Machine learning and intelligent communications part i 2017

The conference tracks were: MainTrack, Machine Learning; Track 1, Intelligent Positioning and Navigation; Track 2,Intelligent Multimedia Processing and Security; Track 3, Intelligent Wir

Second International Conference, MLICOM 2017

Weihai, China, August 5–6, 2017

Proceedings, Part I

226

for Computer Sciences, Social Informatics

University of Florida, Florida, USA

Xuemin Sherman Shen

University of Waterloo, Waterloo, Canada

Bo Li (Eds.)

Machine Learning

and Intelligent

Communications

Second International Conference, MLICOM 2017

Proceedings, Part I

123

Lecture Notes of the Institute for Computer Sciences, Social Informatics

and Telecommunications Engineering

https://doi.org/10.1007/978-3-319-73564-1

Library of Congress Control Number: 2017963764

© ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2018 This work is subject to copyright All rights are reserved by the Publisher, whether the whole or part of the material is concerned, speci fically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed.

The use of general descriptive names, registered names, trademarks, service marks, etc in this publication does not imply, even in the absence of a speci fic statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use.

The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication Neither the publisher nor the authors or the editors give a warranty, express or implied, with respect to the material contained herein or for any errors or omissions that may have been made The publisher remains neutral with regard to jurisdictional claims in published maps and institutional af filiations.

Printed on acid-free paper

This Springer imprint is published by Springer Nature

The registered company is Springer International Publishing AG

The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland

We are delighted to introduce the proceedings of the second edition of the 2017European Alliance for Innovation (EAI) International Conference on Machine Learningand Intelligent Communications (MLICOM) This conference brought togetherresearchers, developers, and practitioners from around the world who are leveragingand developing machine learning and intelligent communications.

The technical program of MLICOM 2017 consisted of 141 full papers in oralpresentation sessions at the main conference tracks The conference tracks were: MainTrack, Machine Learning; Track 1, Intelligent Positioning and Navigation; Track 2,Intelligent Multimedia Processing and Security; Track 3, Intelligent Wireless MobileNetwork and Security; Track 4, Cognitive Radio and Intelligent Networking; Track 5,Intelligent Internet of Things; Track 6, Intelligent Satellite Communications and Net-working; Track 7, Intelligent Remote Sensing, Visual Computing andThree-Dimensional Modeling; Track 8, Green Communication and Intelligent Net-working; Track 9, Intelligent Ad-Hoc and Sensor Networks; Track 10, IntelligentResource Allocation in Wireless and Cloud Networks; Track 11, Intelligent SignalProcessing in Wireless and Optical Communications; Track 12, Intelligent RadarSignal Processing; Track 13, Intelligent Cooperative Communications and Networking.Aside from the high-quality technical paper presentations, the technical program alsofeatured three keynote speeches The three keynote speeches were by Prof HaijunZhang from the University of Science and Technology Beijing, China, Prof YongWang from Harbin Institute of Technology, China, and Mr Lifan Liu from NationalInstruments China

Coordination with the steering chairs, Imrich Chlamtac, Xuemai Gu, and GongliangLiu, was essential for the success of the conference We sincerely appreciate theirconstant support and guidance It was also a great pleasure to work with such anexcellent Organizing Committee who worked hard to organize and support the con-ference, and in particular, the Technical Program Committee, led by our TPC co-chairs,Prof Xin Liu and Prof Mingjian Sun, who completed the peer-review process oftechnical papers and created a high-quality technical program We are also grateful tothe conference manager, Katarina Antalova, for her support and to all the authors whosubmitted their papers to MLICOM 2017

We strongly believe that the MLICOM conference provides a good forum forresearchers, developers, and practitioners to discuss all the science and technologyaspects that are relevant to machine learning and intelligent communications We alsohope that future MLICOM conferences will be as successful and stimulating, asindicated by the contributions presented in this volume

Gongliang Liu

Bo Li

Steering Committee

Steering Committee Chair

Imrich Chlamtac University of Trento, Create-Net, Italy

Steering Committee

Xin-Lin Huang Tongji University, China

Organizing Committee

General Chairs

Xuemai Gu Harbin Institute of Technology, China

Z Jane Wang The University of British Columbia, CanadaGongliang Liu Harbin Institute of Technology (Weihai), ChinaGeneral Co-chairs

Jianjiang Zhou Nanjing University of Aeronautics and Astronautics,

ChinaXin Liu Dalian University of Technology, China

Web Chairs

Xuesong Ding Harbin Institute of Technology (Weihai), ChinaZhiyong Liu Harbin Institute of Technology (Weihai), ChinaXiaozhen Yan Harbin Institute of Technology (Weihai), ChinaPublicity and Social Media Chair

Aijun Liu Harbin Institute of Technology (Weihai), ChinaSponsorship and Exhibits Chair

Chenxu Wang Harbin Institute of Technology (Weihai), ChinaPublications Chairs

Xin Liu Dalian University of Technology, China

Bo Li Harbin Institute of Technology (Weihai), ChinaPosters and PhD Track Chair

Xiuhong Wang Harbin Institute of Technology (Weihai), China

Local Chair

Bo Li Harbin Institute of Technology (Weihai), ChinaConference Manager

Katarina Antalova EAI - European Alliance for Innovation

Technical Program Committee

Technical Program Committee Chairs

Z Jane Wang University of British Columbia, Canada

Xin Liu Dalian University of Technology, China

Mingjian Sun Harbin Institute of Technology (Weihai), ChinaTPC Track Chairs

Machine Learning

Xinlin Huang Tongji University, China

Rui Wang Tongji University, China

Intelligent Positioning and Navigation

Mu Zhou Chongqing University of Posts

and Telecommunications, ChinaZhian Deng Dalian Maritime University, China

Min Jia Harbin Institute of Technology, China

Intelligent Multimedia Processing and Security

Bo Wang Dalian University of Technology, China

Fangjun Huang Sun Yat-Sen University, China

Wireless Mobile Network and Security

Shijun Lin Xiamen University, China

Yong Li Tsinghua University, China

Cognitive Radio and Intelligent Networking

Yulong Gao Harbin Institute of Technology, China

Weidang Lu Zhejiang University of Technology, China

Huiming Wang Xi’an Jiaotong University, China

Intelligent Internet of Things

Xiangping Zhai Nanjing University of Aeronautics and Astronautics,

ChinaChunsheng Zhu The University of British Columbia, CanadaYongliang Sun Nanjing Tech University, China

Intelligent Satellite Communications and Networking

Kanglian Zhao Nanjing University, China

Zhiqiang Li PLA University of Science and Technology, China

Intelligent Remote Sensing, Visual Computing, and Three-Dimensional ModelingJiancheng Luo Institute of Remote Sensing and Digital Earth,

Chinese Academy of Sciences, China

Bo Wang Nanjing University of Aeronautics and Astronautics,

ChinaGreen Communication and Intelligent Networking

Jingjing Wang Qingdao University of Science and Technology, ChinaNan Zhao Dalian University of Technology, China

Intelligent Ad-Hoc and Sensor Networks

Bao Peng Shenzhen Institute of Information Technology, ChinaDanyang Qin Heilongjiang University, China

Zhenyu Na Dalian Maritime University, China

Intelligent Resource Allocation in Wireless and Cloud Networks

Feng Li Zhejiang University of Technology, China

Jiamei Chen Shenyang Aerospace University, China

Peng Li Dalian Polytechnic University, China

Intelligent Signal Processing in Wireless and Optical Communications

Wei Xu Southeast University, China

Enxiao Liu Institute of Oceanographic Instrumentation,

Shandong Academy of Sciences, ChinaGuanghua Zhang Northeast Petroleum University, China

Jun Yao Broadcom Ltd., USA

Intelligent Radar Signal Processing

Weijie Xia Nanjing University of Aeronautics and Astronautics,

ChinaXiaolong Chen Naval Aeronautical and Astronautical University,

ChinaIntelligent Cooperative Communications and Networking

Deli Qiao East China Normal University, China

Jiancun Fan Xi’an Jiaotong University, China

Lei Zhang University of Surrey, UK

Contents – Part I

Machine Learning

An Effective QoS-Based Reliable Route Selecting Scheme

for Mobile Ad-Hoc Networks 3Jiamei Chen, Yao Wang, Xuan Li, and Chao Gao

Space Encoding Based Compressive Tracking with Wireless

Fiber-Optic Sensors 12Qingquan Sun, Jiang Lu, Yu Sun, Haiyan Qiao,

and Yunfei Hou

Moving Object Detection Algorithm Using Gaussian Mixture Model

and SIFT Keypoint Match 22Hang Dong and Xin Zhang

Low-Complexity Signal Detection Scheme Based on LLR for Uplink

Massive MIMO Channel 30Xifeng Chen, Liming Zheng, and Gang Wang

Accurate Scale-Variable Tracking 40Xinyou Li, Wenjing Kang, and Gongliang Liu

Sparse Photoacoustic Microscopy Reconstruction Based on Matrix

Nuclear Norm Minimization 49Ying Fu, Naizhang Feng, Yahui Shi, Ting Liu, and Mingjian Sun

Clustering Analysis Based on Segmented Images 57Hongxu Zheng, Jianlun Wang, and Can He

Channel Estimation Based on Approximated Power Iteration Subspace

Tracking for Massive MIMO Systems 76Liming Zheng, Donglai Zhao, Gang Wang, Yao Xu, and Yue Wu

BER Performance Evaluation of Downlink MUSA over Rayleigh

Fading Channel 85Yao Xu, Gang Wang, Liming Zheng, Rongkuan Liu,

and Donglai Zhao

Intelligent Positioning and Navigation

Privacy Protection for Location Sharing Services in Social Networks 97Hui Wang, Juan Chen, Xianzhi Wang, Xin Liu, and Zhenyu Na

A Non-line-of-Sight Localization Method Based on the Algorithm

Residual Error Minimization 103Sunan Li, Jingyu Hua, Feng Li, Fangni Chen, and Jiamin Li

WLAN Indoor Localization Using Angle of Arrival 112Zengshan Tian, Yong Li, Mu Zhou, and Yinghui Lian

Defect Detection of Photovoltaic Modules Based on Convolutional

Neural Network 122Mingjian Sun, Shengmiao Lv, Xue Zhao, Ruya Li,

Wenhan Zhang, and Xiao Zhang

An Effective BLE Fingerprint Database Construction Method

Based on MEMS 133

Mu Zhou, Xiaoxiao Jin, Zengshan Tian, Haifeng Cong,

and Haoliang Ren

Intelligent Multimedia Processing and Security

A New Universal Steganalyzer for JPEG Images 145

Ge Liu, Fangjun Huang, Qi Chen, and Zhonghua Li

Double JPEG Compression Detection Based on Fusion Features 158Fulong Yang, Yabin Li, Kun Chong, and Bo Wang

Complexity Based Sample Selection for Camera Source Identification 168Yabin Li, Bo Wang, Kun Chong, and Yanqing Guo

Wireless Mobile Network and Security

Lattice Reduction Aided Linear Detection for Generalized

Spatial Modulation 181Chungang Liu, Chen Wang, and Wenbin Zhang

Radio Frequency Fingerprint Identification Method

in Wireless Communication 195Zhe Li, Yanxin Yin, and Lili Wu

Cognitive Radio and Intelligent Networking

Short Term Prediction Models of Mobile Network Traffic Based

on Time Series Analysis 205Yunxue Gao, Liming Zheng, Donglai Zhao, Yue Wu,

and Gang Wang

Calculation Method of Field Strength in the Case of Side Obstacles 212

Lu Chen, Fusheng Dai, Yonggang Chi, and Ji Zhou

Variable Dimension Measurement Matrix Construction for Compressive

Sampling via m Sequence 221Jingting Xiao, Ruoyu Zhang, and Honglin Zhao

Signal Quality Assessment of Wireless Signal Based

on Compressed Sensing 231Fei An and Fusheng Dai

Distributed Compressive Sensing Based Spectrum Sensing Method 239Yanping Chen, Yulong Gao, and Yongkui Ma

Recent Advances in Radio Environment Map: A Survey 247Jingming Li, Guoru Ding, Xiaofei Zhang, and Qihui Wu

Elimination of Inter-distract Downlink Interference Based on

Autocorrelation Technique 258Hui Kang, Hongyang Xia, and Fugang Liu

Intelligent Internet of Things

Application of Cooperative Communications with Dual-Stations

in Wireless Mobile Environments 271Ershi Xu, Xiangping Zhai, Weiyi Lin, and Bing Chen

Design for Attendance System with the Direction Identification

Based on RFID 282Hongyuan Wang

A Geo-Based Fine Granularity Air Quality Prediction Using

Machine Learning and Internet-of-Things 291Hang Wang, Yu Sun, and Qingquan Sun

Research on Key Technology in Traditional Chinese Medicine (TCM)

Smart Service System 300Yongan Guo, Tong Liu, Xiaomin Guo, and Ye Yang

Application of Wireless Sensor Network in Smart Buildings 315Mingze Xia and Dongyu Song

Distributed System Model Using SysML and Event-B 326

Qi Zhang, Zhiqiu Huang, and Jian Xie

Intelligent Satellite Communications and Networking

A Full-Protocol-Stack Testbed for Space Network Protocol Emulation 339Xiaoqin Ni, Kanglian Zhao, and Wenfeng Li

Application Layer Channel Coding for Space DTN 347Dongxu Hou, Kanglian Zhao, and Wenfeng Li

Routing Optimization of Small Satellite Networks Based

on Multi-commodity Flow 355Xiaolin Xu, Yu Zhang, and Jihua Lu

Modeling of Satellite-Earth Link Channel and Simulating in Space-Ground

Integrated Network 364Beishan Wang and Qi Guo

A Deep Learning Method Based on Convolutional Neural Network

for Automatic Modulation Classification of Wireless Signals 373

Yu Xu, Dezhi Li, Zhenyong Wang, Gongliang Liu, and Haibo Lv

Modeling and Performance Analysis of Multi-layer Satellite Networks

Based on STK 382

Bo Li, Xiyuan Peng, Hongjuan Yang, and Gongliang Liu

Artificial-Neural-Network-Based Automatic Modulation Recognition

in Satellite Communication 394Yumeng Zhang, Mingchuan Yang, and Xiaofeng Liu

Licklider Transmission Protocol for GEO-Relayed Space Networks 405Wenrui Zhang, Chenyang Fan, Kanglian Zhao, and Wenfeng Li

Intelligent Remote Sensing, Visual Computing

and Three-Dimensional Modeling

Design of LED Collimating Optical System 417Yihao Wang, Yuncui Zhang, Xufen Xie, and Yuxuan Zhang

Global Depth Refinement Based on Patches 423

Xu Huang, Yanfeng Zhang, Gang Zhou, Lu Liu,

and Gangshan Cai

3D Surface Features Scanning System with UAV-Carried Line Laser 434Yilang Sun, Shuqiao Sun, Zihao Cui, Yanchao Zhang,

and Zhaoshuo Tian

Contourlet Based Image Denoising Method Combined Recursive

Cycle-Spinning Algorithm 444Hongda Fan, Xufen Xie, Yuncui Zhang, and Nianyu Zou

Green Communication and Intelligent Networking

A Resource Allocation Algorithm Based on Game Theory in UDN 453Changjun Chen, Jianxin Dai, Chonghu Cheng, and Zhiliang Huang

Optimal Relay Selection Algorithm for Combining Distance and Social

Information in D2D Cooperative Communication Networks 463Kaijian Li, Jianxin Dai, Chonghu Cheng, and Zhiliang Huang

Linear Massive MIMO Precoding Based on Nonlinear

High-Power Amplifier 475Xudong Yin, Jianxin Dai, Chonghu Cheng, and Zhiliang Huang

Linear Precoding for Massive MIMO Systems with IQ Imbalance 484Juan Liu, Jianxin Dai, Chonghu Cheng, and Zhiliang Huang

Research on Insurance Data Analysis Platform Based on the

Hadoop Framework 494Mingze Xia

SNR Analysis of the Millimeter Wave MIMO with Lens Antenna Array 505Min Zhang, Jianxin Dai, Chonghu Cheng, and Zhiliang Huang

Cross-Entropy Optimization Oriented Antenna Selection for Clustering

Management in Multiuser MIMO Networks 516Xinyu Zhang, Jing Guo, Qiuyi Cao, and Nan Zhao

Subcarrier Allocation-Based Simultaneous Wireless Information

and Power Transfer for Multiuser OFDM Systems 524Xin Liu, Xiaotong Li, Zhenyu Na, and Qiuyi Cao

Intelligent Ad-Hoc and Sensor Networks

A 100 MHz SRAM Design in 180 nm Process 535Zhuangguang Chen and Bei Cao

A Modified AODV Protocol Based on Nodes Velocity 545Tong Liu, Zhimou Xia, Shuo Shi, and Xuemai Gu

RSA Encryption Algorithm Design and Verification Based

on Verilog HDL 555Bei Cao, Tianliang Xu, and Pengfei Wu

A Novel High Efficiency Distributed UEP Rateless Coding Scheme

for Satellite Network Data Transmission 564Shuang Wu, Zhenyong Wang, Dezhi Li, Qing Guo,

and Gongliang Liu

A New Class of Unequal Error Protection Rateless Codes with Equal

Recovery Time Property 574Shuang Wu, Zhenyong Wang, Dezhi Li, Gongliang Liu,

and Qing Guo

Stochastic Geometry Analysis of Ultra Dense Network

and TRSC Green Communication Strategy 584Guoqiang Wang and Bai Sun

Reputation-Based Framework for Internet of Things 592Juan Chen, Zhengkui Lin, Xin Liu, Zhian Deng,

and Xianzhi Wang

Gain-Phase Error Calculation in DOA Estimation for Mixed

Wideband Signals 598Jiaqi Zhen, Yong Liu, and Yanchao Li

Mutual Coupling Estimation in DOA Estimation for Mixed

Wideband Signals 606Jiaqi Zhen, Yong Liu, and Yanchao Li

Efficient Data Gathering with Compressed Sensing Multiuser Detection

in Underwater Wireless Sensor Networks 614Rui Du, Wenjing Kang, Bo Li, and Gongliang Liu

An Efficient Data Collection and Load Balance Algorithm in Wireless

Sensor Networks 626Danyang Qin, Ping Ji, Songxiang Yang, and Qun Ding

RFID Based Electronic Toll Collection System Design

and Implementation 635Yang Li and Peidong Zhuang

Design and Implementation of Survey Vehicle Based on VR 641Weiguang Zhao and Peidong Zhuang

Development of the Embedded Multi Media Card Platform

Based on FPGA 648Songyan Liu, Ting Chen, Shangru Wu,

and Cheng Zhang

An Implementation of Special Purpose SSD Device 657Songyan Liu, Shangru Wu, Ting Chen, and Cheng Zhang

Performance Evaluation of DTN Routing Protocols in Vehicular

Network Environment 666Yongliang Sun, Yinhua Liao, Kanglian Zhao, and Chenguang He

Benefits of Compressed Sensing Multi-user Detection

for Spread Spectrum Code Design 675Yan Wu, Wenjing Kang, Bo Li, and Gongliang Liu

Application of Time-Varying Filter in Time-Frequency

Resource Allocation 682Zhongchao Ma, Liang Ye, and Xuejun Sha

Secure Communication Mechanism Based on Key Management

and Suspect Node Detection in Wireless Sensor Networks 692Danyang Qin, Songxiang Yang, Ping Ji, and Qun Ding

Research on the Pre-coding Technology of Broadcast Stage

in Multi-user MIMO System 701Guoqiang Wang and Shangfu Li

Author Index 711

Contents – Part II

Intelligent Resource Allocation in Wireless and Cloud Networks

The Application of Equivalent Mean Square Error Method in Scalable

Video Perceptual Quality 3Daxing Qian, Ximing Pei, and Xiangkun Li

Spectrum Allocation in Cognitive Radio Networks by Hybrid Analytic

Hierarchy Process and Graph Coloring Theory 8Jianfei Shi, Feng Li, Xin Liu, Mu Zhou, Jiangxin Zhang,

and Lele Cheng

Spectrum Pricing in Condition of Normally Distributed User Preference 15

Li Wang, Lele Cheng, Feng Li, Xin Liu, and Di Shen

Allocation Optimization Based on Multi-population Genetic

Algorithm for D2D Communications in Multi-services Scenario 23Xujie Li, Xing Chen, Ying Sun, Ziya Wang, Chenming Li,

and Siyang Hua

Agricultural IoT System Based on Image Processing

and Cloud Platform Technology 33Yaxin Zheng and Chungang Liu

Extension of 2FSK Signal Detection Utilizing Duffing Oscillator 43Dawei Chen, Enwei Xu, Shuo Shi, and Xuemai Gu

An Efficient DOA Estimation and Network Sorting Algorithm

for Multi-FH Signals 53Xin-yong Yu, Ying Guo, Kun-feng Zhang, Lei Li, Hong-guang Li,

and Ping Sui

Study on Correlation Properties of Complementary Codes and the

Design Constraints of Complementary Coded CDMA Systems 61Siyue Sun, Guang Liang, and Kun Wang

A Novel Structure Digital Receiver 71Zijian Zhang, Dongxuan He, and Yulei Nie

Analysis of Passive Intermodulation Effect

on OFDM Frame Synchronization 79

Yi Wang, Xiangyuan Bu, Xiaozheng Gao, and Lu Tian

Variable Tap-Length Multiuser Detector for Underwater

Acoustic Communication 87Zhiyong Liu, Yinghua Wang, and Yinyin Wang

Two-Phase Prototype Filter Design for FBMC Systems 96Jiangang Wen, Jingyu Hua, Zhijiang Xu, Weidang Lu,

and Jiamin Li

A Fine Carrier Phase Recovery Method for 32APSK 106Yulei Nie, Zijian Zhang, and Peipei Liu

Intelligent Radar Signal Processing

Interferometric-Processing Based Small Space Debris Imaging 117Yuxue Sun, Ying Luo, and Song Zhang

Sparse Representation Based SAR Imaging Using Combined Dictionary 124Han-yang Xu and Feng Zhou

Parametric Sparse Recovery and SFMFT Based M-D Parameter Estimation

with the Translational Component 132Qi-fang He, Han-yang Xu, Qun Zhang, and Yi-jun Chen

A New Radar Detection Effectiveness Estimation Method Based

on Deep Learning 142Feng Zhu, Xiaofeng Hu, Xiaoyuan He, Kaiming Li, and Lu Yang

A Novel Parameter Determination Method for Lq Regularization Based

Sparse SAR Imaging 150Jia-cheng Ni, Qun Zhang, Li Sun, and Xian-jiao Liang

Downward-Looking Sparse Linear Array Synthetic Aperture Radar 3-D

Imaging Method Based on CS-MUSIC 160Fu-fei Gu, Le Kang, Jiang Zhao, Yin Zhang, and Qun Zhang

Adaptive Scheduling Algorithm for ISAR Imaging Radar

Based on Pulse Interleaving 169

Di Meng, Han-yang Xu, Qun Zhang, and Yi-jun Chen

Direction of Arrive Estimation in Spherical Harmonic Domain Using Super

Resolution Approach 179Jie Pan, Yalin Zhu, and Changling Zhou

Adaptive Mainlobe Interference Suppression in Sparse Distributed Array

Radar Based on Synthetic Wideband Signal 188Jian Luo, Honggang Zhang, and Yuanyuan Song

Wideband MIMO Radar Waveform Optimization Based on Dynamic

Adjustment of Signal Bandwidth 198Yi-shuai Gong, Qun Zhang, Kai-ming Li, and Yi-jun Chen

Learning Algorithm for Tracking Hypersonic Targets in Near Space 206Luyao Cui, Aijun Liu, Changjun Yv, and Taifan Quan

Coherent Integration Algorithm for Weak Maneuvering Target Detection

in Passive Radar Using Digital TV Signals 215Ying Zhou, Weijie Xia, Jianjiang Zhou, Linlin Huang,

and Minling Huang

High-Resolution Sparse Representation of Micro-Doppler Signal in Sparse

Fractional Domain 225Xiaolong Chen, Xiaohan Yu, Jian Guan, and You He

Estimating of RCS of Ionosphere for High Frequency Surface

Wave Radar 233Yang Xuguang, Yu Changjun, Liu Aijun, and Wang Linwei

Intelligent Cooperative Communications and Networking

Joint Mode Selection and Beamformer Optimization for Full-Duplex

Cellular Systems 243Fangni Chen, Jingyu Hua, Weidang Lu, and Zhongpeng Wang

Construction of Emergency Communication Network with Multi

Constraints Based on Geographic Information 254Yuan Feng, Fu-sheng Dai, and Ji Zhou

Design of Turntable Servo Control System Based on Sliding Mode

Control Algorithm 263Zongjie Bi, Zhaoshuo Tian, Pushuai Shi, and Shiyou Fu

Joint Power Allocation and Relay Grouping for Large MIMO Relay

Network with Successive Relaying Protocol 273Hong Peng, Changran Su, Yu Zhang, Linjie Xie, and Weidang Lu

The Second Round

Generation of Low Power SSIC Sequences 285Bei Cao and Yongsheng Wang

Intrusion Detection with Tree-Based Data Mining Classification

Techniques by Using KDD 294Mirza Khudadad and Zhiqiu Huang

Night Time Image Enhancement by Improved Nonlinear Model 304Yao Zhang, Chenxu Wang, Xinsheng Wang, Jing Wang,

and Le Man

Research on Non-contact Heart Rate Detection Algorithm 316Chenguang He, Yuwei Cui, and Shouming Wei

Lorentzian Norm Based Super-Resolution Reconstruction

of Brain MRI Image 326Dongxing Bao, Xiaoming Li, and Jin Li

A Virtual Channel Allocation Algorithm for NoC 333Dongxing Bao, Xiaoming Li, Yizong Xin, Jiuru Yang, Xiangshi Ren,

Fangfa Fu, and Cheng Liu

A Two-Layered Game Approach Based Relay’s Source Selection

and Power Control for Wireless Cooperative Networks 343Yanguo Zhou, Hailin Zhang, Ruirui Chen, and Tao Zhou

A Novel Method of Flight Target Altitude Attributes

Identification for HFSWR 351Shuai Shao, Changjun Yu, and Kongrui Zhao

A Minimum Spanning Tree Clustering Algorithm Inspired by P System 361Xiaojuan Guo and Xiyu Liu

Transfer Learning Method for Convolutional Neural Network in Automatic

Modulation Classification 371

Yu Xu, Dezhi Li, Zhenyong Wang, Gongliang Liu, and Haibo Lv

Pulse Compression Analysis for OFDM-Based Radar-Radio Systems 381Xuanxuan Tian, Tingting Zhang, Qinyu Zhang, Hongguang Xu,

and Zhaohui Song

Implementation of Video Abstract Algorithm Based on CUDA 391Hui Li, Zhigang Gai, Enxiao Liu, Shousheng Liu, Yingying Gai,

Lin Cao, and Heng Li

Realization of Traffic Video Surveillance on DM3730 Chip 402Xin Zhang and Hang Dong

Fertilization Forecasting Algorithm Based on Improved BP

Neural Network 410Tong Xue and Yong Liu

Green Resource Allocation in Intelligent Software Defined

NOMA Networks 418Baobao Wang, Haijun Zhang, Keping Long, Gongliang Liu,

and Xuebin Li

An Algorithm for Chaotic Masking and Its Blind Extraction of Image

Information in Positive Definite System 428Xinwu Chen, Yaqin Xie, Erfu Wang, and Danyang Qin

Instruction Detection in SCADA/Modbus Network Based

on Machine Learning 437Haicheng Qu, Jitao Qin, Wanjun Liu, and Hao Chen

A Joint Source-Channel Error Protection Transmission Scheme

Based on Compressed Sensing for Space Image Transmission 455Dongqing Li, Junxin Luo, Tiantian Zhang, Shaohua Wu,

and Qinyu Zhang

Local Density Estimation Based on Velocity and Acceleration Aware

in Vehicular Ad-Hoc Networks 463Xiao Luo, Xinhong Wang, Ping Wang, Fuqiang Liu,

and Nguyen Ngoc Van

Research on Millimeter Wave Communication Interference Suppression

of UAV Based on Beam Optimization 472Weizhi Zhong, Lei Xu, Xiaoyi Lu, and Lei Wang

Global Dynamic One-Step-Prediction Resource Allocation Strategy

for Space Stereo Multi-layer Data Asymmetric Scale-Free Network 482Weihao Xie, Zhigang Gai, Enxiao Liu, and Dingfeng Yu

Machine Learning Based Key Performance Index Prediction Methods

in Internet of Industry 490Haowei Li, Liming Zheng, Yue Wu, and Gang Wang

An Auction-Gaming Based Routing Model for LEO Satellite Networks 498Ligang Cong, Huamin Yang, Yanghui Wang, and Xiaoqiang Di

Parameters Estimation of Precession Cone Target Based

on Micro-Doppler Spectrum 509MingFeng Wang, AiJun Liu, LinWei Wang, and ChangJun Yu

Automated Flowering Time Prediction Using Data Mining

and Machine Learning 518Runxuan Li, Yu Sun, and Qingquan Sun

Spatial Crowdsourcing-Based Sensor Node Localization in Internet

of Things Environment 528Yongliang Sun, Yejun Sun, and Kanglian Zhao

Influence of Inter-channel Error Distribution on Mismatch

in Time-Interleaved Pipelined A/D Converter 537Yongsheng Wang, Chen Yin, and Xunzhi Zhou

Distributed Joint Channel-Slot Selection for Multi-UAV Networks:

A Game-Theoretic Learning Approach 546Jiaxin Chen, Yuhua Xu, Yuli Zhang, and Qihui Wu

Ship Detection in SAR Using Extreme Learning Machine 558Liyong Ma, Lidan Tang, Wei Xie, and Shuhao Cai

Obtaining Ellipse Common Tangent Line Equations by the Rolling Tangent

Line Method 569Naizhang Feng, Teng Jiang, Shiqi Duan, and Mingjian Sun

On Sampling of Bandlimited Graph Signals 577

Mo Han, Jun Shi, Yiqiu Deng, and Weibin Song

Data Association Based Passive Localization in Complex

Multipath Scenario 585Bing Zhao and Ganlin Hao

Design and Implementation of Multi-channel Burst Frame Detector 595Bing Zhao

Research on Cache Placement in ICN 603

Yu Zhang, Yangyang Li, Ruide Li, and Wenjing Sun

The Digital Chaos Cover Transport and Blind Extraction

of Speech Signal 612Xinwu Chen, Yaqin Xie, and Erfu Wang

A Multi-frame Image Speckle Denoising Method Based on Compressed

Sensing Using Tensor Model 622Ruofei Zhou, Gang Wang, Wenchao Yang, Zhen Li, and Yao Xu

Frequency-Hopped Space-Time Coded OFDM over Time-Varying

Multipath Channel 634Fangfang Cheng, Jiyu Jin, Guiyue Jin, Peng Li, and Jun Mou

Dynamic Characteristic Analysis for Complexity of Continuous Chaotic

Systems Based on the Algorithms of SE Complexity and C0Complexity 647Xiaolin Ye, Jun Mou, Zhisen Wang, Peng Li, and Chunfeng Luo

Design and Implemention of an Emulation Node for Space Network

Protocol Testing 658Sichen Zhao, Yuan Fang, Wenfeng Li, and Kanglian Zhao

Optimization Spiking Neural P System for Solving TSP 668Feng Qi and Mengmeng Liu

Author Index 677

Machine Learning

Selecting Scheme for Mobile Ad-Hoc Networks

Jiamei Chen1(&), Yao Wang2, Xuan Li1, and Chao Gao2

Communication Department, Shenyang Artillery Academy,

No 31 Dongdaying Avenue, Shenhe Area, Shenyang 110161, China

Abstract In mobile ad-hoc networks, the random mobility of nodes will result

in unreliable connection In addition, the bandwidth resource limit will affect thequality of service (QoS) critically In this paper, an effective QoS-based reliableroute selecting scheme (QRRSS) is proposed to alleviate the above problems.The route reliability can be estimated by received signal strength and the controlpacket overhead can be decreased by selecting more reliable link that satisfiesthe QoS requirements Simulation results indicate that the reliable routeselecting scheme presented in this paper shows obvious superiority to the tra-ditional ad-hoc QoS on-demand routing (AQOR) in the packet successfuldelivery rate, the control packet overhead and the average end-to-end delay

Keywords: Mobile ad-hoc networks
Quality of service (QoS)
QRRSS

Reliability
AQOR

1 Introduction

This instructionfile for Word users (there is a separate instruction file for LaTeX users)may be used as a template Kindly send thefinal and checked Word and PDF files ofyour paper to the Contact Volume Editor This is usually one of the organizers of theconference You should make sure that the Word and the PDFfiles are identical andcorrect and that only one version of your paper is sent It is not possible to updatefiles

at a later stage Please note that we do not need the printed paper

We would like to draw your attention to the fact that it is not possible to modify apaper in any way, once it has been published This applies to both the printed book andthe online version of the publication Every detail, including the order of the names ofthe authors, should be checked before the paper is sent to the Volume Editors.With the development of mobile ad-hoc networks and continuous improvement ofuser demands, the limited bandwidth resource becomes difficult to guarantee high QoSfor users [1] Although such issues can get some improvement by a serial of QoSrouting algorithms [2,3] recently, no effective discussion of link reliability is available.Due to the link breakage caused by random mobility of nodes, source nodes needcontinue to trigger the route discovery process, which will lead to sharp increase in the

© ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2018

X Gu et al (Eds.): MLICOM 2017, Part I, LNICST 226, pp 3–11, 2018.

https://doi.org/10.1007/978-3-319-73564-1_1

control overhead, the probability of packet discard, and average end-to-end delay.Therefore, it will have a serious impact on the QoS We can see that under theprecondition of urgent QoS requirement, to establish a reliable end-to-end route fornodes is very important and necessary [4].

Many pertinent researches of route in mobile ad-hoc networks have been proposed.Nodes in Associative-Based Routing Protocol (ABR) measure the route reliability bysending pilot signal periodically, and meanwhile, ABR supposes that it must exist astable period after an unstable period During the stable time all nodes restart to moveafter experiencing an immobile time [5] Obviously, this supposition is opposite to thereal situation because of the random mobility of nodes in mobile ad-hoc networks LinkLife Based Routing Protocol (LBR) attains link lifetime by estimating the distance andmaximum speed of the nodes When link fails, proactive maintenance is started up torecover the route However, estimating route lifetime is invalidation owing to the linkfailure Consequently, the reliability of backup route may be hard to guarantee [6].Entropy-Based Long-Life Distributed QoS Routing Protocol (EBLLD) algorithmproposes an idea of using entropy metric to weigh the route reliability and select thelonger lifetime path, where the entropy for a route is a function about the relativepositions, velocities, and the transmission ranges of the nodes [7] Although thesealgorithms can be applied to the mobile ad-hoc networks better than the statisticalmodels, they need the premise of assumption that the relative positions all nodes areknown accurately, which is not realistic in most of the mobile ad-hoc networks.With the gradual maturation of the signal strength measurement technology, theapplication of signal strength has come to the top in domains of the control of wirelessnetworks [8], measuring distance and orientation [9] Considering that the signalstrength can reflect the connection state of the link indirectly, this paper proposes amethod of estimating route reliability based on received signal strength and establishes

an effective QoS-based reliable route selecting scheme QRRSS QRRSS selects morereliable link that satisfies QoS requirement by adding relative information to (RouteRequest, RREQ)/(Route Reply, RREP), So that it can decrease control packet overhead

by reducing frequent route discovery

2 Effective Qos-Based Reliable Route Estimation Algorithm

A mobile ad-hoc network can be depicted as an undirected graph G = (V,E) Where,

V is the set of nodes and E is the set of bidirectional links between the nodes Any link

lði; jÞ 2 E can be given by residual Bandwidth B(l), Delay D(l) and Link ReliabilityLR(l) The path from one node s to another node d can be described as

Pðs; dÞ ¼ ðs; lðs; xÞ; x; lðx; yÞ; y; ; lðz; dÞ; dÞ, where x; y; ; z are some points in thepath The connection between any two nodes is made up of a serial of all possiblepaths, which is Pðs; dÞ ¼ fP0; P1; Pi; ; Png Accordingly, we can define a certainpath Pi between s and d, whose delay, bandwidth and reliability satisfy the require-ments as (1),

DelayðPiÞ ¼ P

l2P i

DðlÞBandðPiÞ ¼ minfBðl0Þ; Bðl1Þ; BðliÞ; ; BðlnÞgReliabilityðPiÞ ¼ Q

ReliabilityðPmÞ ¼ maxfReliabilityðP0Þ; ReliabilityðP1Þ;

ReliabilityðPmÞ; ; ReliabilityðPnÞg8BandðPmÞ  Db

Table 1 The parameters and meanings in this paperParameters Meanings

RxThr Reception threshold of received signal strength, we assume it is same for all nodes

SS1i,j Current received signal strength for the link between nodes i and j

SS2i,j The received signal strength stored in neighbor information table for the link

between nodes i and j, periodically updated by SS1i,j

Thr1 If a node receives signal with strength≥ Thr1, then the link can be assumed to be

very reliable

Thr2 If a node receives signal with strength < Thr2, then the link can be assumed to be

unreliable to transfer the data

DSSi,j The difference of signal strength between nodes i and j to indicate the changes of

the signal strength

m1, m2 m1is a threshold for DSS to indicate small environment variations in signal

strength, and that m2(>m1) is used to detect whether two nodes are leaving awayfrom each other fast

LRi,j Link reliability between nodes i and j, and LRi,j2 [0, 1]

LUi,j Link uncertainty between nodes i and j, means that the link’s reliability cannot be

determined due to lack of SS2i,jin neighbor information table

As a consequence, nodes can obtain the relative parameters from received packets,and estimate route reliability with DR The packet, whose signal strength is less than orequal to Thr2, is discarded We define the route reliability and uncertainty as (3),

esti-0

1 I

0.8

0.8

0.6 S

S

ARU ARR ADELY

0.025 0.32

F 0 0.64 0.028

Pre Hop

1.0 0.8 0.8

0.8 0.8

0.8 0.5

A B C

E F

RREQ

0

B 0 ARU ARR

RRFT maintained at node C

ADELY 0.025 0.32 RREQ

RREQ

RREQ

F 0.64 0.028 RREQ

RREQ RREQ

(a) Node S broadcasts RREQ packet (b)Mediate node C processes and forwards

RREQ packet

(c)Mediate node C receives tow RREQ packets

(d)Dstination node D sends RREP packet (with boldfaced line to represent) (e)The route has established

RRFT maintained at node C

RRFT maintained at node C

0

1.0 0.8 0.8 0.8

0.8

0.8 0.5

A B C

E F

RREQ

0

1 I 0.8

0.8

0.8 0.5

D

A B C

E F

RREQ

I 1.0 0.8

B 0 ARU ARR ADELY

0.025 0.32

1.0 0.8 0.8

0.8

0.8

0.8 0.5

A B C

E F

0.8

0.8 0.5

D

C

E F

Fig 1 The principlefigure of route establishment

can see that the numbers above the links represents the current reliability of the links.The detailed route discovery process is shown as following:

(1) Firstly, the source node S broadcasts the RREQ packet (including the information

of bandwidth and delay requirements), which is shown in Fig.1(a), and sets theinitiate value of parameters as: Accumulated Delay of route, ADELY = 0;Accumulated Route Reliability, ARR = 1; Accumulated Route Uncertainty,ARU = 0 After sending the RREQ packet, S starts a timer of 3 Dmax to waitthe RREP packet

(2) As shown in Fig.1(b), mediate node C estimates the route reliability and updatesthe RREQ packet after receiving the RREQ packet Before forwarding thisreceived RREQ packet, node C sets the reverse route timer to 3 Dmax andstores relative information of RREQ into the Route Request ForwardTable (RRFT) RRFT of mediate node C has: ADELAY = 0.025, ARR = 0.32,ARU = 0 For the sake of selecting more reliable route, the RREQ packets arealso disposed during a certain time, as shown in Fig.1(c) Mediate node C re-ceives another RREQ packet from node F and registers the information as below:ADELAY = 0.028, ARR = 0.64, ARU = 0 Obviously, we can see that this routereliability is higher

In summary, if a mediate node receives an RREP packet, it firstly finds out theRRFT of relevant RREQ packet and selects a most reliable route Secondly, it estimatesthe route reliability and updates ARR and ARU of RREP packet, since ARR and ARUcan represent the current route reliability Finally, before forwarding the RREP packet,

it sets the RRFT timer to 3 Dmax and stores relative information into the route table(3) The destination node D may receive many RREQ packets from different paths,like the mediate node C And it also estimates the route reliability with the same

DR On receiving thefirst RREQ packet, node D waits a period time, called RouteReply Latency (RRL), to receive other RREQ packets andfind a more reliableroute to satisfy the QoS requirements Next, node D copies the value of QoS,ARR, and ARU to the RREP packet Simultaneously, node D sets the RRFT timer

to 3 Dmax and stores relative information into the route table, which is shown

in Fig.1(d) Eventually, node D will select the route including node F to send theRREP packet via route selecting algorithm As a consequence, the route fromsource node S to destination node D that can guarantee the QoS requirements hasbeen established, as shown in Fig.1(e)

In this section, we compare our reliable route selecting scheme to a traditionalreal-time-flow based QoS routing protocol, AQOR, which is constrained by bandwidthand delay Then, we give out the performance evaluation from packet successfuldelivery rate, control packet overhead and average end-to-end delay Packet successfuldelivery rate is the ratio of the data packets successfully received at the destinations andthe total data packets that are actually sent to the network Control packet overhead is

the ratio of the control packets sent to the network and the total data packets cessfully delivered at the destinations Average end-to-end delay is the average time ofdelivered time that all data packets have successfully arrived destinations NS2 basedsimulation gives the performance evaluation to QRRSS The simulation results areshown in Figs.2,3,4and the detailed simulation parameters are shown in Table2.

suc-The route failure is one of the most important factors affecting the packet successfuldelivery rate When the route fails, upriver nodes will store the data packets in buffersand wait until the route is established again During this time, the buffers of nodes arefilled in quickly, which will result in the subsequently discarding of the received datapackets Figure2shows the packet successful delivery rate performance of AQOR andour QRRSS at low/high load respectively We can see that QRRSS can increase thepacket successful delivery rate about 10% when the nodes move quickly, and also

0.82 0.84 0.86 0.88 0.9 0.92 0.94 0.96 0.98 1

Fig 2 Packet successful delivery rate

Table 2 The parameters and values in the simulation

significantly improve the delivery performance of the whole network The reason isthat by establishing reliable end-to-end route connection, QRRSS can effectively avoidthe data packets discarded extensively due to the route failure, no matter in low or highload environment.

From Fig.3, it can be seen that the packet control overhead in QRRSS has reducesand especially in high load and nodes moving fast it reduces nearly 12% The reasonseems to be obvious, destination node in AQOR will send many RREP replies so thatsource node can select a most optimization route, but at the same time it will lead to thecontrol overhead increasing With contrast to the AQOR, QRRSS not only increases

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5

Fig 3 Control packet overhead

0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045

Fig 4 Average end-to-end delay

the route reliability and reduces the ratio of route failure, but also reduces the routeoverhead indirectly from some kind of degree.

From Fig.4, we can observe that the average end-to-end delay of AQOR andQRRSS are both not up to 0.04 s, and obviously, QRRSS has better delay performancethan AQOR That is because the algorithm sets the link uncertaintyðLUi ;jÞ and otherparameters to different values under different conditions, which makes QRRSS canguarantee the route reliability to some extent and decrease the probability of routerediscovery

QRRSS proposed in this paper selects more reliable route connection that can antee the QoS requirements by adding relative information to RREQ/RREP Thescheme does not depend on the orientation equipments like GPS and the mobilitymodel of network nodes Simulation results indicate that QRRSS shows obviousperformance improvements with contrast to traditional AQOR in packet successfuldelivery rate, control overhead and average end-to-end delay

guar-Acknowledgments This research was supported by National Natural Science Foundation ofChina (Grant No 61501306), Liaoning Provincial Education Department Foundation (Grant

3 Foster, I., Kesselman, C.: The Grid: Blueprint for a New Computing Infrastructure MorganKaufmann, San Francisco (1999)

4 Czajkowski, K., Fitzgerald, S., Foster, I., Kesselman, C.: Grid information services fordistributed resource sharing In: 10th IEEE International Symposium on High PerformanceDistributed Computing, pp 181–184 IEEE Press, New York (2001)

5 Foster, I., Kesselman, C., Nick, J., Tuecke, S.: The physiology of the grid: an open gridservices architecture for distributed systems integration Technical report, Global GridForum (2002)

6 National Center for Biotechnology Information.http://www.ncbi.nlm.nih.gov

7 Zhai, C., Zhang, W., Mao, G.: Cooperative spectrum sharing between cellular and ad-hocnetworks IEEE Trans Wirel Commun 13(7), 4025–4037 (2014)

8 Song, Y., Xie, J.: BRACER: a distributed broadcast protocol in multi-hop cognitive radioad-hoc networks with collision avoidance IEEE Trans Mob Comput 14(3), 509–524(2015)

9 Laursen, A.L., Mousten, B., Jensen, V., Kampf, C.: Using an Ad-Hoc corpus to write aboutemerging technologies for technical writing and translation: the case of search engineoptimization IEEE Trans Prof Commun 57(1), 56–74 (2014)

10 Rios, M.: Variable route expiration time based on afixed probability of failure for ad-hocnetworks routing applications IEEE Latin Am Trans 13(1), 383–389 (2015)

11 Conti, M., Giordano, S.: Mobile ad-hoc networking: milestones, challenges, and newresearch directions IEEE Commun Mag 52(1), 85–96 (2014)

12 Khalili-Shoja, M.R., Amariucai, G.T., Wei, S., Deng, J.: Secret common randomness fromrouting metadata in ad-hoc networks IEEE Trans Inf Forensics Secur 11(8), 1674–1684(2016)

13 Haque, I.T.: On the overheads of ad-hoc routing schemes IEEE Syst J 9(2), 605–614(2015)

14 Bello, L., Bakalis, P., Rapajic, P., Anang, K.A.: Optimised adaptive power on-demandrouting protocol for mobile ad-hoc wireless network IET Netw 3(4), 245–251 (2014)

15 Celimuge, W., Ji, Y., Liu, F., Ohzahata, S., Kato, T.: Toward practical and intelligent routing

in vehicular ad-hoc networks IEEE Trans Veh Technol 64(12), 5503–5519 (2015)

16 Wang, Z., Chen, Y., Li, C.: PSR: a lightweight proactive source routing protocol for mobilead-hoc networks IEEE Trans Veh Technol 63(2), 859–868 (2014)

17 Tang, F., Guo, M., Guo, S., Cheng-Zhong, X.: Mobility prediction based joint stable routingand channel assignment for mobile ad-hoc cognitive networks IEEE Trans Parallel Distrib.Syst 27(3), 789–802 (2016)

with Wireless Fiber-Optic Sensors

Qingquan Sun1(B), Jiang Lu2, Yu Sun3, Haiyan Qiao1, and Yunfei Hou1

Abstract This paper presents a distributed, compressive multiple

tar-get localization and tracking system based on wireless fiber-optic sensors.This research aims to develop a novel, efficient, low data-throughput mul-tiple target tracking platform The platform is developed based on threemain technologies: (1) multiplex sensing, (2) space encoding and (3) com-pressive localization Multiplex sensing is adopted to enhance sensingefficiency Space encoding can convert the location information of multi-target into a set of codes Compressive localization further reduces thenumber of sensors and data-throughput In this work, a graphical model

is employed to model the variables and parameters of this tracking tem, and tracking is implemented through an Expectation-Maximization(EM) procedure The results demonstrated that the proposed system isefficient in multi-target tracking

sys-Keywords: Human tracking·Multiplex sensing

Indoor environments monitoring has been demanded in many areas The cations include human counting, tracking, identification, activity recognition,and situation perception, etc The purposes are to provide secure and intelligentworking and living spaces to users through the surveillance of the environments.Among these applications, human tracking is a very challenging but interestingapplication, and is receiving more and more attentions Traditional human track-ing systems in indoor environments are based on video cameras Such systemshave been widely applied due to its visual characteristic [1] Nowadays, somewireless sensor based human tracking systems have been developed and demon-strated with a satisfied performance especially under severe conditions such aspoor illumination, low computation, disguise, and so on

appli-The wireless sensor based human tracking systems are advantageous in (1)large surveillance area; (2) low data throughput; (3) robustness; (4) multiple

c

 ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2018

X Gu et al (Eds.): MLICOM 2017, Part I, LNICST 226, pp 12–21, 2018.

Whatever sensor is used to form a human tracking system, the goals are toimplement low-data-throughput and energy-efficient sensing Recently compres-sive sensing technology has been proposed and applied in image processing andinformation retrieval [6,7] It has been proved that compressive sensing can fur-ther reduce the data samples but still guarantee the successful reconstructions.Inspired by this technique, we propose a wireless sensor based human trackingplatform using compressive sensing Furthermore, we extend compressive sensingconcept from data processing to sensing mode and sampling geometry, namely,

we start compress measurements in sensing and sampling phases

Other than the typical wireless sensor based human tracking systems, mainlythe PIR sensor based systems, in this paper, we propose to use a new sensingmodality, fiber-optic sensors to implement human tracking Compared with PIRsensors, fiber-optic sensors are more suitable to human tracking By adoptingmultiplex sensing, space encoding and compressive localization, the sensing effi-ciency and data compression are enhanced The multi-target tracking is achievedthrough a graphical model and expectation-maximization (EM) approach

As we know, sensing is the process that converting physical information intosignals that can be read and observed by an instrument The fiber-optic sensorscan be used to convert the presence and pressure information of targets intolight intensities to enable localization and tracking Multiplex sensing technique

is inspired by the antenna of insects which is able to increase the utilization ratio

of single sensor cells Here, in our system, we employ multiplex sensing to enableeach fiber-optic sensor to detect multiple regions rather than just one region Inthis way, all the sensors can be fully utilized and the number of sensors neededcan be reduced dramatically Such a method can improve the sensing efficiencybut at a price of increasing ambiguities in localization The fiber-optic sensingformats are shown in Fig Compared with simplex sensing (Fig.1(a)), multiplexsensing (Fig.1(b), (c)) consumes less sensors to cover the same size regions

Space encoding is to segment the monitored area into different blocks and use acertain sensors to encode each block Thus, when a target appears in a certainblock, the corresponding code indicates the target’s location The purpose ofusing space encoding technology is to enhance the feasibility and efficiency ofmonitoring Fiber-optic sensors are appropriate for space encoding due to itsflexibility and detection modality There are multiple space encoding schemessuitable for fiber-optic sensors The ideal encoding scheme is named decimalencoding, in which a single block is encoded by only one sensor Apparently, thisencoding scheme is able to get a high accuracy with a minimum of ambiguity Thenumber of sensors, however, could be very large for a wide area In comparison,binary encoding scheme can reduce the sensor consumption dramatically Forexample, encoding a 4 blocks area, decimal encoding scheme needs 4 sensors,while binary encoding scheme only needs 2 sensors, as shown in Fig.2

Supposen fiber-optic sensors are available in the system, and they are used to

monitor a space which is divided intom blocks γ = {γ1, γ2, · · · , γ m } Each block

Fig 3 Illustrations of space encoding for (a) one target case; (b) two targets case.

γ i is encoded byn fiber-optic sensors, and the corresponding code will be a

n-bit binary string, represented by C i ={c i1 , · · · , c in }, as shown in Fig.3 c ij isgenerated when a target presents inj th block, so

whereI(·) is a logic function whose output is “0” or “1”; Ω is the sampling

geom-etry of sensori; ϕ(r) is the target at location r; and ∩ represents bit-wise AND

operation Therefore, withn fiber-optic sensors deployment, the observation area

is encoded into a set ofn-bit codes.

When only one target presents within the observation area, the measurement

y, which is a n × 1 vector, is given by

whereC = [c ij]T, which is an × m matrix, and x1=I(r ∈ γ), which is a m × 1

binary vector with only one ‘1’ element

WhenK targets present within the observation area, the measurement n × 1

if the upper bond is 1 andI is a matrix with only ‘1’s The example of the binary

measurement sequence for one and two targets cases are shown in Fig.3

The complexity of the compressive localization for multiple targets comes fromthe bit-wise OR operation in Eq.3 To localize K targets with small errors, it

requires a high degree of independence among the codes However, an increase

of the independence will lead to an increase of sensors

Given the space codes matrixH, the binary compressive localization problem

is solved by [8]

ˆ

x = argmin

where y is the binary measurement For simplicity purpose, the nonlinear

con-straint,y = H  x, can be replaced by a linear constraint, y = HX by rounding

the real number valued solution to a binary vector Alternatively, the constraintcan be further replaced by the binary compressive sensing constraint,y = H ⊕x.

The original problem is finalized as

ˆ

x = argmin

x x1 s.t y = [H2I][x; z] (5)whereI is the identity matrix and z > 0 is an auxiliary vector.

The selection of two solutions is determined by the number of targets andthe code matrix

Multi-target Tracking

Multiple target tracking is a challenging issue due to the involvement of a bunch

of unknown variables and complex conditions With different characteristics ofthese variables in multi-target tracking systems, the system models under variousconditions can be summarized to:

Fig 4 Multi-target tracking model with unknown number of false alarms.

Case 1 - known data-to-target association

Case 2 - unknown data-to-target association

Case 3 - unknown tracker-to-tracker association

Case 4 - unknown detection failures

Case 5 - unknown false alarms

Case 6 - varying number of targets

t , · · · , z m

t ) denotesm observations at time

t, which are related and dependent upon X t The hidden variables are given asfollows:

V t data-to-tracker association matrix

W t tracker-to-tracker association matrix

κ t number of targets

τ t number of detectable targets

ζ t number of false alarms

The first case is the simplex tracking model, in which correct data-to-trackerassociation can be achieved Specifically, the k th tracker X k

t is associated withmeasurement Z k

t correctly, and the current states of trackers can be associatedwith previous states of the same trackers correctly As for such cases, the mul-tiple targets can be tracked with a high accuracy While for other cases, if thedata-to-tracker association, tracker-to-tracker association, or detection failure isunknown, then the tracking model becomes more complicated and correspond-ingly the tracking error will be larger In this work, we establish a more compli-cated tracking model to investigate the case that the false alarms are unknown.The system model is shown in Fig.4 For the cases of unknown false alarms,the number of false alarms is denoted as ζ t, which is a Poisson random variablewith an average value of λ The location of false alarms yields a uniform dis-

tribution with a density value of O1, whereO is the volume of the observation

space All the false alarms belong to a clutter trackerX0; hence, the dimension

of the association matrixV becomes m t × (K + 1) Assuming the measurements

are reordered such that

z j | ∈ [m t − ζ(V t) + 1, mt] (6)where z j is a false alarm, then the clutter tracker model is given by

whereK − τ tcolumns of the association matrix,V , are all-zero vectors.

The joint probability density function ofX, Z, V , W , τ, ζ is given by

The challenge of multi-target tracking is that some hidden variables exist inthe sequential estimation and prediction process such as the number of detectedtargets, the number of trackers, the number of false alarms, and data-to-targetassociation LetH represent all the hidden variables, then the general solution

can be obtained by using Expectation-Maximization (EM) optimization

1 E-step: estimate the distribution of hidden variables from the predicted

tar-get state, ˆx t, and measurements, z, by conditioning the joint distribution, p(H, x, z), which is represented by

p(H|ˆx, z) = p(H), ˆx, z

Σ H p(z|ˆx, H)p(ˆx|z, H)p(H) (12)

2 M-step: estimate the distribution of the target state,x, from measurements,

z, by marginalizing hidden variables, H, that is

it is easy to remove the false alarms introduced by the scheme itself Although