Dynamic Node Collaboration for Mobile Target Tracking in Wireless Camera Sensor Networks Liang Liu†,‡, Xi Zhang†, and Huadong Ma‡ † Networking and Information System Laboratory Department of Electrical and Computer Engineering Texas A&M University, College Station, TX 77843, USA ‡ Beijing Key Lab of Intelligent Telecomm. Software and Multimedia, Beijing University of Posts and Telecomm., Beijing , China Study group 3/10 Jason
Outline Introduction Motion model of the target Sensing model of camera sensors Target tracking by SMC Dynamic Node Collaboration Selection of the Cluster Member Simulation Results
Introduction 2 advantages of camera tracking 1. Most informative sensor 2.High-level analysis using image processing 2 characteristics of camera tracking 1. Cooperative localization algorithm 2. Forming a dynamic cluster 1)Electing of cluster head(CH) 2)Selecting of cluster members(Cm)
Introduction 1)Electing of cluster head (CH) 2)Selecting of cluster members(Cm) 1)Electing of cluster head (CH) - SMC based tracking procedure 2)Selecting of cluster members(Cm) - Optimization-based algorithm (considering about accuracy and energy consumption)
Motion model of the target Target location Target velocity
Sensing model of camera sensors Li: Location of cameras r: Radius of sensing Vi: unit vector of Alpha: offset angle
Sensing model of camera sensors Background Subtraction
Sensing model of camera sensors
Target tracking by SMC Tracking problem consists – Posterior function – Estimating location by expectation At t-1, we’ve known
Target tracking by SMC Posterior function updating Estimating location by expectation
Dynamic Node Collaboration Motion Model of Target
Selection of Cluster Member Maximum utility: maximize the localization accuracy under the specified cost Minimum cost: minimize the cost so as to attain a specified accuracy
Simulation Results(Accuracy)
Simulation Results(Energy)