to some specific positions and the use of machine learning, a revolutionary wireless AI platform can be built to enable many cutting-edge IoT applications that have been envisioned for a long time, but have never been achieved. We will show the world’s first ever centimeter-accuracy wireless indoor positioning systems that can offer indoor GPS-like capability to track human or any indoor objects without any infrastructure, as long as WiFi/LTE/5G is available. Such a technology forms the core of a wireless AI platform that can be applied to device-free non-obtrusive home/office monitoring/security, radio human biometrics, vital signs detection, sleep monitoring, and fall detection. In essence, in the future of wireless world, communication, as we see it, will be just a small component of what’s possible. There are many more magic-like smart applications that can be made possible by the emerging field of wireless AI, allowing us to decipher our surrounding world with a new “sixth sense”. Some demo videos will be shown to illustrate the future of smart radios for smart life. 
BIOGRAPHY
K. J. Ray Liu is a Distinguished University Professor and a Distinguished Scholar-Teacher of University of Maryland, College Park, where he is Christine Kim Eminent Professor of Information Technology. He is the founder of Origin Wireless, Inc., a high-tech start-up pioneering wireless AI for smart life. Dr. Liu was a recipient of the 2016 IEEE Leon K. Kirchmayer Award on graduate teaching and mentoring, IEEE Signal Processing Society 2014 Society Award for “influential technical contributions and profound leadership impact”, IEEE Signal Processing Society 2009 Technical Achievement Award, and more than a dozen best paper awards. Recognized by Web of Science as a Highly Cited Researcher, he is a Fellow of IEEE, AAAS, and US National Academy of Inventors. As the founder of Origin Wireless, his inventions won 2017 CEATEC Grand Prix Award and CES 2020 Innovation Award, with products available over 150 countries worldwide. He also received teaching and research recognitions from University of Maryland including university-level Invention of the Year Award (three times), and college-level Poole and Kent Senior Faculty Teaching Award, Outstanding Faculty Research Award, and Outstanding Faculty Service Award, all from A. James Clark School of Engineering (each award honors one faculty per year from the entire college). Dr. Liu is a candidate of IEEE President-Elect. He was IEEE Vice President, Technical Activities, Division IX Director of IEEE Board of Director, President of IEEE Signal Processing Society, where he has served as Vice President - Publications and Editor-in-Chief of IEEE Signal Processing Magazine.
Keynote Speaker 2
Time: 10:30~11:30, Saturday, August 22, 2020
Session Chair:  Fen Hou, Macau University
Reinforcement Learning for Resource Management in Space-Air-Ground Integrated Vechicular Network
Xuemin (Sherman) Shen
Department of Electrical and Computer Engineering,
University of Waterloo, Canada
Abstract
Modern society depends on faster, safer, and environment friendly transportation system. Vehicular communications network in terms of vehicle to vehicle, vehicle to infrastructure, vehicle to pedestrian, vehicle to cloud, and vehicle to sensor, can provide a solution to such transportation system. In this talk, we first introduce the all connected vehicles. We then present the applications, challenges and scientific research issues of vehicular communications network. We also explain the role of vehicular networking in the automated driving era. We conclude the talk by discuss the future Space-Air-Ground (SAG) Integrated vehicular networks.
BIOGRAPHY
Xuemin (Sherman) Shen is a University Professor and Associate Chair for Graduate Studies, Department of Electrical and Computer Engineering, University of Waterloo, Canada. Dr. Shen is also the Fellow of IEEE, Engineering Institute of Canada, Canadian Academy of Engineering, and Royal Society of Canada. Dr. Shen’s research focuses on wireless resource management, wireless network security, wireless body area networks, smart grid and vehicular ad hoc and sensor networks. He is the Editor-in-Chief of IEEE IoT Journal. He serves as the General Chair for Mobihoc’15, the Technical Program Committee Chair for IEEE GC’16, IEEE Infocom’14, IEEE VTC’10, the Symposia Chair for IEEE ICC’10, the Technical Program Committee Chair for IEEE Globecom’07, the Chair for IEEE Communications Society Technical Committee on Wireless Communications. Dr. Shen is the elected VP Publication of IEEE ComSoc, and was the chair of IEEE ComSoc Distinguish Lecturer selection committee and a member of IEEE ComSoc Fellow evaluation committee. Dr. Shen received the Excellent Graduate Supervision Award in 2006, and the Premier’s Research Excellence Award (PREA) in 2003 from the Province of Ontario, Canada. Dr. Shen is a Distinguished Lecturer of IEEE Vehicular Technology Society and Communications.

ICSPCC 2020 Conference Program
21 Aug (Fri) 
6:00 pm - 7:30 pm 		 Welcoming and  Testing Session
22 Aug (Sat)
9:00 am - 12:30 pm		 Opening and Keynotes
Paper Presentation Track 1 Track 2 Track 3
22 Aug (Sat)
13:30-18:30		 STPC SPG01  SPG02
23 Aug (Sun)
09:00-13:00 pm		 COM01 SPG03 CPT
23 Aug (Sun) 
13:30-18:30 		 COM02 SPG04 　
COM - Communications Session
CPT - Computing Session
SPG - Signal Processing Session
STC - STudent Paper Contest Session
Presentation Session - Each paper should have a 20 minutes time-slot for the author to present his paper as scheduled in this Program.   Authors are requested to prepare a 15 minutes video script which should spell out the theme, the research method, the advantage, possible drawback and any future works.  Presentations should be in English.  Then followed by a 5 minutes Q&A.
Welcoming and Testing Session - for virtual conference test trial with the OC Chairs and any questions raised. 


Presentation Session
Track 1: - STPC Session, 22 Aug (Sat) 13:30-18:30 
Session: - STPC-01 (Paper No.2121)
Paper Title: - A Highly Parallel Constant-Time Almost-Inverse Algorithm
Authors: - Daniele Venier and Ray C. C. Cheung
Abstract: - In this paper, we present a highly parallel and area-efficient constant-time inversion algorithm over the r-th degree polynomial ring, derived from Schroeppel's Almost Inverse algorithm. We propose a first constant time version, from which we derive a highly-parallel and a faster algorithm, while still preserving the constant-time property.
This constitutes an alternative and relatively unexplored approach to inversion, compared to the more common multiplicative approach by Itoh and Tsuji, and has extensive application in algorithms such as the BIKE proposal for quantum-resistant cryptography.  Our approach is extremely area-efficient, with a constant area with respect to the polynomial degree r.
Session: - STPC-02 (Paper No.2144)
Paper Title: - A Novel DNN Based Channel Estimator for Underwater Acoustic Communications with IM-OFDM
Authors: - Mingzhang Zhou, Junfeng Wang, Haixin Sun, Jie Qi, Xiao Feng and Hamada Esmaiel
Abstract: - Performance of acoustic communication system in shallow sea is influenced by complicated interferences. Multipath with large delays and strong reflections leads to serious transmission error. To support reliable and efficient transmission in the background above, in this paper, a deep learning based channel estimator for underwater index modulated OFDM is proposed. A deep neural network is designed and trained with real channels tested in Xiamen sea area. The extracted real channels are collected and analyzed, constituting a mixed database with the channels generated using real parameters. Via a half-physical simulation, the performance is evaluated with different channel estimators. The results prove stability of the performance with the proposed channel estimator in different communication distances of shallow water. In conclusion, the deep learning based underwater IM-OFDM channel estimator obtains significant performance in target shallow sea scenarios, which is promising as a solution to the adaptive scheme in changing underwater environment.
Session: - STPC-03 (Paper No.2147)
Paper Title: - Time Synchronization and Clock Parameter Estimation for Wireless Sensor Networks with Unequal Propagation Delays
Authors: - Guopeng Liu, Shefeng Yan, Linlin Mao and Zeping Sui
Abstract: - Time synchronization is a prerequisite and also a critical support technique for wireless sensor networks (WSNs). Traditional synchronization protocols have some drawbacks, such as limited synchronization accuracy, high energy consumption, and strict node deployment requirement, which limits their application in underwater WSNs. In view of the shortcomings of the traditional schemes, we propose an energy-efficient synchronization protocol which explores the nature of broadcasting to save energy. Based on the proposed protocol, the minimum variance unbiased estimation (MVUE) of the clock offsets is further derived to obtain higher synchronization accuracy. Remarkably, the proposed protocol can be applied to randomly deployed WSNs with unequal propagation delays, especially underwater WSNs. Simulation results confirm the superiority of the proposed protocol in both terrestrial and underwater WSNs.
Session: - STPC-04 (Paper No.2150)
Paper Title: - Multi-Channel Audio Statistical Restoration
Authors: - Yinhong Liu and Simon Godsill
Abstract: - Phonograph record is an analogue sound storage medium that has played an important part in sound history. Modulated spiral groves on the disc are usually inevitably damaged by scratches and dust, which lead to noises with different statistical characteristics. Among those different type of noises, there is one common click-shape noise that have some degree of correlation between channels. Its origins have been explained as a deviation of the probe needle means it shifts closer to one wall but further from another with similar distances. Our aim is to restore this type of noises on dual mono audio signals with a statistical method. There are previous researches on restoring single-channel noisy data based on generative models. However multi-channel noisy data contain more audio information than single-channel data, due to its channel correlation. To exploit this property, we propose multivariate Gaussian models for both signal and noise models. Then we derive the Maximum A Posteriori(MAP) estimation of the underlying data. Additionally, we also use the Maximum Likelihood method to optimize model parameters. In the end we compare the restoration performances between our model and the baseline model on both synthetic and real-life noisy music data.
Session: - STPC-05 (Paper No.2164)
Paper Title: - Restoration of sparse multispectral single photon LiDAR data
Authors: - Songmao Chen, Wei Hao, Xiuqin Su and Zhenyang Zhang
Abstract: - This paper presents a new algorithm to restore the sparse multispectral data acquired by single photon LiDAR. Regarding to its sparsity, the robustness of depth estimation is improved by exploiting the correlations between wavelengths and multi-scale information. Furthermore, the non-local spatial correlations between pixels/patches are learned by the affinity graph and used as prior information to prompt the smoothness for both depth and reflectivity images.
To reduce the computational cost, a non-uniform sampling algorithm and clustering strategy are adopted, where flexible sampling points were assigned to reduce the size of the graph while minimizing the loss in details and different spatial clusters can be processed in parallel. Finally, the restoration were achieved by optimizing a cost function which account for both multi-scale information and non-local spatial correlations. The cost function is efficiently solved by ADMM algorithm that present fast convergence.
Results on simulated data showed the benefits of the proposed algorithm by comparing with state-of-the-art algorithms, and the fast computation  justified the supreme performance.
Session: - STPC-06 (Paper No.2181)
Paper Title: - One-Step Backtracking Algorithm Based on Viterbi Algorithm in GMSK Demodulation
Authors: - Doudou Song, Rugui Yao, Huaihai Ma, Xiaoya Zuo, Ye Fan and Juan Xu
Abstract: - Viterbi algorithm is a commonly used algorithm for Gaussian filtered minimum frequency shift keying (GMSK) demodulation, but there exist some problems such as large
delay, high overhead, and data overflow in the hardware implementation. Aiming at the problems of delay and overhead, we propose a novel update rule to optimize the stored transfer-state information table and realize one-step backtracking in this paper.
It also ensures the realization of pipeline operation, reduces demodulation delay, and saves hardware resources. Furthermore, we adopt combinational logic to perform pre-decision, which not only satisfies the timing requirements, but also achieves the antioverflow.
The hardware implementation results demonstrate the feasibility and correctness of the design.
Session: - STPC-07 (Paper No.2186)
Paper Title: - VARIANCE REDUCED DIFFUSION ADAPTATION FOR ONLINE LEARNING OVER NETWORKS
Authors: - Mengfei Zhang, Danqi Jin and Jie Chen
Abstract: - The stochastic variance reduced gradient (SVRG) algorithm has shown its effectiveness in accelerating the convergence of stochastic gradient algorithms. Considering the emergent applications of distributed
estimation, it is interesting to investigate the way to adapt this algorithm to distributed learning with streaming data. For this purpose, in this work we first propose a time-averaging SVRG algorithm
that fits into the context of streaming data processing. Then, we integrate this algorithm with the diffusion adaptation to enhance the performance of distributed estimation over networks. Theoretical analysis of the resulted algorithm is conducted to characterize its stability. We also provide the simulation results to illustrate its favorable performance.
Session: - STPC-08 (Paper No.2218)
Paper Title: - Convolution-Enhanced LSTM Neural Network Post-Equalizer used in Probabilistic Shaped Underwater VLC System
Authors: - Zhongya Li, Fangchen Hu, Guoqiang Li, Peng Zou, Chaofan Wang and Chi Nan
Abstract: - We proposed a convolution-enhanced long short-term memory (CE-LSTM) neural network as the equalizer to overcome severe linear and nonlinear distortion in underwater visible light communication (UVLC) system with probabilistic shaping technologies. By comparing the proposed equalizer with the traditional Volterra-series based equalizer and normal LSTM-based equalizer, we demonstrated the efficiency and performance advantages of using a CE-LSTM based equalizer. To the best of our knowledge, this is the first practice to combine convolution network with LSTM in equalization scheme of UVLC and the highest achievable information rate (AIR) of 3.53 Gbit/s is achieved for single LED-based UVLC with PS.
Session: - STPC-09 (Paper No.2234)
Paper Title: - Multi-head Attention Networks for Nonintrusive Load Monitoring
Authors: - Nan Lin, Binggui Zhou, Guanghua Yang and Shaodan Ma
Abstract: - In this paper, we proposed two multi-head attention neural networks for Nonintrusive Load Monitoring (NILM). The proposed networks are more suitable for the processing of sequential data by implementing the attention mechanism to learn the complex patterns and long-term dependencies. Compared with existing neural NILM schemes, the proposed multi-head attention networks achieve better disaggregation accuracy for different domestic appliances, are more robust to the dynamics of the aggregated data and more efficient for training.
Session: - STPC-10 (Paper No.2240)
Paper Title: - High Resolution Localization and Vertical Angle Scattering Spectrum Reconstruction of Targets in Shallow Water Waveguide
Authors: - Xiaomeng Fan, Ting Zhang and Jingning Jiang
Abstract: - In complex ocean environment, reconstructing low-frequency scattering pattern (frequency-angle spectrum) of underwater targets can provide further powerful support for target classification. This paper describes the reconstruction of target scattering pattern with unknown position, from received data on a vertical line array (VLA) in shallow water waveguide. The point spread function (PSF) in conventional matching field processing (CMFP) does not have global strict shift invariance. This paper proves that with the same effective mode number, the PSF retains the shift invariant property and can thus be deconvolved from CMFP by Richardson-Lucy (R-L) algorithm. The proposed deconvolved matched field processing (DMFP) can achieve high resolution along both range and depth direction and low side lobes. With the estimated target range and depth, and the estimated mode amplitude at each frequency by mode decomposition, the broadband target scattering spectrum with respect to vertical angle is accurately reconstructed with moderate number of modes.
Track 2: - SPG01  Session, 22 Aug (Sat) 13:30-18:30 
Session: - SPG01-01 (Paper No.2090)
Paper Title: - GEV Beamforming with BAN Integrating LPS Estimation and Post-filtering
Authors: - Shuhao Deng, Changchun Bao and Rui Cheng
Abstract: - Beamforming method can effectively remove background noise, even in the complex environment, so it is widely used in speech enhancement. We propose a novel Generalized Eigenvalue (GEV) beamforming with Blind Analytic Normalization (BAN) method. In this method, the GEV beamformer coefficients are constructed by estimating logarithmic power spectrum (LPS), which are used to filter multichannel speech signals, and post filter technology is used to further remove noise in the beamformed signals. Firstly, in order to estimate the LPS of speech signal in each channel, we use the data-driven method to train the deep neural network (DNN) model. Then, we use the well trained DNN model to estimate LPS, which is used to calculate the power spectral density (PSD) matrix of speech, and further obtain the coefficients of the GEV beamformer. Since the GEV beamformer will cause speech distortion, the BAN is employed to post-process the beamformed signal. Furthermore, single channel speech enhancement is used to reduce residual noise. Our experiment is conducted in 8-channel simulation data set. The experimental results show that, compared with some existing speech enhancement methods, the proposed method can effectively remove background noise and achieve better speech enhancement effect.
Session: - SPG01-02 (Paper No.2092)
Paper Title: - Multi-channel Speech Enhancement Based on the MVDR Beamformer and Postfilter
Authors: - Dujuan Wang and Changchun Bao
Abstract: - Deep neural network (DNN) based ideal ratio mask (IRM) estimation methods have yielded good performance in monaural speech enhancement. Meanwhile, these methods have also shown considerable potential for beamforming and multichannel speech enhancement. It is crucial for minimum variance distortionless response (MVDR) beamformer to estimate the covariance matrix of the speech and noise accurately. The accurate estimation of time-frequency (T-F) mask has significant impact on the estimation of the covariance matrices. So, in this paper, a complex real and imaginary ratio mask (CRIRM) based MVDR beamformer for speech enhancement using residual network is proposed. First, the real and imaginary masks of speech and noise are estimated by taking advantage of a residual neural network. After that, the estimations of speech and noise are obtained by using the estimated masks. Finally, the covariance matrices of speech and noise are estimated, and applied into the MVDR beamformer. In addition, in order to further reduce residual noise interference, the output of the MVDR beamformer is further processed by an end-to-end monaural speech enhancement module. Experiments show that, the proposed method can better improve the quality and intelligibility of the enhanced speech.
Session: - SPG01-03 (Paper No.2094)
Paper Title: - A Weekly Supervised Speech Enhancement Strategy using Cycle-GAN
Authors: - Yang Xiang, Changchun Bao and Jing Yuan
Abstract: - Nowadays, due to the application of deep neural network (DNNS), speech enhancement (SE) technology has been significantly developed. However, most of current approaches need the parallel corpus that consists of noisy signals, corresponding speech signals and noise on the DNNs training stage. This means that a large number of realistic noisy speech signals is difficult to train the DNNs. As a result, the performance of the DNNs is restricted. In this research, a new weakly supervised speech enhancement approach is proposed to break this restriction, using the cycle-consistent generative adversarial network (CycleGAN). There are two stage for our methods. In training stage, a forward generator is employed to estimate ideal time-frequency (T-F) mask and an inverse generator is utilized to acquire noisy speech magnitude spectrum (MS). Additionally, two discriminators are used to distinguish the real clean and noisy speech from generated speech, respectively. In enhancement stage, the T-F mask is directly estimated by using the well-trained forward generator for speech enhancement. Experimental results indicate that our strategy can not only achieve satisfied performance for non-parallel data, but also acquire the higher score in speech quality and intelligibility for the DNN-based speech enhancement using parallel data.
Session: - SPG01-04 (Paper No.2098)
Paper Title: - Prediction of NMF-based Wiener Filter for Speech Enhancement Using Deep Neural Networks
Authors: - Zhigang Bai, Changchun Bao and Zihao Cui
Abstract: - In this paper, a novel approach is presented to predict a training target called NMF-based Wiener filter using deep neural networks (DNN) in the nonnegative matrix factorization (NMF) based speech enhancement. The NMF-based Wiener filter, as a masking-based target, is easier than the encoding vectors used in previous algorithms for parameter estimation. The intermediate error of the NMF-based speech enhancement process was reduced due to direct prediction of the NMF-based Wiener filter. The encoding vectors of noisy speech were extracted with the NMF algorithm and normalized to obtain more discriminative input features. The DNN was trained to learn a nonlinear mapping from the encoding vector of noisy speech to the NMF-based Wiener filter. At test stage, the predicted NMF-based Wiener filter was used to enhance noisy speech. The objective evaluations demonstrated that the proposed algorithm outperforms some existing NMF-based and DNN-based methods at various input signal-to-noise ratio (SNR) levels.
Session: - SPG01-05 (Paper No.2116)
Paper Title: - Impact of Multimodal Cognitive Processes on Second Language Acquisition
Authors: - Yizhou Lan and Will Xiangyu Li
Abstract: - Second Language (L2) speech perception is believed to be a function of the perceived distance between the second and first languages (L1). Graphic influence on the orthography during the learning experience of a series of L1 sounds, though understudied, can play an equally important role in L1-L2 perceptual assimilation. The present study attests the hypothesis that early-age orthographic anchors, or inputs, can interplay with L1 phonology and affect the perception of L2 on a level parallel to phonetic percepts. We have recruited 70 English-learning subjects from China, whom had acquired Chinese through the pinyin Romanization system when started school, to perform a perceptual identification task of categorizing English phonemes as Mandarin ones. We have thus harvested participants’ identification mappings together with subjective appraisal of the goodness of mappings. The findings show that learners performed in certain tasks significantly inaccurate and yet highly confident in identifications where graphic confusions of English-Chinese existed, but not in others. And the effect is clearly separate from acoustic distances between L1-L2, affirming our hypothesis. The study has significant implications on the theorizing of human speech perception and language education policy.
Session: - SPG01-06 (Paper No.2131)
Paper Title: - Spoken Language Identification With Hierarchical Topic Model and Multi-levels Discriminative Cues
Authors: - Linjia Sun
Abstract: - The discriminative cues and effective representations are two important factors for spoken language identification (LID). In this paper, the multi-levels language cues, including phonetic, phonotactic and prosodic, are studied, and a hierarchical representation based on supervised topic model is further designed to organize the multi-levels cues. The difference of languages is distinguished by using the distribution of prosodic topics and phonemic topics in each language. We built the proposed LID method and report the test results on NIST LRE07 datasets. Compared with the most advanced LID methods, our proposed method shows the competitive performance in the LID task. In particular, our method helps to capture robust discriminative information for short duration language identification.
Session: - SPG01-07 (Paper No.2138)
Paper Title: - Research on Classification and Recognition of Underwater Targets Based on Spark's Decision Tree Technology
Authors: - Yue Sun, Yuan Peng and Guijuan Li
Abstract: - This paper proposes a Spark-based decision tree classification algorithm that effectively utilizes the memory-based distributed parallel processing capabilities of the Spark big data platform. For water and underwater targets, the algorithm's recognition accuracy is different when the method of measuring the impureness of the decision tree nodes, the maximum depth of the decision tree, and the maximum number of branches of each node of the decision tree are different. We can find the parameter combination with the highest recognition accuracy of the algorithm.
        First, we introduce operating mechanism of Spark and the principle of parallel computing. Then, we analyze the implementation basis of parallelization of the decision tree classification algorithm and use the big data-oriented decision tree parallel machine learning method in the Spark cloud computing environment to process underwater acoustic data features. Finally, under the condition of cross-validation, we give the influence of different combinations of the maximum depth of the decision tree and the maximum number of branches on each node on the recognition accuracy of underwater acoustic signals.
        This paper presents the test results of the algorithm on the recognition accuracy and running time under different parameter combinations. The results show that for the existing underwater acoustic data, when the information gain is used as the evaluation criterion, the maximum depth of the decision tree is 11 and the maximum number of each node branch is 8, the recognition accuracy is the highest. At this time, the recognition achieved on the training set is accurate. The rate is 87.9%, and the recognition accuracy on the test set is 84.9%. The experiments show that the overall classification recognition effect and operating efficiency have been significantly improved.
Session: - SPG01-08 (Paper No.2209)
Paper Title: - The Novel Improving Algorithms on DRA Audio Entropy Coding
Authors: - Jianxin Yan and Lei Wang
Abstract: - DRA (Digital Rise Audio) is referred to as specification for multi-channel digital audio coding technology which was issued as a Chinese national standard in 2008. There novel algorithms which are the differential Huffman, context-based Huffman and context-based arithmetic entropy coding, respectively, are presented to improve the compression efficiency of DRA entropy coding with 3%, 11.11% and 13.52% at the expense of the different implementing complexities and the intra-frame and/or inter-frame error propagation in live transmission.
Session: - SPG01-09 (Paper No.2220)
Paper Title: - Investigation of CNN-based models for Frog Calling Activity Detection
Authors: - Jie Xie, kai hu, Harry Hines, Jinglan Zhang, Ya Guo and Jinghu Yu
Abstract: - Frog plays an important role in monitoring and evaluating the health of wetland. Therefore, it is valuable to first detect frog activity, then classify species, and lastly predict frog population. In this study, we propose a CNN-based model for detecting frog calling activity. Specifically, continuous frog recordings are first segmented into segments having a fixed duration. Then, different CNN architectures and various time-frequency representations of frog calls are investigated. Next, different CNN modes are evaluated using different train and test datasets to test the robustness of the proposed model. Experimental results show that the best-averaged macro F1-score and accuracy obtained is 98.41% and 97.99%, which are obtained using Mel-spectrogram and VGG-style CNN. In addition, when the train and test data are collected on different days, the averaged macro F1-score is reduced to 82.12% and 71.38%, respectively.
Session: - SPG01-10 (Paper No.2237)
Paper Title: - Study on the Acoustic Signal Characteristics of Feeding Activity of Penaeus Vannamei
Authors: - Maochun Wei, Yating Lin, chen keyu and Wei Su
Abstract: - To solve the problem of intelligent feeding of Penaeus Vannamei, this paper presents a study on the characteristics of acoustic signals during feeding, which can adaptively control the feeding amount. In this experiment, a sound acquisition system was designed for the weak sound signal of feeding of Penaeus Vannamei. The sound signals of different feeding states of Penaeus Vannamei were collected under the condition of artificial feeding through several feeding box experiments. Through a large number of data analysis, we get and confirm the typical sound signal of feeding of Penaeus Vannamei. The frequency band of sound signal distribution in feeding of Penaeus Vannamei was studied. It provides theoretical support for further study on the feeding pattern of Penaeus Vannamei.
Session: - SPG01-11 (Paper No.2190)
Paper Title: - Classification of Steering Wheel Contacts from Electrocardiogram Signals Using Machine Learning
Authors: - Meghan McConnell, Belinda Schwerin, Nikaela Podolsky, Matthew Lee, Brent Richards and Stephen So
Abstract: - With the current day advancements in both computational power and machine learning (ML) techniques, there is a fundamental shift toward the application of new and smarter technologies. Worldwide incidents of motor vehicle crashes cause financial and emotional distress, along with physical injury, and even death, often stemming from driver fatigue. Nowadays, advanced ML techniques can be combined with electrocardiogram signals recorded from hand-contact with the motor vehicle steering wheel, to accurately detect the onset of driver fatigue. However, the signal recorded is only viable for fatigue analysis when two hands are in contact with the wheel. This work aims to carry out a comparative evaluation on a selected set of ML algorithms, when considering their ability to determine the number of contacts on a steering wheel. The ML classifiers considered in this study include the unsupervised methods K-means clustering, and Gaussian Mixture Model, and the supervised methods Support Vector Machine (SVM), Linear Discriminant Analysis, and Convolutional Neural Network (CNN). The evaluation is carried out based on both standard ML evaluation metrics including accuracy, precision, specificity, and computational cost. The experimental results show that the CNN produced the highest-level accuracy (>99%), but also had the highest computational cost. The SVM method presented the most balanced performance with a low computational cost and the second highest-level accuracy (94%). This paper assesses the viability of ML algorithms to eliminate the non-viable segments within ECGs that are used to determine driver fatigue. This is done by evaluating the techniques ability to consistently detect the number of contacts on a steering wheel, and its ability to be implemented in real-time, through the analysis of computational cost.
Session: - SPG01-12 (Paper No.2203)
Paper Title: - A DEEP LEARNING MODEL FOR THE AUTOMATIC DETECTION OF MALIGNANCY IN EFFUSION CYTOLOGY
Authors: - Shajahan Aboobacker, Deepu Vijayasenan, Sumam David, Pooja K. Suresh and Saraswathy Sreeram
Abstract: - The excessive accumulation of fluid between layers of pleura covering lungs is known as pleural effusion. Pleural effusion may be due to various infections, inflammations or malignancy. The cytologists visually examine the microscopic slide to detect the malignant cells. The process is time-consuming, and interpretation of reactive cells and cells with ambiguous levels of atypia may differ between pathologists. Considerable research is happening towards the automation of fluid cytology reporting. We propose an integrated approach based on deep learning, where the network learns directly to detect the malignant cells in effusion cytology images. Architecture U-Net is used to learn the malignant and benign cells from
the images and to detect the images that contain malignant cells. The model gives a precision of 0.96, recall of 0.96, and specificity of 0.97. The AUC of the ROC curve is 0.97. The model can be used as a screening tool and has a malignant cell detection rate of 0.96 with a low false alarm rate of 0.03.
Session: - SPG01-13 (Paper No.2214)
Paper Title: - Classification of EEG signals based on time-frequency analysis and spiking neural network
Authors: - Qinghua Wang, LIna Wang and Song Xu
Abstract: - Electroencephalogram (EEG) is one of the most effective and essential tools for analyzing and diagnosing epilepsy. However, there is a challenging task for detecting seizures from EEG signals, which is due to the non-stationary nature of EEG signals. This paper proposes a novel automatic EEG signal recognition method to assist epilepsy detection. Specifically, the multi-wavelet basis function (MWBF)    expansion method is first adopted to construct a time-varying autoregressive (TVAR) model of EEG signals, and a robust ultra-orthogonal least squares (UOLS) algorithm aided by derivative information of EEG signals is then utilized for model structure detection; besides, the power spectral density  (PSD) estimation method is applied to extract high-resolution time-frequency features; particularly, to fully exploited the spatiotemporal information of the extracted features, features were fed into the spiking neural networks (SNN) for classification. Experimental results on a widely-used benchmark dataset show that proposed methods outperform other related methods in terms of classification performance.
Session: - SPG01-14 (Paper No.2208)
Paper Title: - Investigation of tri-planar range of motion of ankle according to gender using photogrammetry technique
Authors: - Ali Al-kharaz and Albert Chong
Abstract: - The risk of ankle sprain during activitiesis very high. Therefore, studying ankle range of motion is important in order prevent any injuries. Photogrammetry is used to investigate tri-planar range of motion of the ankle. This study determines the differences in ankle mobility between genders. Twenty participants (10 females and 10 males) were recruited, and fourteen retro-reflective targets were mounted on the skin of each participant’s right foot. Imaging sensors were self-calibrated using a bundle adjustment technique and the images were downloaded with Australis photogrammetric software. The results show that the females have a higher angular plantar flexion than males, and that the maximum mean angular of inversion for males is higher than for females. The maximum mean angle of internal/external rotation of the ankle for males and females is very similar (there is no significant difference). These results may contribute to a better understanding of the influence of gender on the mobility of the ankle.
Session: - SPG01-15 (Paper No.2120)
Paper Title: - Performance Analysis of Deficient Length Normalized LMS Algorithm
Authors: - Wei GAO, Meiru Song, Lihuan Huang and Lingling Zhang
Abstract: - In the practical applications of adaptive filters, the system to be identified is often modeled as a finite impulse response (FIR) filter. When the a priori information of unknown system is unavailable, the length of transversal adaptive filter is usually shorter than that of actual system impulsive response. However, the deserted weight coefficients of FIR have significant impact on the convergence behavior of the deficient length adaptive filters. This paper studies the performance analysis of deficient length normalized LMS (DL-NLMS) algorithm for the correlated input data, which allows us to further understand its convergence behavior. Simulations illustrate the effectiveness and correctness of the derived theoretical results.
Track 3: - SPG02 Session, 22 Aug (Sat) 13:30-18:30 
Session: - SPG02-01 (Paper No.2140)
Paper Title: - Measurement-Aided Cubature Weight Correction GM-PHD algorithm for Muti-target tracking
Authors: - zhen Ren and Hao Wu
Abstract: - So as to improve the accurate multi-target tracking, this paper proposes a Measurement-Aided Cubature Weight Correction GM-PHD (CKF-MCGMPHD) algorithm. The algorithm introduces the measurement auxiliary weight correction into the GM-PHD algorithm, and uses the estimated value to correct the updated Gaussian weights to fix the Gaussian weights, so as to further improve the accuracy of the filtering algorithm, and reduce the calculation at the same time ; The CKF method is introduced into the GM-PHD algorithm, and uses cubature point to predict and update the GM-PHD, so as to improve the nonlinearity and stability of the filtering algorithm. The simulation results show that compared with the traditional algorithm, CKF-MCGMPHD can obtain more accurate target tracking results under different clutter environments, and the computational complexity is smaller.
Session: - SPG02-02 (Paper No.2153)
Paper Title: - Unsupervised Learning Based Acoustic NLOS Identification for Smartphone Indoor Positioning
Authors: - Wentao Xue, Zhixin Hu, Nan Wang and Lei Zhang
Abstract: - The NLOS phenomenon seriously impair the performance of acoustic indoor localization for smartphones in the real indoor environments. Through identifying and discarding the NLOS measurements, the positioning performance can be improved by incorporating only the LOS measurements. Even though the method based on SL and SSL can obtain a satisfactory accuracy, those methods are very hard to be apply because they need a large number of labeled signals. In this paper, an UL based acoustic NLOS identification approach is proposed. Based on the features extracted from the characteristics of acoustic channel, the performance of UL based methods, such as $k$-means, GMM  and AGNES clustering algorithm, are evaluated and compared under accuracy criterion. The results show that the optimal size of sample set is 100, the accuracy of UL clustering algorithm based on GMM is higher than $k$-means and AGNES. The optimal feature combination is $\{  \tau_{med}, \tau_{rms}, s, g_m, g_{rms} \}$.
Session: - SPG02-03 (Paper No.2165)
Paper Title: - A Pilot Workload Evaluation Method Based on EEG Data and Physiological Data
Authors: - Jun Chen, Lei Xue and Zuocheng Liu
Abstract: - With the increasing complexity of today's battlefield environment, the workload level of the pilots greatly affects the smooth execution of flight tasks. When a pilot's workload is too high, it can cause his ability to sense the surrounding environment to be reduced, and even cause extremely dangerous flight accidents. Based on experiments and statistics, the sensitivity of the pilot's EEG (electroencephalography) and physiological factors under different flight tasks is analyzed, and a pilot workload evaluation model is established based on the SVM (support vector machines) classifier.
Session: - SPG02-04 (Paper No.2177)
Paper Title: - A Duffing Detection Method of Eliminating Noise False Alarm Based on Detection Statistics
Authors: - CAIHONG LIU
Abstract: - Abstract—Duffing chaotic systems are sensitive to certain signals and immune to noise, the properties of which demonstrate their potential application in weak signal detection. Determination of critical value for abrupt change of the Duffing system is important theory basis for Duffing oscillator system. A solving formula of the critical value of Duffing oscillator system is obtained by using harmonic balance analysis method. The simulation results show that the theoretical solution and numerical simulation are preferably coherence. Noise can cause the false alarm detection. Based on the sensitivity of damping ratio, the variable damping ratio Duffing oscillator detection method with variable amplitude coefficients is introduced to reduce the false alarm caused by noise. To overcome the shortcomings of the phase diagram discrimination method in quantitative analysis, a detection statistics identification method is proposed. The results of the experiment verify the feasibility and accuracy of the new approach.
Session: - SPG02-05 (Paper No.2194)
Paper Title: - A novel database for plant diseases and pests classification
Authors: - Qiyao Wang, Guiqing He, Feng Li and Haixi Zhang
Abstract: - In agricultural field, the research, detection, and treatment of plant diseases and pests play a very important role. Prevention or early treatment of diseases and pests can significantly increase crop yields. With rich variety, plants form a mature hierarchical structure based on taxonomic methods. Thus, in the process of computer vision, plant classification and identification have attracted many researchers, and then the detection system of plant diseases and pests came into being. In this paper, the keyword retrieval is used to obtain images of plant diseases and pests from the keyword search engine to build a novel database. After that, a hierarchical multitask learning is proposed to classify plant diseases and pests by leveraging the relationship between different plant species and pests. The experimental results confirmed the feasibility and reliability of the classification of plant diseases and pests using deep learning model.
Session: - SPG02-06 (Paper No.2200)
Paper Title: - Study of circular recognition algorithm in statistics of microbubbles
Authors: - Wenyuan Sun, Yuchen Lei and Yongzhi Sun
Abstract: - Abstract—In view of the results of large-scale preparation of microbubbles, as well as the digital requirements for microbubble size and quality control, this paper uses the algorithm based on the Hough circular transformation, assisted by the related pre-processing and the selection constraints setting of the results to realize the identification, detection and statistics of large-volume microbubbles. On the basis of the theoretical analysis and the experimental method, the algorithm proposed in this paper uses the computer vision library (OpenCV) to realize the program and achieves the purpose of judging the quality control of the preparation results by measuring and comparing the particle size distribution of the microbubbles obtained from a large number of preparations.
Session: - SPG02-07 (Paper No.2210)
Paper Title: - A Novel Detection and Localization Approach for Nonlinear Faults in Two-Dimensional Structures
Authors: - Quankun Li and Mingfu Liao
Abstract: - In this study, a novel detection and localization approach for nonlinear faults (loosening bolts or fatigue cracks) in two-dimensional structures is proposed and verified. In this new approach, a two-dimensional discrete multi-degree-of-freedom (MDOF) model, which simulates nonlinear faults as nonlinear damper-spring elements between each mass, is applied to characterize the dynamic behaviour of two-dimensional structures. With the nonlinear output spectrum and multi-position excitation method, damage features correspond to properties of substructures are derived, and local damage indexes are defined for nonlinear fault detection and localization in substructures. Numerical results of case study demonstrate that the proposed approach is realizable and sensitive to faults in the substructure under diagnosis even though there are faults in other substructures.
Session: - SPG02-08 (Paper No.2216)
Paper Title: - Classification and Recognition of Underwater Target Based on MFCC Feature Extraction
Authors: - Yuze Tong, Xin Zhang and Yizhou Ge
Abstract: - The key to underwater target recognition is to extract the effective features of underwater target radiation noise. This paper presents an effective method for underwater target recognition and classification by extracting Mel-Frequency Cepstral Coefficients (MFCCs) features of underwater target radiation noise. Compared with traditional spectral analysis methods, MFCC makes full use of the non-linear auditory effect of the human ear with different perception capabilities for sounds of different frequencies. In this paper, the classification experiment of the radiated noise of the three types of measured underwater targets is done, where the MFCC feature vectors of the three types of targets are extracted, and the K-Nearest Neighbor (K-NN) algorithm is used to classify and identify them. Finally, the experimental results show that the method is effective.
Session: - SPG02-09 (Paper No.2202)
Paper Title: - AOPNet: Anchor Offset Prediction Network for Temporal Action Proposal Generation
Authors: - Fan Peng, Kun Li, Xueliang Liu and Dan Guo
Abstract: - Temporal action proposal generation is a hot topic in the field of video understanding, aiming to extract action instances in long untrimmed videos. Although various approaches have been proposed, the challenge of proposal generation with reliable confidence scores and precise boundaries still exists. To solve this challenge, we propose a novel proposal generation framework called Anchor Offset Prediction Network (AOPNet), which utilizes multi-layer Bidirectional RNN (BiRNN) to predict multi-scale action proposals and fine-tune temporal boundaries with the predicted anchor offsets. Experimental results demonstrate that though with the use of anchor boundary offset prediction module, the AOPNet can remarkably improve the classification accuracy of temporal anchors and finally generate temporal proposals with high recall. Ablation studies on the THUMOS-14 dataset demonstrated the effectiveness of the proposed approach.
Session: - SPG02-10 (Paper No.2205)
Paper Title: - Localization of steel strand damage based on Empirical Mode Decomposition algorithm
Authors: - Yongju Chu, Ke He, Haiyan Wang and Xiaopeng Lv
Abstract: - In order to accomplish rapid detection of the damage in mooring steel strand of the offshore platform, this paper proposes to use the ultrasonic guided wave method for the online damage detection. Considering the non-stationary characteristics of the ultrasonic signal, this paper proposes an Empirical Mode Decomposition (EMD)-based algorithm to detect and localize the position of steel strand damage. To be specific, the detect signal is firstly decomposed by EMD into multiple Intrinsic Mode Functions (IMFs). Then, in order to find the weak damage signal which is usually buried in noise, a weighted synthesis procedure is adopted, i.e., assigning more weights to the dominant components of IMFs. By doing so, the reconstruction signal is denoised and the damage signal can be much more easily captured. Simulation results show that the proposed method can effectively detect the ultrasonic guided wave signal and realize the accurate localization of the weak damage to the steel strand by comparing with the traditional method of directly selecting several natural IMFs.
Session: - SPG02-11 (Paper No.2139)
Paper Title: - Indoor Object Identification based on Spectral Subtraction of Acoustic Room Impulse Response
Authors: - Haitao Wang, Xiangyang Zeng, Ye Lei, Shuwei Ren, Feng Hou and Ningjuan Dong
Abstract: - Object identification in the room environment is a key technique in many advanced engineering applications such as the unidentified object recognition in security surveillance, human identification and barrier recognition for AI robots. The identification technique based on the sound field perturbation analysis is capable of giving immersive identification which avoids the occlusion problem in the traditional vision-based method. In this paper, a new insight into the relation between the object and the variation of the sound field is presented. The sound field difference before and after the object locates in the environment is analyzed using the spectral subtraction based on the room impulse response. The spectral subtraction shows that the energy loss caused by the sound absorption is the essential factor which perturbs the sound field. By using the energy loss with high uniqueness as the extracted feature, an object identification technique is constructed under the classical supervised pattern recognition framework. The experiment in a real room validates that the system has high identification accuracy. In addition, based on the feature property of position insensitivity, this technique can achieve high identifying accuracy with a quite small training data set, which demonstrates that the technique has potential to be used in real engineering applications.
Session: - SPG02-12 (Paper No.2124)
Paper Title: - Enhancement of Mammograms with Periphery Brightness Compensation
Authors: - Yuting Tang and Junfeng Wang
Abstract: - Mammography is an effective way for early diagnosis of breast cancer. The enhancement of mammograms is essentially important for the diagnosis. This can be achieved by high-pass filtering and grayscale mapping. However, the details in the periphery of the breast may be lost because the breast is bright except its dark periphery in the mammogram. In this paper, we present a technique for the enhancement of mammograms with periphery brightness compensation. First, the image is decomposed into a base image and a detail image. Especially, the decomposition is adaptive such that no artifacts are made on the boundary of the breast in the detail image. Then, the periphery brightness is compensated by a Gamma correction in the base image, and the detail image is linearly amplified. Finally, the base image and the detail image are summed and mapped to obtain an image with significantly enhanced details, even in the periphery of the breast. Some results indicate the effectiveness and the advantage of this technique.
Session: - SPG02-13 (Paper No.2173)
Paper Title: - CBC and DBC Counter Using Image Processing and Analysis
Authors: - Jennifer Dela Cruz, Janette Fausto, Errol Ace Agatep, Emmanuel John Manlangit and John Renielle Valenzuela
Abstract: - In this study, the researchers developed a cost-effective way of performing Complete Blood Count (CBC) and Differential Blood Count (DBC) by implementing image processing, analysis, and machine learning algorithms using a Raspberry pi B+ microcontroller in conjunction with a microscope that has Short Message Service (SMS) feature to be used in rural areas for remote transmission of results to a health professional. Blood image input from the raspberry pi’s camera will be used for the counting and classification needed by the system. The capturing and analysis of the images will be integrated on a graphical user interface (GUI) to provide user-friendly access. HSV color space conversion, thresholding, morphological operations, and Connected Component Labeling (CCL) are used for CBC. For DBC, an additional process of Convolutional Neural Network (CNN) classification is done. The study shows a remarkable result by having no significant difference from conventional method which is manual counting.
Session: - SPG02-14 (Paper No.2187)
Paper Title: - A Memory and Computation Efficient Local Stereo Algorithm Suited for Mobile Robot Applications
Authors: - Jinjin Shi and Pan Liu
Abstract: - Range sensing is a key component for a mobile robot to interact with the environment. This paper proposes a memory and computation efficient local stereo algorithm for mobile robot applications. The memory reduction is done by enforcing local smoothness constraint during the weighted cost aggregation to make the difference between neighboring aggregated matching costs small and then applying predictive coding scheme to compress the original representations.In addition,by performing cost aggregation only at the sampled positions, the run time is largely reduced and the memory requirement is further decreased. Finally, a high resolution disparity map is derived by an efficient disparity upsampling algorithm utilizing approximated geodesic distance. The tradeoff between matching accuracy, memory cost and computational complexity is extensively investigated using Kitti dataset. Experimental results demonstrate that our method provides high-quality disparity maps with low memory cost and computational effort.
Session: - SPG02-15 (Paper No.2215)
Paper Title: - Reticular shield detection algorithm using superpixel segmentation
Authors: - Zelin Yu, Lingyun Cai, Weizheng Jin and Tang Jin
Abstract: - In this paper, we propose a novel reticular shield detection algorithm to restore the images that are obscured by reticular shield such as barbed wires and fences. The whole framework is composed of three stages: stage-1 for accurate superpixel segmentation, stage-2 for deriving the mask of reticular shield and stage-3 for restoring the image by introducing the algorithm. In the former 2 stages, to obtain the accurate mask of the shield, we firstly conduct superpixel segmentation of the image, introduce the joint feature of color and texture histogram to describe the superpixels, and convert the classification problem based on pixel into the classification problem based on superpixels, so as to suppress the misclassification caused by local color changes. Then, the graph cut algorithm is used to classify the superpixels, so that the reticular structure can be extended along smooth edges and is not restricted by fixed shapes, which improves the detection accuracy of irregular reticular structure and does not rely on the manual input required by some traditional algorithm. Additionally, a new joint feature is used to train support vector machine (SVM) classifier and classify all unclassified superpixels to obtain an accurate reticular shield mask. In the final stage, we introduce an image restoring algorithm, to restore the image.

　 　 　 　
　 Track 1: -  COM01 Session, 23 Aug (Sun) 09:00 - 1300 　
　 　 　 　
　 Session: - COM01-01 (Paper No.2122) 　
　 Paper Title: - An Improved Proportionate Normalized LMS Based on the L0 Norm for Underwater Acoustic Channel Estimation 　
　 Authors: - Xin Hu, Yingmin Wang, FEIYUN WU and Aiping Huang 　
　 Abstract: - The improvement proportionate normalized least-mean-square (IPNLMS) algorithm has been developed for acoustic echo cancellation technology. Nevertheless, the sparsity of the channel is not sensitive enough to be identified by this algorithm. In this paper, a new channel estimation algorithm is proposed by making full use of channel sparsity for underwater acoustical (UWA) channel estimation. The proposed algorithm integrates L0 norm into its cost function of the IPNLMS algorithm based on L0 norm, firstly. And then, we apply IPNLMS, IPNLMS-L0, L0-IPNLMS and the proposed algorithm in channel estimation of a shallow water acoustic channel. The performance of the algorithm we proposed achieves fast convergence and low misadjustment in a channel estimation simulation experiment. Simulation results show that this algorithm is better, compared with other IPNLMS-type algorithms. 　
　 　 　 　
　 Session: - COM01-02 (Paper No.2179) 　
　 Paper Title: - Recognition method of key nodes in SN-UASN based on TOPSIS 　
　 Authors: - Yifan Yuan, Xiaohong Shen, Ke He, Haiyan Wang, Lin Sun and Shilei Ma 　
　 Abstract: - In Underwater Acoustic Sensor Networks (UASN), certain key nodes in the network will cause network partition, communication failure, packet loss once they failed to work. Also, because the key nodes are key positions of the UASN, once the malicious behavior of the key nodes occurs, the network performance will drop sharply. This paper first puts forward the key node recognition method for UASN. Also, we propose node usage and hop distance factors to evaluate node’s importance. Five factors at the topological level are combined, and using Technique for Order Performance by Similarity to Ideal Solution (TOPSIS) method to evaluate the importance of each node in the UASN. In the experiment, the key nodes are replaced by attack nodes to verify the performance of the algorithm. Experiment results show that the algorithm has a good and effective value. 　
　 　 　 　
　 Session: - COM01-03 (Paper No.2206) 　
　 Paper Title: - An Improved BICM-ID Receiver for the Time-Varying Underwater Acoustic Communications with DDPSK Modulation 　
　 Authors: - Hongyu Cui, Changxin Liu, Xiaoping Hong, Jie Wu and Dajun Sun 　
　 Abstract: - Double differential phase shift keying(DDPSK) modulation is an efficient method to compensate the Doppler shifts, whereas the phase noise will be amplified which results in the signal-to-noise ratio (SNR) loss. In this paper, we propose a novel receiver architecture for underwater acoustic communications with Doppler shifts. The proposed method adopts not only the DDPSK modulation to compensate the Doppler shifts, but also the improved bit-interleaved coded modulation with iterative decoding (BICM-ID) algorithm for DDPSK to recover the SNR loss. The improved DDPSK demodulator adopts the multi-symbol estimation to track the channel variation, and an extended trellis diagram is constructed for DDPSK demodulator. Theoretical simulation shows that our system can obtain around 10.2dB gain over the uncoded performance, and 7.4dB gain over the hard-decision decoding performance. Besides, the experiment conducted in the Songhua Lake also shows that the proposed receiver can achieve lower BER performance when Doppler shifts exists. 　
　 　 　 　
　 Session: - COM01-04 (Paper No.2217) 　
　 Paper Title: - Path-Planning Analysis of AUV-Aided Mobile Data Collection in UWA Cooperative Sensor Networks 　
　 Authors: - Shihan Chen, Yougan Chen, Jianying Zhu and Xiaomei Xu 　
　 Abstract: - The autonomous underwater vehicle (AUV) aided mobile data collection is an effective method for reducing the energy consumption of the underwater acoustic (UWA) sensor networks. In this paper, we propose an AUV-aided path-planning scheme using cooperative transmission mechanism for a medium-scale UWA sensor network. In the proposed scheme, we analyze not only the energy consumption, but also the task duration and path-planning cost comprehensively for practical applications. We analyze four different path-planning schemes in terms of energy consumption of UWA sensor nodes and travel cost of AUV. The simulation results show that the lawn mower path-planning scheme has lower energy consumption of UWA sensor networks. But the circle path-planning scheme has lower working time and path energy consumption of AUV. Therefore, in view of different needs, we should make a comprehensive selection. 　
　 　 　 　
　 Session: - COM01-05 (Paper No.2230) 　
　 Paper Title: - A Test of Cross-border Magnetic Induction Communication from Water to Air 　
　 Authors: - Zhiying Tian, Xin Zhang and Haochen Wei 　
　 Abstract: - In the Underwater Sensor Network (UWSN), the cross-border communication between the underwater sensor carrier and the water data center has important practical significance. At present, under-water acoustic communication and sea surface relay is mostly adopted by the cross-border communication, but the communication performance will be challenged by the complex propagation characteristics of mixed channels near the sea surface. The magnetic induction field can directly propagate across the air-water interface and is insensitive to time-varying, multipath channels. This paper analyzes the propagation characteristics of the sea-air transboundary magnetic induction field and develops a transboundary magnetic induction communication test system and conducts a transboundary communication test from underwater to air. The test results show that cross-border magnetic induction communication is feasible and reliable from water to air. 　
　 　 　 　
　 Session: - COM01-06 (Paper No.2231) 　
　 Paper Title: - Soft Frequency Reuse with Large Propagation Delays for Interference Mitigation in Underwater Acoustic Networks 　
　 Authors: - Yuzhi Zhang, Rui Fan and Anyi Wang 　
　 Abstract: - The available bandwidth for underwater acoustic (UWA) communication and networks is very limited, and frequency reuse is one of the important techniques to improve spectrum efficiency and system capacity. In this paper, a soft frequency reuse (SFR) system is proposed for UWA sensor networks, with the consideration of physical layer channel features including propagation attenuation and delay. As the propagation speed of UWA signal is very slow, the time arrived at one user from multiple senders are obviously different, herein a new idea to mitigate inter-cell interference based on both sub-bands allocation and propagation delay difference has been proposed for UWA networks. The simulation results have shown that, with the proposed methods, UWA sensor network can achieve better SINR and spectrum efficiency in comparison to traditional FFR and SFR. 　
　 　 　 　
　 Session: - COM01-07 (Paper No.2254) 　
　 Paper Title: - Routing Design Based on Data Importance Rating in Underwater Acoustic Sensor Networks 　
　 Authors: - Changjing Xiong, Yougan Chen, Jianying Zhu and Xiaomei Xu 　
　 Abstract: - For the underwater acoustic senor networks (UWA-SN), the data carried by different sensor nodes usually has different importance at different moments. In this paper, we propose a data importance rating routing (DI2R) protocol using data importance index for UWA-SN to concern the above issue. In the proposed DI2R scheme, we take not only the data importance rating, but also the residual energy of sensor node and packet loss probability comprehensively into account. Specifically, we analyze two different implementation schemes according to whether to adjust the minimum transmitting power (Pmin) of each sensor node in UWA-SN for the proposed DI2R protocol, in terms of energy consumption, death node number, packet loss probability, and transmission distance. The simulation results show that the dynamic Pmin DI2R routing scheme has lower energy consumption of UWA-SN and lower packet loss probability compared to the static Pmin DI2R routing scheme; But the static Pmin DI2R routing scheme has lower node mortality and shorter transmission distance. 　
　 　 　 　
　 Session: - COM01-08 (Paper No.2266) 　
　 Paper Title: - Mapping Diversity for High Bandwidth Efficiency CCK in Underwater Acoustic Channels 　
　 Authors: - Han Wang, Lianyou Jing, Chengbing He and Jianguo Huang 　
　 Abstract: - This paper presents an optimal design of QAMCCK modulation based on maximizing mapping diversity for
ARQ transmission scheme, aiming at improving the unsatisfying
performance of QAM-CCK in underwater channels with single
transmission. In order to achieve the optimal BER performance,
this paper first derives the pair-wise error probability after
re-transmission and then obtains the optimal constrcution of
QAM-CCK by minimizing the target function. Simulation results
demonstrate that optimum desigend QAM-CCK achieves 2 dB
SNR gain compared to randomly selected QAM-CCK when
BER equals 10 4
. Furthermore, optimum desigend QAM-CCK
achieves 5 dB SNR gain compared to randomly selected QAMCCK when BER equals 10 5
under tested acoustic channels from
UNet06 sea experiments.
Index Terms—CCK, QAM-C		 　
　 　 　 　
　 Session: - COM01-09 (Paper No.2227) 　
　 Paper Title: - Strong Barrier Coverage in Underwater Directional Sensor Network 　
　 Authors: - Juan Chang 　
　 Abstract: - Barrier coverage is one of the research hotspots in underwater wireless sensor networks (UWSNs). However, the barrier coverage in underwater directional sensor networks(UDSNs) has not been paid much attention. In the past works, barrier coverage algorithms in UWSNs are based on directional barrier graph(DBG). Judging the position relationship between two directional sensors is the core of establishing DBG. In this paper, the geographical relations is exploited to build DBG. Based on the DBG, the adjacency matrix-based algorithm[11 in UDSNs is discussed.  The following experimental studies demonstrate the effectiveness of the proposed algorithm. 　
　 　 　 　
　 Session: - COM01-10 (Paper No.2253) 　
　 Paper Title: - Underwater Magnetic field and temperature sensing with a Michelson-like interferometer of core-offset and FBG 　
　 Authors: - Xueyi Jin, Xiaokangx Li, Dongshi Wang, Haili Yang, Minyuan Xian and Dexing Yang 　
　 Abstract: - An underwater magnetic field and temperature sensing structure consisting of a core-offset and fiber Bragg grating (FBG) immersed into magnetic fluid (MF) is proposed and investigated experimentally. The structure is analogous to a Michelson-like interferometer, where cladding modes are excited by a large core-offset (around 8 μm) and recoupled back to the core by the downstream FBG. The area of the cladding modes indicates the mode intensity. Via cladding mode area and FBG wavelength interrogations, magnetic field and temperature sensing with sensitivities of -0.3427 /mT and 7.34 pm/℃ for magnetic field and temperature changes, respectively, are achieved. This structure is low-cost, easy-fabrication and will find more applications in multiparameter sensing if other sensitive materials used. 　
　 　 　 　
　 Session: - COM01-11 (Paper No.2099) 　
　 Paper Title: - The Study on Path Tracking Control Method Based on Fuzzy-CMAC for Autonomous Vehicle in Rural Environment 　
　 Authors: - jiajia chen, wei han, pan zhao, zhenya wei and xingtao liu 　
　 Abstract: - In complex outdoor environments, the traditional control methods based on dynamic model were difficult to meet the special control requirement of mobile robots. In this paper, a humanoid control strategy was used and the composite control method based on CMAC neural network and Fuzzy control was constructed. The neural network improves the self-learning ability of controller when the model and input signal change unpredictability，and the fuzzy logic was used to simulate the experience of experts and intuition of human brain, maintain the stability and robustness of system in rural environment. The experiments showed that this method could improve the high mobility and stability performance in mobile robot path tracking problem in different paths, terrain and geological conditions. 　
　 　 　 　
　 Session: - COM01-12 (Paper No.2102) 　
　 Paper Title: - An improved Visual Indoor Navigation Method based on Fully convolutional Neural Network 　
　 Authors: - Chengqing Wen, Chenning Li and Haowei Xu 　
　 Abstract: - In this paper, a new method of indoor location based on convolution network (FCN) is proposed, which greatly improves the accuracy of contour recognition in image recognition, and thus improves the final location results. In addition, this paper also proposes a pre trained CNN (revolution neural network) to get a rough location to solve the location problem caused by the distribution of fuzzy layer and similar objects in the existing algorithm. The experimental results show that the proposed method can effectively improve the success rate from 85% to 93% and the positioning error from 2.6m to 1.4m indoor positioning algorithm compared with the previous TM based and r-cnn based positioning algorithms. 　
　 　 　 　
　 Session: - COM01-13 (Paper No.2126) 　
　 Paper Title: - Facial expression recognition based on the ensemble learning of CNNs 　
　 Authors: - Jia Chen, Li Li Chu and Ying Zhou 　
　 Abstract: - With the improvement of computer computing resources and the continuous enrichment of deep learning theory, automatic facial expression recognition technology has been further developed. Because of the complexity and subtlety of expressions, real-time facial expression recognition is still a big problem. In this paper, we proposes a facial expression recognition method based on the ensemble learning of CNNs. Our model is composed of three sub-networks, and uses the SVM classifier to synthesize the output of the three networks to obtain the final predicted value. The recognition accuracy of the model's expression on the FER2013 dataset reached 71.27%. The results show that the method has high test accuracy and short prediction time, and can realize real-time, high-performance facial recognition. 　
　 　 　 　
　 Session: - COM01-14 (Paper No.2170) 　
　 Paper Title: - Credit Risk Evaluation Based on Data Mining and Integrated Feature Selection 　
　 Authors: - Yuanjie Deng, Ying Wei and Yujun Li 　
　 Abstract: - Credit risk assessment is an important process for every financial institution. Most current studies in this area focus on the development of the classifiers. However, besides the classifiers, data analysis and mining also play an important role in improving the accuracy. In this paper, an integrated feature selection method is proposed combing the analysis of the data in one feature but belongs to different labels and the analysis of the relationship between a single feature and labels. Back propagation neural network is used as the classification model. Experiments are done on Australian credit approval dataset. The results show that the proposed integrated feature selection can improve the accuracy of assessment compared with other previous work. 　
　 　 　 　
　 Session: - COM01-15 (Paper No.2204) 　
　 Paper Title: - Development of Machine Learning-based Predictive Models for Wireless Indoor Localization Application with Feature Ranking via Recursive Feature Elimination Algorithm 　
　 Authors: - Jennifer Dela Cruz and Timothy Amado 　
　 Abstract: - The next-generation wireless technologies offer a vast array of services, from ultra-high-speed data communications to internet of things (IoT) massive machine-to-machine communications. However, the current network setup cannot handle such variety of services. Consequently, it is becoming a trend to apply artificial intelligence, machine learning and deep learning to wireless networks. And one of these applications is on wireless indoor localization. Wireless indoor localization takes advantage of wireless access points (WAPs) received signal strength indicators (RSSI) values to pinpoint the location of a user, similar to concept of GPS but indoors. The goal of this paper is to apply machine learning techniques to develop predictive models that can be used to predict the location of a user based on RSSI readings that his smartphone receives. In the study, four machine learning algorithms are used which are support vector machines, random forest, Naïve-Bayes classifier and neural networks. All models are successfully implemented each having an accuracy of 97.83%, 97.67%, 98.50% and 97.33% respectively. Also, a recursive feature elimination algorithm is also used to determine the predictor that has the least impact amongst all other features and it is found out in the study that WAP2 is contributes the least influence when the predictive models are developed. 　
　 　 　 　
　 　 　 　
　 Track 2: - SPG03 Session, 23 Aug (Sun) 09:00 - 13:00 　
　 　 　 　
　 Session: - SPG03-01 (Paper No.2103) 　
　 Paper Title: - Discrimination of FM Signal Types Based on Time-Frequency Analysis 　
　 Authors: - weihao cao, zhixiang yao and wenjie xia 　
　 Abstract: - To solve the problem of discrimination of non-cooperative pulse signals type, a frequency modulation pulse type discrimination method based on time-frequency analysis is proposed. The instantaneous frequencies of linear frequency modulation (LFM) pulse and hyperbolic frequency modulation (HFM) pulse are estimated by using Rife interpolation short-time fractional Fourier transform-variable weight fitting method. By comparing the root mean square error between the estimated instantaneous frequency and the reciprocal fitting instantaneous frequency, the types of LFM pulse and HFM pulse signals are accurately distinguished. The simulation results show that this method has higher accurate discrimination probability under lower signal-to-noise ratio (SNR). 　
　 　 　 　
　 Session: - SPG03-02 (Paper No.2107) 　
　 Paper Title: - A New Target Classification Method for Synthetic Aperture Radar Images Based on Wavelet Scattering Transform 　
　 Authors: - Hongliang Zhu, Tat Wong, Nan Lin, Howong Lung, Zhayuan Li and Segios Thedoridis 　
　 Abstract: - Target Classification for SAR images is becoming a hot issue since the emergence of deep learning technology. Many kinds of different deep neural networks have been developed to handle this problem. With the strong capability of learning and generalization, the neural network could extract the features from SAR images easily. However, training an active deep neural network usually calls for a large number of data, which is a shortcoming for SAR images data. It is so precious to acquire such large number of SAR data for its high cost and limitation. Aiming at this problem, this paper discusses and proposes a brand-new target classification method for SAR images based on wavelet scattering transform, which extracts the features from SAR images by using the wavelet scattering. This kind of feature extractor has a strong capability of shift-invariance and local stability that constructs representation with rotation, translation, and scale invariance in 2D images. Finally, we construct a PCA + SoftMax shallow neural network to finish the classification task by using the obtained features. The experiment results on the MSTAR dataset show that our proposed method achieves an average accuracy of 99.9% in ten target categories, which outperforms other methods with less time consuming. 　
　 　 　 　
　 Session: - SPG03-03 (Paper No.2112) 　
　 Paper Title: - Radar and Communication Integration Based on OFDM Signal 　
　 Authors: - Min Jiang, Linlin Qi, Yuan Yao, Guangxin Wu and Baotao Huang 　
　 Abstract: - In modern electronic warfare, integration of radar
and communication plays an important role in maximizing
resource utilization and reducing electromagnetic interference,
especially for the integrated waveform design technology based
on signal sharing. In this paper, we consider orthogonal frequency
diversity multiplexing (OFDM) into the design of an integrated
radar and communication system with P4 complementary code
serving as modulation code. An analytical expression of the
ambiguity function for OFDM signal with each pulse consists
of multiple consecutive symbols is derived. The idea of minimum
spacing matching to demodulation is then proposed. Moreover,
an integrated modelling is established, of which radar and
communication performance is evaluated.		 　
　 　 　 　
　 Session: - SPG03-04 (Paper No.2132) 　
　 Paper Title: - A study on the cubic range model for MEO SAR ground moving target imaging 　
　 Authors: - Yongkang Li, Feng Wang, Lina Zeng and Hongmeng Chen 　
　 Abstract: - Due to the advantages of large coverage, and short revisit time, medium-earth-orbit (MEO) synthetic aperture radar (SAR) is an attractive tool for ground moving target indication (GMTI). This paper studies the derivation and the accuracy of the third-order Taylor approximated range equation for MEO SAR ground moving target imaging. First, the coordinates of a ground accelerating target and MEO SAR in the earth-centered rotating coordinate system, as well as the target’s range equation are developed. Second, the third-order Taylor approximated range equation is derived. Finally, the accuracy and the application scope of the third-order Taylor approximated range equation are investigated via numerical simulations. 　
　 　 　 　
　 Session: - SPG03-05 (Paper No.2176) 　
　 Paper Title: - Range Ambiguity Suppression for Underwater Distributed Localization Based on Total Support Degree 　
　 Authors: - Qing Li 　
　 Abstract: - Underwater distributed synchronous positioning can be adopted to locate moving targets at a high speed. But the problem of range ambiguity always exists, as synchronous period of the system is short. A range ambiguity suppression method based on total support degree (TSD) is proposed for solving the range ambiguity problem with multipath interference. The time of arrival and phase difference measurements are used to calculate alternative coordinates of the target. Then total support degree is employed to evaluate the majority agreement of alternative coordinates. Finally, ambiguity periods are distinguished by comparing the total support degrees, thus unambiguous localization under complex underwater multipath circumstance can be guaranteed. Simulation results demonstrate the effectiveness and robustness of the proposed method. 　
　 　 　 　
　 Session: - SPG03-06 (Paper No.2198) 　
　 Paper Title: - Secondary Surveillance Radar Signal Processing Based on Two-channel Deep Residual Network 　
　 Authors: - Xue Du, Xiaofeng Shen and Kuo Liao 　
　 Abstract: - The secondary surveillance radar system is the main surveillance method for national defense and domestic civil aviation. It will be interfered by noise during the transmission process, affecting the clarity of signal transmission, and also leading to a reduction in the stability and reliability of radio transmission. Although the traditional denoising method is effective, there is still a lot of noise remaining, which affects the signal detection. This paper proposes a secondary radar signal processing method based on the two-channel deep residual neural network (2CD-NET), which can effectively extract the deep features of the secondary radar signal and predict the response timing signal. It is verified by experiments that under the normal working noise environment of the secondary radar, the two-channel deep residual neural network has strong noise suppression performance. When the signal-to-noise ratio is 10dB, the signal-to-noise ratio of the test set can reach 99.98%. When the signal-to-noise ratio is 5dB, the test set accuracy rate can reach 94.71%. 　
　 　 　 　
　 Session: - SPG03-07 (Paper No.2239) 　
　 Paper Title: - Orthogonal Waveform Design of MIMO Radar Based on Niche Genetic Algorithm 　
　 Authors: - Bingnan Pei, Haosheng zhang and Tengda Pei 　
　 Abstract: - In order to optimize the orthogonal design of MIMO radar waveform, an improved algorithm based on niche genetic algorithm is proposed. The algorithm is able to eliminate individuals with similar structures in the population or data, maintain a certain distance between individuals, and effectively maintain the diversity of them. It thus eliminates the phenomenon of premature convergence, strengthens the global search ability of the algorithm, and improves its convergence speed. It is shown that The algorithm has the advantages of overall search ability of genetic algorithm and the ability of the niche genetic algorithm to discover multiple optimal solutions. Simulation was given to show that the algorithm was cable to reduce the side-lobe of the auto-correlation function of the orthogonal polyphase code and the peak value of the cross-correlation function of the code with good performance of autocorrelation and cross-correlation obtained 　
　 　 　 　
　 Session: - SPG03-08 (Paper No.2257) 　
　 Paper Title: - The Design of Resolution Improvement Based on TKIS-I Helmet-mounted Colorful Imaging Sonar 　
　 Authors: - Xiangkai Jia, Jing Wang, Yucong Deng, JingLi Wang and Wenjiao Shan 　
　 Abstract: - TKIS-I is a helmet-mounted single-beam mechanically scanned color imaging sonar with a working frequency of 678KHz, the beam width of 2.2°, and a maximum detection distance of 100m.It can be carried by a single soldier to complete the operation. It is mainly used for underwater salvage, search and rescue, etc. It has been used in naval forces and has been widely recognized.TKIS-I helmet-mounted colorful imaging sonar has a lower angular resolution effect on close-range targets, and cannot distinguish between closerange targets.In this paper, FPGA is added on the basis of the system circuit of TKIS-I helmet-mounted colorful imaging sonar to process high-speed signal, which makes the system
work at a frequency of 2.25MHz.It can effectively improves the angular resolution of close-range underwater targets and enhances the signal strength of target objects.Finally, the imaging test was performed on the 2.25Mhz high-frequency imaging sonar system. From the comparison results, it can be
concluded that the angular resolution of the target object has been enhanced.		 　
　 　 　 　
　 Session: - SPG03-09 (Paper No.2265) 　
　 Paper Title: - A Method of SAS with Sparse Aperture Based on Compressive Sensing 　
　 Authors: - shi yi, xia ji and Guijuan Han 　
　 Abstract: - Abstract—High resolution synthetic aperture sonar (SAS) is limited by some factors such as size of the sonar carrier, carrier speed, underwater sound velocity and sound propagation characteristics, so obviously the mapping efficiency of SAS is less than that of SAR.  In order to break through the problem of low mapping efficiency, this paper proposes a method of sampling sparse in azimuth to obtain higher operation efficiency. The signal is transformed into SAS imaging by stripe along the range direction to solve the effect of range migration caused by the motion of carrier. And the scattering coefficient is reconstructed by one of the compressed sensing method called best orthogonal matching pursuit (CS) algorithm, in order to obtain the final SAS imaging. The simulation and experimental data imaging results show that the method can still realize the imaging without ambiguity under the condition of the enormous down sampling along the azimuth direction. It solves the problem of the grating lobe effect caused by sparse sampling, improves the mapping efficiency greatly and has strong effectiveness and practicability.
Key words: SAS, compressed sensing, sparse aperture		 　
　 　 　 　
　 Session: - SPG03-10 (Paper No.2142) 　
　 Paper Title: - A Missile-Borne Computer Design Method for AGM with Multiple-Type Warheads 　
　 Authors: - Ruping Zou, Shichao Chen and Ming Liu 　
　 Abstract: - Focusing on the problem of the designing of the missile-borne computer, which is suitable for the air-to-ground missile (AGM) with multiple-type warheads, a designing method using universal current-limiting resistance and coding pattern recognition is proposed in this paper. The current-limiting resistance suitable for different warhead missiles is determined by intersecting the resistance scopes of various current loops. The determined current-limiting resistance can achieve stable ignitions for AGMs with different warheads. The warhead type is recognized by coding pattern comparison obtained from circuit loop searching, giving instructions for the choice of the guidance model of the missile. The proposed method can avoid redundancy design and improve the robustness of the missile-borne computer for AGMs with multiple-type warheads. The effectiveness of the proposed method is validated on real collected data. 　
　 　 　 　
　 Session: - SPG03-11 (Paper No.2199) 　
　 Paper Title: - Dynamic Twiddle Factors Split-Radix Fast Fourier Transform for Monobit Receivers 　
　 Authors: - PENGFEI JI, QINGZHAN SHI and NAICHANG YUAN 　
　 Abstract: - This paper gives a new method to replace the discrete Fourier transform (DFT) for monobit receivers by using dynamic twiddle factors based on the split-radix fast Fourier transform (SRFFT) to optimize performance of monobit receivers. Two main advantages from this method can be concluded as: a. Simplify the design and calculation. b. Improve spurious-free dynamic range. From the results of the experiments in this paper, the spurious-free dynamic range of the proposed algorithm can be optimal when the number of effective bits of input signals and SNR are determined. The results are presented by simulating a 4-bit ADC with 2.5 GHz sampling rate and 1024-point SRFFT. 　
　 　 　 　
　 Session: - SPG03-12 (Paper No.2245) 　
　 Paper Title: - Selection-Channel-Aware Deep Neural Network to Detect Motion Vector Embedding of HEVC Videos 　
　 Authors: - Xiongbo Huang, Yongjian Hu, Yufei Wang, Beibei Liu and Shuowei Liu 　
　 Abstract: - It is well established that using the selection channel, the probabilities with which the elements in cover are modified during message embedding, would improve the performance of steganalysis. Most video steganographical algorithms embed secret messages in the compressed domain by modifying the motion vectors having less impact on video visual quality, which can be considered as a form of selection channel. Recently, deep neural networks have been rapidly developed for multimedia steganalysis. Although there have been some selection-channel-aware networks for image steganalysis, they cannot be simply extended to video steganalysis because there are great differences between image and video steganographic modification. To our best knowledge, there have been no selection-channel-aware networks for video steganalysis in literature. In this article, we propose a selection-channel-aware deep neural network for video steganalysis. Considering that video structure is quite different from that of image, we focus on the construction of input data matrix for deep convolutional neural network, the definition of probability for motion vector modification, and the network structure of using the selection channel knowledge. Experimental results have demonstrated that the proposed method benefits from selection channel and has satisfactory performance on testing HEVC videos. 　
　 　 　 　
　 Session: - SPG03-13 (Paper No.2256) 　
　 Paper Title: - Research on Imaging Methods of Single-beam Mechanical Scanning Sonar 　
　 Authors: - JingLi Wang, Jing Wang, Xiangkai Jia, Wenjiao Shan and Jing Guan 　
　 Abstract: - In order to solve the problem that imaging is greatly disturbed by noise in the application of TKIS series helmet-mounted colorful imaging sonar, anti-noise ability research was conducted on three imaging methods commonly used in imaging sonar systems. In this paper, based on TKIS series helmet-mounted colorful imaging sonar system, the traditional fast Fourier transform (FFT) method, orthogonal-demodulation-based method and linear frequency modulation (LFM) cross-correlation method were used for comparative analysis of anti-noise capability. The imaging experiment of the traditional FFT method has been completed, and the imaging experiment of the other two methods will be completed in the future. The results show that the orthogonal-demodulation-based method has strong noise suppression ability in all scopes of the system, and is more suitable for the upgrade of TKIS series helmet-mounted colorful imaging sonar system. 　
　 　 　 　
　 Session: - SPG03-14 (Paper No.2119) 　
　 Paper Title: - Random Fourier Features Multi-Kernel LMS Algorithm 　
　 Authors: - Wei GAO, Meiru Song, Jie Chen and Lingling Zhang 　
　 Abstract: - Multi-kernel based methods have better and more flexible performance due to more freedom degrees and united features than the single kernel methods. In this paper, we present the random Fourier multi-kernel least-mean-square (RFF-MKLMS) algorithm, and derive its analytical models in the mean and mean-square error sense to characterize the transient and steady-state stochastic behaviors. The analytical predictions consistently match with the Monte Carlo simulations in the transient and steady-state phases, which validate the accuracy of theoretical findings. 　
　 　 　 　
　 Session: - SPG03-15 (Paper No.2104) 　
　 Paper Title: - Research on infrared signal processing circuit of large size infrared touch screen with interference rejection 　
　 Authors: - Yuxin Liu, Huajie Lin, Lina Zeng and Jie Liu 　
　 Abstract: - Infrared touch screen realizes touch operation by locating touch position on display screen with occlusion of infrared light. Infrared transmission diodes and photodiodes are placed in the frame of the screen to form an equivalent infrared optical network on the surface of the screen. Large size infrared touch screen will increase the number of transmission diodes and photodiodes, as well as the scanning speed and transmission distance of infrared light, it causes signal interference increases and signal-to-noise ratio decreases. This paper presents an infrared signal processing circuit with interference rejection, which improve the signal-to-noise ratio while ensuring the signal processing speed. The infrared light is transmitted in switch mode at a fixed frequency. The infrared signal processing circuit uses resonant circuit to restore the transmitted infrared signal which has the same resonance frequency. The output of the infrared processing circuit intensity extracted by band-pass filter and switch lock-in amplifier. The experiment results show that proposed circuit has 150μs fast response time, the output voltage varies with the ambient light intensity by 1.5 percent per 1000 lux, and the circuit has well characteristic of frequency selection which limited noise bandwidth. The interference rejection is improved effectively. 　
　 　 　 　
　 　 　 　
　 　 　 　
　 Track 3: - CPT Session, 23 Aug (Sun) 9:00 - 12:30 　
　 　 　 　
　 Session: - CPT-01 (Paper No.2197) 　
　 Paper Title: - Advanced Predictive Techniques for detection of Cyber-Attacks in Water Infrastructures 　
　 Authors: - Teik Toe Teoh, Han Yi Lim and Edwin Franco Myloth Josephlal 　
　 Abstract: - Critical infrastructures such as the water industry are now more computerised than before making them susceptible to cyber-attacks. In this paper we have developed suitable predictive models for successful early detection of cyber-attacks. It also looks to find key variables in water system infrastructure that are especially prone to cyber-attacks to safeguard them effectively. We demonstrated this using the iTrust Lab's Secure Water Treatment (SWaT) dataset. We carried out data exploration and data cleaning before creating the predictive models for cyber-attack detection. Three different models were explored for the prediction of cyber-attacks which are Neural Network, Multivariate Adaptive Regression Splines (MARS) and Random Forest. We evaluated these models based on their predictive accuracy, robustness and the ratio of false-positive and false-negative results. Key variables were also determined by looking at the variable importance from the MARS and Random Forest models. Our models are able to improve the detection rate for cyber-attacks which allows the water infrastructure to quickly activate cyber-security measures to limit damage and improve security. 　
　 　 　 　
　 Session: - CPT-02 (Paper No.2183) 　
　 Paper Title: - A ZYNQ-based anti-high overload radar data recorder 　
　 Authors: - Jingchao Zhang, Fengzuo Guo and Liyan Qiao 　
　 Abstract: - As an important phase of radar test, sled test or flight test are very critical to the structure and strength of radar test equipment. The ZYNQ-based anti-high overload acceleration radar data recorder designed to solve this problem can record the test data under the complex conditions, which provides a solution for radar data recording. After simulation, the combination of its data logger and anti-high overload protection shell have been verified. The maximum stress at each position is below the yield strength or tensile strength, and has a safety factor of tens or even hundreds of times. It proves that the structure has better structural stability and safety under the high acceleration environment. The structure can bear vertical overload of 11g acceleration. 　
　 　 　 　
　 Session: - CPT-03 (Paper No.2109) 　
　 Paper Title: - Automatic Detection and Classification of Road Conditions Using Statistical Model for Autonomous Driving 　
　 Authors: - Ricky Hou 　
　 Abstract: - Autonomous car is driverless; it creates lots of new possibilities, but also imposes new challenges. A major challenge for autonomous cars is safety. The previous studies showed that road condition is one of the main factors of traffic accidents. This paper proposes a method to automatically collect the status of road conditions from a group of cars, and uses a statistical model to quantitatively classify the road conditions. The processed information is then shared with peer cars to prevent from potential dangers. The performance study showed that the proposed algorithm could effectively classify road conditions. The algorithm is particularly useful for uncertain road conditions in remote and developing areas. 　
　 　 　 　
　 Session: - CPT-04 (Paper No.2115) 　
　 Paper Title: - Miniaturized Base-station Antenna Element using Folded Structure and Meta-surface 　
　 Authors: - Chenyu Ding, Hailiang Zhu, Jingxin Ye, Borui Leng, Aihu Song and Yaoting Yan 　
　 Abstract: - This paper presents a novel dual-polarized base-station antenna element with a small metal ground and a layer of metamaterial. The radiators apply square-wave shaped structure and it has a dimensions of 83mm * 83mm * 75.2mm (0.23 λ0 * 0.23 λ0 * 0.21 λ0). The area of the ground is also reduced to 230mm * 230mm (0.63 λ0 * 0.63 λ0). The antenna is designed to work at band of 690M-960MHz, which is commonly used  for cellular communication. Simulated results show that high gain, stable radiation pattern and high port isolation can be achieved over its working band. 　
　 　 　 　
　 Session: - CPT-05 (Paper No.2141) 　
　 Paper Title: - BROAD BAND POLARIZATION CONVERTING METASURFACE FOR C AND X BAND APPLICATIONS 　
　 Authors: - Babar Kamal, Jingdong Chen, Yingzing Yin, Jian Ren and Sadiq Ullah 　
　 Abstract: - A metasurface with broadband polarization conversion is presented in this paper. The proposed reflective polarization converter unit cell is designed with a metallic pattern of square ring resonators (SRRs) over a dielectric substrate (Rogers R04003). This presented converter can achieve a 90◦ linear polarization at three operational frequencies and pure circular polarized transformation at two resonant frequencies. The mechanism of polarization conversion of the presented surface is studied through the analysis of the surface current distribution at three resonant frequencies, i.e., 5.9 GHz, 7.6 GHz, and 9.6 GHz. Simulations are performed and the results show that the presented converter under normal incidence achieves a polarization conversion ratio of 93% for broadband polarization conversion within a bandwidth of 4.4 GHz (5.6 GHz–10 GHz), which covers the C-band and a portion of the X-band. Consequently, the designed converter can be used for linear and circular polarization conversion applications in the C and X-bands. 　
　 　 　 　
　 Session: - CPT-06 (Paper No.2091) 　
　 Paper Title: - Application of Graph Database for the Storage of Knowledge Map of Power Grid Model 　
　 Authors: - Han Wu, Yujun Wang, Peng Chen, Haoqiu Shi and Tao Wu 　
　 Abstract: - The future regulation system puts forward higher requirements for the storage and management of large power data. According to the natural topological structure, dynamic change and intrinsic complex data characteristics of power grid model, the knowledge map of power grid model is stored by introducing graph database Neo4j, which promotes the evolution of dispatching automatic analysis and decision-making algorithm to knowledge guidance. The key to build knowledge map is to extract the knowledge unit of power grid model accurately. By studying the data storage mode of Neo4j, this paper proposes a general method of building knowledge map of power grid model based on Python and storing it in graph database. The feasibility of the method is verified by practical cases, and the internal relationship of model data in knowledge map is explored by using Cypher query language. 　
　 　 　 　
　 Session: - CPT-07 (Paper No.2159) 　
　 Paper Title: - Numerical Simulation and Computing the properties of radiation noise from underwater single-propeller and contra-rotating propellers 　
　 Authors: - ma fuqiang 　
　 Abstract: - In order to study the properties of radiation noise of single-propeller and contra-rotating propellers, the numerical simulation and computing was accomplished by using RNG turbulence model, LES turbulence model, FW-H acoustic analog equation and  dynamic grid model. The open water performance of single propeller and contra-rotating propellers is calculated respectively, and the calculated results are in good agreement with the published results. Then the flow field properties  of single propeller and contralateral propeller are compared and analyzed, on this basis, acoustic signature analysis and computing on properties of radiation noise from underwater single-propeller and contra-rotating propellers was accomplished, the properties of pulsating sound pressure at the monitoring point are solved. From the computing results, it can be seen that there are line spectral components related to front and rear blade frequency in the noise power spectrum, the low frequency line spectrum of contra-rotating propellers is more abundant than that of single propeller. 　
　 　 　 　
　 Session: - CPT-08 (Paper No.2160) 　
　 Paper Title: - Effect of Reflector Material on the Characteristics of UPSS Bunching Sound Field 　
　 Authors: - Xiaolong Liu, Ning Li and Hongbing Li 　
　 Abstract: - Curved reflection bunching technology is an effective method to achieve directional radiation of intense acoustic shock wave and improve the acoustic intensity in the designated area for high-power underwater intense sound source. The material properties of reflector are one of the main factors affecting the distribution characteristics of reflection bunching sound field. Based on the nonlinear finite element method, curved reflection bunching and fluid-structure interaction theory, the underwater plasma sound source (UPSS) bunching sound field model is established, and the distribution law of the underwater plasma sound source bunching sound field with different reflector material properties is numerically simulated. The influence of material properties on the performance of ellipsoidal reflector, such as bunching gain, peak pressure of bunching wave and acoustic axis distribution, is analyzed in detail, which provides guidance for further design of reflector, selection of reflector material and improvement of bunching performance parameters. 　
　 　 　 　
　 Session: - CPT-09 (Paper No.2162) 　
　 Paper Title: - Dynamic Behavior of Reflector and Its Effect on the Characteristics of Bunching Sound Field 　
　 Authors: - Xiaolong Liu, Kaizhuo Lei and Xuguang Xu 　
　 Abstract: - The nonlinear finite element method (FEM) is used to numerically simulate the characteristics of bunching sound field of underwater plasma sound sources (UPSS). Based on the acceleration histories at different positions on the outer surface of the reflector, the dynamic behavior of the ellipsoid reflector under the action of intense acoustic shock wave and its influence on the bunching gain, the peak pressure of the maximum bunching wave and the focus position are emphatically analyzed. The results show that the elastic deformation of the ellipsoid reflector increases gradually from the bottom to the outlet of the reflector. In order to improve the stress safety margin, the thickness of the reflector should be increased appropriately. By optimizing the reflector profile, the bunching gain can be effectively improved, the ellipsoid reflector can maintain a more ideal shape, and the actual focus is closer to the geometric focus. 　
　 　 　 　
　 Session: - CPT-10 (Paper No.2236) 　
　 Paper Title: - Vision-Based Metal Oxide Semiconductor Transistor-Level Layout Error Classification Using EfficientNet Model 　
　 Authors: - Lorena Ilagan, Ronnie II Concepcion, Melvin Cabatuan and Christian Roque 　
　 Abstract: - In response to the emerging challenges of providing intelligent dynamic integrated circuit (IC) layout checking, computer vision in IC design and constraint engineering highlights the opportunities of computational intelligence solutions. In this study, vision-based system architecture is integrated with deep transfer learning models to classify metal oxide semiconductor (MOS) transistor cell-level layout error using one-vs-rest (OvR) multilabel classification. Two layout errors, namely missing contact layer and excess structure around the gate, are generated using the developed tool command language (TCL) script that is executed on Synopsys platform. Missing contact layer error is furtherly subcategorized to metal 1 in place and not fully placed, off-positioned contact and its combination. Excess structure around the transistor gate is characterized by excess p-type implant (PIMP) and n-type implant (NIMP) with misaligned diffusion (DIFF) and polysilicon (PO) layers. Feature extraction for MOS-level error classification explored on using MobileNetV2 and EfficientNet variants. It was found that EfficientNetB7 best MobileNetV2 and other variants of EfficientNet in predicting IC layout errors based on nine error subcategories. Hamming loss was found to decrease and inference time to increase as the input image size is increased. The deep transfer network EfficientNetB7 has accuracy of 96.889 %, precision of 88.778 %, recall of 97.444 % and F1-score of 91.667 in predicting transistor-level layout errors. Overall, the developed approach in predicting MOS transistor cell-level layout error using integrated computer vision and deep learning proved to be accurate and easy to be replicated for further enhancement to provide advanced layout evaluation. 　
　 　 　 　
　 Session: - CPT-11 (Paper No.2249) 　
　 Paper Title: - Promising Bio-Authentication Scheme for E2E S2S in IoT-Cloud System 　
　 Authors: - Mustafa A. Al Sibahee, Songfeng Lu, Zaid Ameen Abduljabbar, Erasmus (Xin Liu), Yanli Ran, Ahmed Abdulelah Jasim Al-ashoor, Mohammed Abdulridha Hussain and Zaid Alaa Hussien 　
　 Abstract: - Document integrity and origin for E2E S2S in IoT-cloud have recently received considerable attention because of their importance in the real-world fields. Maintaining integrity could protect decisions made based on these message/image documents. Authentication and integrity solutions have been conducted to recognise or protect any modification in the exchange of documents between E2E S2S (smart-to-smart). However, none of the proposed schemes appear to be sufficiently designed as a secure scheme to prevent known attacks or applicable to smart devices. We propose a robust scheme that aims to protect the integrity of documents for each user’s session by integrating HMAC-SHA-256, handwritten feature extraction using a local binary pattern, one-time random pixel sequence based on RC4 to randomly hide authentication codes using LSB. The proposed scheme can provide users with one-time bio-key, robust message anonymity and a disappearing authentication code that does not draw the attention of eavesdroppers. Thus, the scheme improves the data integrity for a user’s messages/image documents, phase key agreement, bio-key management and a one-time message/image document code for each user’s session. The concept of stego-anonymity is also introduced to provide additional security to cover a hashed value. Finally, security analysis and experimental results demonstrate and prove the invulnerability and efficiency of the proposed scheme. 　
　 　 　 　
　 Session: - CPT-12 (Paper No.2258) 　
　 Paper Title: - Downhole robot path planning based on improved D* algorithmn 　
　 Authors: - Tian Ma, Jiahao Lv and Ming Guo 　
　 Abstract: - In order to address the possible problems of the traditional D* algorithm in the path planning process, such as the proximity to obstacles, this paper improves both the distance function and the selection of child nodes. Firstly, the traditional D* algorithm using Euclid distances was improved to Manhattan distances that are more suitable for the special environment in the mine, thus reducing the cost of computing time. Then, improve the selection of child nodes, increase the danger factor as the node attribute, and exclude the node whose danger factor is 1, thereby increasing the safety distance of the planned route and reducing the number of search points in the path planning process. The experimental results show that for the downhole roadway environment simulation map, the calculation efficiency of the method is significantly improved and the planned path is safer and reasonable. 　
　 　 　 　

　 　 　 　
　 Track 1: - COM02 Session, 23 Aug (Sun)  13:30-18:30  　
　 　 　 　
　 　 　 　
　 Session: - COM02-01 (Paper No.2143) 　
　 Paper Title: - The Performance of Relay Communication Based on Energy Harvesting 　
　 Authors: - Xuhu Wang, Shengnan Zhang and Enyu Li 　
　 Abstract: - Using the decoding-and-forwarding (DF) relay network, the problem of limited energy in wireless relay network was designed. The exact expression of outage probability and approximate expression for constant reference channel and Rayleigh channel under high SNR are derived. The exact expression of interception probability with interception node is obtained. The results and simulation show that they are correct, and the maximum diversity order is 1 when the energy collection channel is a constant parameter, while the maximum diversity order under Rayleigh channel is less than 1. The optimal time allocation factor algorithm is given to optimize the network outage performance. 　
　 　 　 　
　 Session: - COM02-02 (Paper No.2166) 　
　 Paper Title: - Electrical Specialty Experiment Teaching Reform Method with Flipped Classroom 　
　 Authors: - chengkai tang, Huajie Lin, lingling zhang, tao bao, yi zhang and Zhiyu Liu 　
　 Abstract: - In view of the low interaction between teachers and students in the existing electrical Specialty experimental teaching, the experimental teaching has become a water course. As a new mixed teaching mode based on information technology, the flipped classroom subverts the traditional classroom teaching mode, pays attention to students as the main body, and gathers many advantages in improving the quality of teaching and learning. Combined with the characteristics of the electronic experiment course system, this paper proposes the electronic experiment teaching method based on the flipped classroom. Through the flipped classroom model design before, during and after class, the two-way interaction between teachers and students in the electronic experiment teaching is realized, which effectively improves students' interest in learning, promotes students' cooperation and self-learning ability. The application results show that the flipped classroom experiment teaching method of electronic class provides a new way of thinking and exploration for the experiment teaching under the new engineering. 　
　 　 　 　
　 Session: - COM02-03 (Paper No.2168) 　
　 Paper Title: - Electrical and Electronic Experiment Platform Teaching Method With Remote Control 　
　 Authors: - chengkai tang, huajie lin, lingling zhang, tao bao, yi zhang and zhiyu liu 　
　 Abstract: - The use of modern Internet of Things, Internet and communication technology to promote the development of educational technology is the trend of the development of new engineering, and electrical and electronic experiments are difficult to conduct distance teaching due to their differences from theoretical courses. In order to improve the teaching mode of traditional electrical and electronic experiments and improve the quality of teaching, this paper proposes an electrical and electronic experimental teaching platform based on remote control, which realizes analog circuits and high-frequency circuits through the separate design of the experimental side, equipment side and server side And remote experiment control of courses such as comprehensive design of electrical and electronic, and tested on the system based on ARM microcontroller. This system combines the cutting-edge technology of electronics and communication technology and the Internet, fully embodying the tenet of "technology serves teaching". 　
　 　 　 　
　 Session: - COM02-04 (Paper No.2184) 　
　 Paper Title: - Secure Precoding Scheme for Full-Duplex MIMO Untrusted Relay Network with Energy Harvesting 　
　 Authors: - Ye Fan, Xuewen Liao, Rugui Yao, Xiaoya Zuo and Zhengyu Zhu 　
　 Abstract: - In this paper, we propose a novel secure precoding scheme for the full-duplex multiple-input multiple-output (MIMO) untrusted relay network. To ensure security, the source uses a fraction of power to send artificial noise and the destination collaboratively transmits jamming signal with the harvested energy
to the relay. Aiming at improving the power efficiency and keeping security simultaneously, we jointly maximize the achievable secrecy rate by optimizing the precoders of the information signal, jamming signal, artificial noise, amplification matrix, and power splitting ratio subject to the power constraints. Furthermore, to make the optimization problem solvable, we propose an iterative algorithm and deal with the non-convex constraints through concave convex procedure (CCCP). Simulation results show that
the proposed precoding scheme has significantly improved the secrecy performance and power efficiency compared with the typical direct link (DL) scheme and non-cooperative jamming (No-CJ) scheme.		 　
　 　 　 　
　 Session: - COM02-05 (Paper No.2188) 　
　 Paper Title: - Relays Selection Algorithm of Collaborative Communication Network on Condition of Channel Statistic Information Error 　
　 Authors: - Yong Zhang, Jun-Wei Li, Kunlin Song and Yong Jin 　
　 Abstract: - To date, an original method has been proposed for regional collaborative relays and relays selection based in cognitive radio networks. Considering the different fading characteristics between users and relays in communication links, the regional collaborative relays are introduced to optimize the allocation of powers through selecting available relays. Moreover, the  l0-norm usually is replaced with  l1-norm which is used as the closet convex approximation. Therefore, the neoteric NP-hard optimal problem can be relaxed as a tractable convex optimization problem. Therefore, a computationally efficient and near-optimal solution is obtained by an iteratively re-weighted algorithm. Simulations show that the proposed algorithm satisfied the predetermined service levels at relatively small excess transmission power in collaborative relay network scenarios. 　
　 　 　 　
　 Session: - COM02-06 (Paper No.2251) 　
　 Paper Title: - User-centric Joint Interference Rejection Method in Ultra Dense Network 　
　 Authors: - Qin haoyang, hui Li and wenjie Zhang 　
　 Abstract: - As one of the key technologies of 5G, ultra dense network (UDN) increases the system capacity of local hotspot areas by deploying wireless network infrastructure more densely. Aiming at the handover and interference of edge users, this paper adopts a user-centric transmission scheme and proposes a user-centric joint interference rejection method. In order to simplify the complicate network topology and reduce the computational complexity of the interference rejection method, a dynamic clustering algorithm based on graph theory is proposed in which a virtual cell cluster of different sizes is formed to achieve joint transmission of all micro-base stations to users in the cluster by introducing a reasonable threshold to divide the network. Besides, for intra-cluster and inter-cluster interference,a joint interference rejection algorithm that uses ZF and MMSE precoding to reject intra-cluster interference and inter-cluster MMSE precoding to further reject inter-cluster interference is proposed. The simulation results show that the proposed virtual small clustering algorithm can greatly increase the system capacity and reduce the system bit error rate (BER) and the spectrum efficiency of the system has been further improved after the interference rejection between clusters. 　
　 　 　 　
　 Session: - COM02-07 (Paper No.2113) 　
　 Paper Title: - Low-Complexity Turbo Equalization for MIMO Underwater Acoustic Communications 　
　 Authors: - Shengjun XIONG, Lisheng Zhou, Zhenduo WANG and Chao WANG 　
　 Abstract: - Underwater acoustic communication (UAC) suffers from inter-symbol interference (ISI) due to multipath propagation, multiple-input-multiple-output (MIMO) UAC also   suffers from co-channel interference (CCI) among data streams transmitted by different antennas. In this paper, we propose a low-complexity turbo equalization (LCTEQ) scheme for MIMO UAC. Turbo equalization in MIMO systems is applied to combat both inter-symbol interference (ISI) and CCI. The proposed MIMO LCTEQ structure includes a time reverse mirror (TRM), a decision feedback equalizer (DFE) and a SISO mapper/demapper, from the point of view of reducing the number of filters and the length of filters, we embed the time reverse mirror (TRM) technology into the LCTEQ scheme. In a N × M MIMO system, only N filters are needed instead of N × M filters, the residual ISI length is greatly shortened by using the focusing effect of the TRM, so the   computational complexity of the TEQ is greatly reduced. The performance of the MIMO LCTEQ has been verified through the undersea experimental in 2019. 　
　 　 　 　
　 Session: - COM02-08 (Paper No.2225) 　
　 Paper Title: - An FPGA Implementation of Orthogonal Matching Pursuit for Random Demodulation Real-time Reconstruction 　
　 Authors: - Jingchao Zhang, Sisheng Yin and Wang Liu 　
　 Abstract: - Random demodulation (RD) is one of the effective architectures utilizing compressed sensing (CS) theory to implement sub-Nyquist sampling. In this paper, we present an implementation of orthogonal matching pursuit (OMP) algorithm for random demodulation based on field-programmable gate arrays (FPGA) to facilitate real-time sub-Nyquist spectrum sensing exploiting RD. The proposed algorithm adopts an incremental QR decomposition (QRD) using Gram-Schmidt orthogonalization method to efficiently solve the least square problem （LSP） and is specified for random demodulation. The hardware implementation on the Virtex7 FPGA makes a good use of the resources to achieve a superior performance. Experiments show the proposed architecture can run at a frequency of 250MHz and reconstruct the signal within 317us for K=200, N=1000, m=40, which is applicable for scenarios requiring real-time spectrum sensing. 　
　 　 　 　
　 Session: - COM02-09 (Paper No.2226) 　
　 Paper Title: - A Blind Recovery Algorithm Based on SPG for Multiband Signals 　
　 Authors: - Jingchao Zhang, Xiangxin Zhang and Liyan Qiao 　
　 Abstract: - Multiband signal is a typical signal in the realm of modern communication, whose spectrum is the sum of several narrow bands in frequency domain. Modulated Wideband Converter (MWC) system, which is based on the emerging theory of Compressed Sensing (CS), can sample multiband signals at sub-Nyquist rate. However, classical reconstruction algorithm for MWC requires the number of carrier frequencies of the original signal in advance, which is a very difficult condition to satisfy. In this paper, a Spectral Projection Gradient (SPG) L1,1 algorithm for MWC is proposed. We use the l 1,1 norm of the matrix to measure the sparsity of the matrix, and transforms the sparse solution problem into a class of constrained extremum problems by minimizing l-norm. Then it is transformed into a linear programming problem which is solved by SPG. The sparsity is determined by assessing the numerical differential iteratively. The algorithm can realize blind reconstruction for MWC without requiring the number of carrier frequencies at the expense of minor increased complexity. Simulations demonstrate that the proposed algorithm has good reconstruction performance, which is superior to the classical Simultaneous Orthogonal Matching Pursuit (SOMP). 　
　 　 　 　
　 Session: - COM02-10 (Paper No.2252) 　
　 Paper Title: - An Improved Sparsity Adaptive CoSaMP with Regularization for Underwater Channel Estimation 　
　 Authors: - Xinyu Wu, Chengbing He, Xiangfei Dai and Yang Zhang 　
　 Abstract: - Considering the underwater acoustic channel has characteristics of  sparse, compressed sensing (CS) is used to deal with channel estimation. While for many of the existing methods, either the sparsity of channel must be required as a prior information, or the accuracy of reconstruction cannot be guaranteed. Therefore, an improved sparsity adaptive regularized compressive sampling matching pursuit (ISAR-CoSaMP) based on multiple judgment is proposed. The new algorithm can achieve an accurate estimation of the channel impulse response (CIR) in the case of unknown sparsity information. Simulation results show the ISAR-CoSaMP algorithm can provide a lower mean squared error (MSE) performance . 　
　 　 　 　
　 Session: - COM02-11 (Paper No.2182) 　
　 Paper Title: - Design of ZYNQ-Based Dynamic Configurable Optical Fiber Communication Test Equipment 　
　 Authors: - Jingchao Zhang, Chuanyue Lv and Huiping Ma 　
　 Abstract: - Under the ZYNQ platform, with the help of the high-bandwidth and low-latency characteristics of optical fiber communication, high-speed data transmission with other devices can be achieved. However, when the optical fiber communication rate needs to be changed, the traditional method needs to modify the logic and recompile the project, which is not conducive to the maintenance of the system. Therefore, this paper proposes a method of dynamic configuration of the fiber communication rate, designing the fiber communication control logic on the PL, transplanting the Linux operating system on the PS and running the fiber communication software interface. With the help of FPGA's reconfigurable technology, by dynamically loading different bitstreams during system operation, different transmission rates of optical fiber communication are realized, which increases the flexibility of the system. 　
　 　 　 　
　 Session: - COM02-12 (Paper No.2117) 　
　 Paper Title: - Topology Management for 5G Small-Cell Communications 　
　 Authors: - Yan Yang 　
　 Abstract: - This report investigates a topology management scheme based on beyond next generation mobile broadband (BuNGee) architecture to improve energy efficiency of cellular networks. The designed scheme utilizes sleep mode to deactivate base stations according to capacity usage. Performance is evaluated by a simulation model and a trade-off is found between system performance and energy saving. Results show that the topology management scheme proposed in this report can reduce 30%-50% energy consumption compared to system without topology management while maintaining the system Quality of Service (QoS). 　
　 　 　 　
　 Session: - COM02-13 (Paper No.2158) 　
　 Paper Title: - A Wideband Cluster-Based SCM Vehicle-to-Vehicle Channel Model in Roadside Scattering Environments 　
　 Authors: - Xin Chen 　
　 Abstract: - A novel cluster-based spatial channel model (SCM) for vehicle-to-vehicle (V2V) communications is proposed. This model is set up for V2V roadside scattering environments. To efficiently illustrate the real-word scenarios and describe non-stationary V2V channels, we divide all effective scatterers into three categories of clusters in terms of relative position of the scattering object. A mathematical expression of channel impulse response (CIR) is derived from an extension SCM and cluster-based models. Furthermore, the spatial and frequency statistical properties of the proposed V2V channel model are thoroughly studied. Finally, numerical simulations are presented to demonstrate the effectiveness of SCM V2V model. 　
　 　 　 　
　 Track 2: - SPG04 Session, 23 Aug (Sun)  13:30-18:30  　
　 　 　 　
　 Session: - SPG04-01 (Paper No.2243) 　
　 Paper Title: - Pileup Scintillation Pulses Processing using Expectation Maximization Deconvolution 　
　 Authors: - Zhenzhou Deng, Xin Zhao, Haijie Ding, Yushan Deng, Xiangzhu Cao and Ka-Veng Yuen 　
　 Abstract: - Radiation detectors often work at high count rates, therefore experiencing a high fraction of pileup. Pileup is the situation of two or more pulses partly or completely overlapping. Pileup effects cause that the energy of each single pulse is harder to be precisely obtained at high count rates. Consequently, it makes the energy resolution (ER) of the detector system degenerate. In this paper, we propose an expectation maximization deconvolution (EMD) method for processing pileup events and recovering the energy of each single pulse. EMD not only has non-negative property and energy conservation, but is independent of the pulses shape. We employ this method to recover pileup events accurately in iteration formation. Specifically, in this method, we first derived an event pulse function to express the average scintillation pulse, then the pileup events signal is constructed as a convolution model. Based on this convolution model, we solve the problem using EM iteration. Lastly, energies are calculated from the separated pulse after EM iteration. To evaluate the performance of EMD, we set up one gamma ray detector utilizing an LYSO crystal and a CR105 photomultiplier tube (PMT). Scintillation pulses are directly read out by a digital storage oscilloscope with high sample rates. For single pulses, measured energies using our EMD method are almost equal to that of the digital gate integrator (DGI). For pileup pulses, the ER obtained by EMD, DGI and the digital delay-line clipping (DDLC) is 16.31%, 29.05%, 19.08% at 511keV, respectively, which illustrates EMD is obviously preferable to that of well-established methods. 　
　 　 　 　
　 Session: - SPG04-02 (Paper No.2106) 　
　 Paper Title: - Modeling of High Frequency Sound Propagation Characteristics in Shallow Sea 　
　 Authors: - zhang liang, li xiaoyuan and meng chunxia 　
　 Abstract: - High frequency sound signal is common signal form of marine acoustic equipment, when the high frequency acoustic signal propagates in seawater; it not only produces reflection and refraction on the seawater interface, but also is obviously affected by the sound absorption of seawater. In order to obtain the law of high frequency sound propagation in shallow sea, this paper uses several self-condenser hydrophones and depth sensors placed vertically at equal intervals as receiving device, fixed point high frequency sound propagation experiments with different frequency and frequency bands were carried out in a shallow sea in China. Then the acoustic parameters of ocean waveguide and the sound absorption coefficient of seawater for different frequency are set up and the model of high frequency near field sound propagation in shallow sea is established by using ray acoustics. Finally, the measured sound propagation loss in the vertical direction of seawater is compared with the propagation loss predicted by the model. The analysis results show that for different frequencies and frequency bands, at different horizontal distances, the measured propagation loss values are in good agreement with the theoretical values, and the sound propagation model established in this paper can accurately describe the high frequency sound propagation characteristics of the sea area. The research method in this paper is of great significance for the offshore calibration and performance prediction of high frequency marine acoustic equipment under shallow sea conditions. 　
　 　 　 　
　 Session: - SPG04-03 (Paper No.2114) 　
　 Paper Title: - Underwater Long Baseline Positioning Algorithm based on Double-Parameter Constraint 　
　 Authors: - Haoming Li, Shefeng Yan and Lijun Xu 　
　 Abstract: - The uneven underwater terrain and the node motion lead to unpredictable changes in the arrival structure of acoustic signal, and thus the jump in the estimation of the propagation
delay of the acoustic signal as well as the “jump” of the positioning track (JPT) of the long baseline (LBL) positioning system based on the time delay estimation. None of the existing methods can eliminate the jump points completely. In this paper, a new underwater LBL positioning algorithm based on the double-parameter constraint (DPC) is proposed. On the basis of traditional LBL positioning algorithms, a trajectory correction algorithm with double judgement based on the target speed and steering speed parameters is considered. By performing cubic spline interpolation on the positioning points judged as jumps, the impact of jumps of the delay estimation on the final positioning trajectory can be eliminate greatly, thereby reducing positioning errors. Both the simulation analysis and the lake experiments verify the effectiveness of the proposed method.		 　
　 　 　 　
　 Session: - SPG04-04 (Paper No.2185) 　
　 Paper Title: - DISTRIBUTED PRINCIPAL COMPONENT ANALYSIS BASED ON RANDOMIZED LOW-RANK APPROXIMATION 　
　 Authors: - Xinjue Wang, Tiande Gao and Jie Chen 　
　 Abstract: - Distributed PCA aims to implement dimension reduction for data stored on multiple agents. The conventional distributed PCA encounters the bottleneck of computation when the dimension of local data is large. In this work, we propose a distributed PCA algorithm with local processing based on randomized methods for the star network topology (master-slave networks) with distributed row observations. Local matrix  approxi-
mation with randomized methods allows to significantly accelerate the computation with acceptable loss of precision. The results of numerical experiments show that the proposed algorithm can achieve satisfactory decomposition results with much lower computational complexity.		 　
　 　 　 　
　 Session: - SPG04-05 (Paper No.2196) 　
　 Paper Title: - An Adaptive Variational Bayesian Algorithm for Measurement Loss for Underwater Navigation 　
　 Authors: - Haoqian Huang, Jiacheng Tang and Yuanfeng Jin 　
　 Abstract: - The marine environment is complex and changeable, and it is difficult but indispensable to study the complex and time-varying environment. The measurement loss has an effect on obtaining the high accuracy navigation in-formation. This paper proposes an adaptive variational Bayesian filter (AVBF) algorithm which takes advantages of the variational Bayesian approach and Kalman filter to deal with the problems of the measurement loss. The proposed AVBF is proved in theory and verified by simulation experiments. Owing to the characteristics of the variational Bayesian approach, the higher precise state information can be acquired by the AVBF compared with traditional Kalman filter. 　
　 　 　 　
　 Session: - SPG04-06 (Paper No.2223) 　
　 Paper Title: - An Optimal Search Algorithm of the Integer Ambiguity in the Satellite Avigation Positioning 　
　 Authors: - Bingnan Pei, Nie Jiang and Tengda Pei 　
　 Abstract: - Abstract: In order to resolve the problem of too long search time caused by poor accuracy of the integer ambiguity floating-point solution, an improved search space algorithm was proposed. It was shown that the efficiency of a fixed integer ambiguity degree is greatly improved, with setting the initial search space and optimally updating of the upper and lower bounds that the algorithm works. Experimental simulations were given to show that the proposed algorithm is able to improve effectively search efficiency of the integer ambiguity, compared with the traditional LAMBDA algorithm and the VB algorithm, 　
　 　 　 　
　 Session: - SPG04-07 (Paper No.2207) 　
　 Paper Title: - A Multipath Matching Pursuit algorithm Based on Improved-Inner Product Matching Criterion 　
　 Authors: - menghang wu, FEIYUN WU, kunde yang and Tian Tian 　
　 Abstract: - Abstract—The traditional multipath matching pursuit (MMP) algorithm generally use inner product matching criteria (IPMc) to select the best atom for signal reconstruction, and often lose some important information of atoms. Therefore, any two similar atoms in the observation matrix will affect the matching process of the residual signal, reducing the quality of signal reconstruction. Using the improved inner product matching criteria (I-IPMc) improves the sensitivity of the MMP algorithm to atom values, optimizes the selection of support sets, and reduces the impact of similar atoms on the matching process. The simulation results show that under the same conditions, the MMP algorithm based on the I-IPMc has better reconstruction quality and higher signal reconstruction probability than the traditional MMP algorithm. 　
　 　 　 　
　 Session: - SPG04-08 (Paper No.2100) 　
　 Paper Title: - Target scattering coefficient measurement system and method 　
　 Authors: - Yangyang Wang 　
　 Abstract: - The traditional RCS measurement needs to meet the far-field conditions or special equipment, which has high cost and many limitations. In order to overcome these drawbacks, it is a feasible method to obtain far-field RCS through near-field measurement. In this article, we design a two-dimensional scanning system, which can obtain the near-field three-dimensional scattering coefficient image of the target. Then, we adopt a near-field to far-field transformation method based on spherical wave compensation and back projection to acquire far-field RCS. Finally, some simulation and experimental results show the effectiveness of the system and the developed approach. 　
　 　 　 　
　 Session: - SPG04-09 (Paper No.2093) 　
　 Paper Title: - Multi-channel Speech Enhancement with Multiple-target GANs 　
　 Authors: - Jing Yuan and Changchun Bao 　
　 Abstract: - In noisy scenes, speech enhancement is an important technology to improve the speech quality. In this paper, a multi-channel speech enhancement algorithm with multiple-target Generative Adversarial Networks (GANs) is proposed. Firstly, using the spatial characteristics of microphone array, the mask of target speech signal is generated by the multiple-target GAN (MT-GAN). Secondly, the mask is estimated based on complex Gaussian mixture model (CGMM), which is combined with the mask predicted by network in an iterative way to obtain a more robust speech enhancement system. Finally, the estimated mask is used to construct beamformer. Thus, the noisy speech is enhanced by the constructed beamformer. The experimental results show that compared with the reference methods, the speech quality and intelligibility of the proposed method are improved effectively. 　
　 　 　 　
　 Session: - SPG04-10 (Paper No.2096) 　
　 Paper Title: - Adaptive Beamforming for Constant Modulus Signal of Interest Based on Acoustic Vector Array 　
　 Authors: - Bingjie Yin 　
　 Abstract: - A constant modulus restoral method is presented herein to adaptively extract the signal-of-interest (SOI) with constant modulus, which is a common feature for underwater acoustic communication signals. Considering that the diagonal loading robust adaptive beamforming is required to determine the diagonal loading level adaptively in many applications, the proposed constant modulus restoral diagonal loading robust adaptive beamformer could select the diagonal loading level fully automatically by introducing temporal smoothing technique. Beyond that, it provides increased capability comparing with other robust adaptive beamformers in rejecting interferences and noise while protecting the desired signal. Furthermore, the computation cost is reduced. Simulation results are included to illustrate the performance of the proposed beamformer. 　
　 　 　 　
　 Session: - SPG04-11 (Paper No.2145) 　
　 Paper Title: - Experimental Study on Direction of Arrival Estimation of Acoustic Signals Based on Hyperbeam Cross Spectrum 　
　 Authors: - Zhang Mingwei and Chunping ZHAI 　
　 Abstract: - The direction of arrival estimation(DOA) technology plays an important role in acoustic beacon fast search and rescue at sea. In this paper,the DOA resolution of the split array beam-forming is improved by using the method of Hyper Beam-Forming(HBF) cross-spectrum. Under the condition of a certain signal-to-noise ratio, HBF cross-spectrum beam-forming method based on split array can effectively improve main lob width, reduce side-lobe effect and improve the resolution of direction finding significantly. Provide an effective judgment basis for the location estimation of acoustic beacons by using the intersection area of multi-point high-precision direction finding rays and the mean intersection. In this paper, the theoretical derivation process of the proposed method is given, and the error analysis of the azimuth estimation is carried out. The simulation and experimental results show that the side-lobe can be obviously suppressed by the split array HBF cross-spectrum method, and the accuracy and reliability of the target DOA estimation can be improved. The fusion method of the intersection area of multi-ray and the intersection point of multi-ray mean can effectively reduce the acoustic beacon search radius and improve search efficiency. 　
　 　 　 　
　 Session: - SPG04-12 (Paper No.2146) 　
　 Paper Title: - Research on uniform linear array output signal and beamforming 　
　 Authors: - Wenjiao Shan, Jing Wang, J X, Jingli Wang and Jing Guan 　
　 Abstract: - Aiming at the problems of low scanning efficiency and difficult real-time imaging of TKIS-I color helmet mechanical scanning single-beam sonar developed in the laboratory at present, research on high-frequency high-resolution multi-beam imaging sonar is proposed. Multi-beam imaging sonar can be pre-formed into multiple beams using electronic scanning technology, which can simultaneously scan and image a large range, which greatly improves the scanning efficiency of sonar. This paper will proceed from the two aspects of the output signal of the transducer array and the principle of beamforming to study some characteristics of the uniform linear array multi-beam sonar, and make a theoretical research basis for the development of high-frequency high-resolution multi-beam imaging sonar. 　
　 　 　 　
　 Session: - SPG04-13 (Paper No.2172) 　
　 Paper Title: - Fourth-order cumulant based direction finding algorithm for non-circular signals using uniform circular array with mutual coupling 　
　 Authors: - deng junwu, qiuping wang and jian xie 　
　 Abstract: - Uniform circular array has been widely used in direction finding systems due to its full azimuth and elevation coverage. However, the direction finding performance may degrade substantially in the presence of mutual coupling. Therefore, in this paper, we propose a direction finding algorithm for non-circular signals using uniform circular array under unknown mutual coupling. Based on the principle of rank reduction, the algorithm utilizes fourth order cumulant to decompose the multidimensional optimization problem into a one dimensional spectral search procedure. Without any multidimensional nonlinear search nor iterative computation, the algorithm can reduce the computational complexity effectively. Moreover, by constructing an extended fourth order cumulate matrix composed of circular covariance matrix and ellipse covariance matrix, the noncircularity of the sources has been fully utilized to improve the direction of arrival (DOA) estimation performance. According to the simulation results, the DOA estimation performance of the proposed algorithm outperforms the traditional fourth order cumulate based algorithm. 　
　 　 　 　
　 Session: - SPG04-14 (Paper No.2211) 　
　 Paper Title: - Method for estimating target orientation in single snapshot and coherent echo space 　
　 Authors: - Zhu Bibo and Guijuan Han 　
　 Abstract: - The three-dimensional imaging sonar with small physical size and high integration can detect underwater formations and form image results with centimeter-level resolution in both the forward direction and the depth direction. Due to the limitation of physical size and signal wavelength, the horizontal azimuth resolution obtained by conventional beam-form processing has become a short board for 3D high-resolution imaging. The traditional high-resolution algorithms of MVDR and MUSIC cannot adapt to the characteristics of stratum sonar echo of single snapshot, strong coherent source and a small number of array elements. These traditional high-resolution algorithms cannot guarantee the performance and stability of azimuth resolution of targets in the horizontal direction. This paper avoids the traditional operations of estimating the co-variance matrix, counting multiple snapshot data samples, and reducing the coherence characteristics of the echo space. Through the estimation process of the maximum posterior probability criterion and the iterative convergence process of the constraint equation, the method in this paper can obtain the optimal waveform estimation of the source signal in the spatial scanning orientation. The magnitude of the waveform energy represents the spatial distribution of the target bearing. After simulation analysis and experimental data processing, the algorithm can achieve high-resolution imaging of the target orientation under the single snapshot and coherent sonar echo. It has stable processing performance and good technical innovation 　
　 　 　 　
　 Session: - SPG04-15 (Paper No.2248) 　
　 Paper Title: - Underdetermined direction of arrival estimation with coprime array constrained by approximated zero norm 　
　 Authors: - Yan Wu, Cheng Wang and Jianyu Liu 　
　 Abstract: - Aiming at the problem of underdetermined DOA estimation of coprime array, a sparse reconstruction algorithm with approximated zero norm constraint is proposed. The method first performs vectorization processing on the covariance matrix of the received data of the coprime array. From the perspective of the difference coarray and through the de-redundancy processing, the output data of the virtual uniform linear array is obtained, using the sparseness of the incident signal relative to the entire airspace, the equivalent virtual signal corresponding to the virtual uniform linear array is represented by a set of overcomplete bases. Finally, a continuous smooth function is selected to approximate the zero norm and the signal is sparsely reconstructed to obtain the direction of arrival estimate. Simulation results show that this method has higher estimation accuracy and resolution than traditional DOA estimation algorithms. 　
　 　 　 　
　 Session: - SPG04-16 (Paper No.2130) 　
　 Paper Title: - A MUSIC-Type DOA Estimation Method Based on Sparse Arrays for a Mixture of Circular and Non-Circular Signals 　
　 Authors: - Jingjing Cai, Wei Liu, Ru Zong and Yangyang Dong 　
　 Abstract: - Sparse arrays have attracted a lot of interests recently for their capability of providing more degrees of freedom than traditional uniform linear arrays. For a mixture of circular and noncircular signals, most of the existing direction of arrival (DOA) estimation methods are based on various uniform arrays. Recently, a new class of DOA estimation algorithms was proposed using sparse arrays for a mixture of circular and noncircular signals. To further improve its performance, in this work, an improved MUSIC (I-MUSIC) algorithm is presented. The I-MUSIC algorithm can resolve the same number of mixed signals, and as demonstrated by simulation results, a better performance is achieved than the earlier proposed class of algorithms. 　
　 　 　 　



 

General Co-Chair
Jianguo HUANG
Northwestern Polytechnical University, Xian
Cheng-Zhong XU
University of Macau


International Advisory Committee
Nim CHEUNG (Hong Kong)
Zhi DING (USA)
Tariq DURRANI (UK)
David FENG (Australia)
Toshio FUKUDA (Japan)
B H JUANG (USA)
Mos KAVEH (USA)
Alex C KOT (Singapore)
Jian Qing LI (Macau)
Ray K. J. LIU (USA)
K. M. LUK (Hong Kong)
Zhi-Quan LUO (USA)
Wan-Chi SIU (Hong Kong)
Lawrence WONG (Singapore)

 

Technical Program Co-Chair
Jingdong CHEN, NPU, Xian
Weijia JIA, University of Macau
Shaodan MA, University of Macau
Jiandong LI, HeFei Univ of Tech.
Peter TAM, IEEE HK SEction

 

Track Co-Chair
Ray CHEUNG, CityU, HK
Fen HOU, University of Macau
Ricky LAU, CityU, HK
Liping QIAN, Zhejiang Univ of Tech
Yuan WU, University of Macau
Qunfei ZHANG, NPU, Xian

 

Special Session Co-Chair
Chengbing HE, NPU, Xian Guanghua YANG, Jinan Univ,

 

Local Arrangement Chair

Weiqiang TAN, Univ of Macau

Shengbin CAO, Univ of Macau

 

Publication Chair
Edward CHEUNG, PolyU, HK

 

Publicity Co-Chair
Zheng SHI, Jinan Univ.
Wentao SHI, NPU, Xian

He XIANG, NPU, Xian

 

Registration Co-Chair
Xiaodong CUI, NPU, Xian
Chengkai TANG, NPU, Xian

 

Student Paper Award Co-Chair
Jing Han, NPU, Xian
Peter TAM, PolyU, HK
Linlin ZHAO, Univ of Macau

 

Finance Co-chair
Chengbing HE, NPU, Xian
Y W LIU, HK CASCOM Chapter
  

Secretary
Lingling ZHANG, NPU, Xian

  

 

Sponsors

IEEE ComSoc Chapter, Macau

IEEE Xi’an Section

IEEE Hong Kong Section

Northwestern Polytechnical University, Xian

 

Technical Sponsors

IEEE HK CASCOM Joint Chapter

IEEE Xi'an SP Chapter

Faculty of Science and Technology, University of Macau

State Key Laboratory of Internet of Things for Smart City, University of Macau

Contact us

Email: ywliuhk@hotmail.com

 

 

备案号 :陕ICP备12013028号-2