Distributed Multi-sensor Target Tracking

The RFS-based multi-target tracking algorithms introduced in the previous chapters are mainly aimed at a single sensor. The recent advances in sensor networking technology have given rise to the development of large-scale sensor networks composed of interconnected nodes (or agents) with sensing, communication and processing capabilities. Given all sensor data, one of the ways of establishing the posterior density of the multi-target state is to send all measurements from all sensing systems to a central station for fusion. Although this centralized scheme is optimal, it is required to send all measurements to a single station, which may result in a heavy communication burden. Moreover, since the entire network will stop working once the central station fails, this method makes the sensor network vulnerable. Another method is to treat each sensing system as a node in the distributed system. These nodes collect and process measurements locally to obtain local estimates, and then these local estimates (rather than original measurements) are periodically broadcast to other nodes for data fusion across the entire network. This method is referred to as the distributed fusion.