Distributed Multi-Target Tracking and Identity ManagementSonghwai Oh, Inseok Hwang, and Shankar SastryThe problem of tracking multiple targets and managing their identities in sensor networks is considered. Each sensor is assumed to have its own surveillance region, and an ability to communicate with its neighboring sensors. We propose a scalable distributed multi-target tracking and identity management (DMTIM) algorithm that can track an unknown number of targets and manage their identities efficiently in a distributed sensor network environment. DMTIM finds a globally consistent solution by maintaining local consistency among neighboring sensors. DMTIM consists of data association, multi-target tracking, identity management, and identity and track fusion. The data association and multi-target tracking problems are efficiently solved by Markov chain Monte Carlo data association (MCMCDA) which can track an unknown number of targets. DMTIM manages identities of targets based on the identity-mass-flow framework. This framework prevents exponential growth in computation and storage of target-track association probabilities. Using identity and track fusion, DMTIM maintains consistent identities and tracks among neighboring sensors. The performance and features of DMTIM are extensively evaluated in simulation.Journal of Guidance, Control, and Dynamics
|
|
|