Network Coding Based Distributed Indoor Target Tracking Using Wireless Sensor Networks
- 144 Downloads
We consider the problem of indoor target tracking using wireless sensor networks. To facilitate ease of deployment and keeping the cost to a minimum we focus on devising a target tracking system based on received signal strength indicator (RSSI) measurements. We adopt a model based approach in which the targets are assumed to evolve in time according to a certain maneuver model and the deployed sensors record RSSI measurements governed by an appropriate observation model. To devise an accurate target tracking algorithm, that would account for the radio environment, we use mixed maximum likelihood (ML)-Bayesian framework. Under this framework the radio environment is estimated using the ML approach and the target tracking is accomplished using a Bayesian filtering technique namely, particle filtering. Next to create a distributed tracking algorithm which warrants that every sensor node has access to RSSI measurements of all the other sensor nodes we introduce a dissemination mechanism for the same based on the technique of random linear network coding (RLNC). In this technique every sensor node encodes RSSI measurements that it has received from other nodes (including its own) to create a network coded packet, which in turn is transmitted using the carrier-sense multiple access based access mechanism. Our simulation results demonstrate that the root mean square tracking error (RMSE) obtained by using RLNC is strictly lower than what was achieved with a competing scheme based on localized aggregation. This can be attributed to the rapid dissemination capability of the RLNC technique. Further, the growth of RMSE in a strongly connected network with noise variance was found to be much slower than in the case of a weakly connected network. This points to the potential of RLNC in improving tracking performance, especially in strongly connected networks.
KeywordsWireless sensor networks Network coding Bayesian filtering and EM algorithm
Funding was provided by Science and Engineering Research Board (Grant No. SR/FTP/ETA-85/2009).
- 3.Kushki, A., & Plataniotis, K. (2010). Nonparametric techniques for pedestrian tracking in wireless local area networks. In: S. Haykin, K. J. R. Liu (Eds.), Handbook on Array Processing and Sensor Networks (pp. 783–803). doi: 10.1002/9780470487068.ch25.