Cooperative Control using the Max-Sum Algorithm

This paper presents research on enabling cooperative decision strategies in a distributed sensor network. The paper begins with an introduction to distributed sensor networks and outlines two levels of cooperation: implicit and explicit cooperation. Implicit cooperation is the process of building and maintaining a common distributed picture where explicit cooperation is the process of negotiating to agree a common distributed plan. This paper focuses on how explicit cooperation can be achieved across a distributed sensor network by exploiting factorisation in the utility function to form a factor graph. The paper describes an efficient approach to both building and solving this factor graph using the max-sum algorithm. Experimental results are presented in a simulated sensor-to-target assignment problem.