Design and application of consensus-based filtering for Local Strongly Coupled & constrained systems

In the brand new era of “big data”, finding creative strategies for information fusion in a complicated system is important. Previous researchers have introduced a model called “Local Strongly Coupled System”, in which every agent stochastically communicates only with its neighbours (i.e. “coupled” nodes). However, the meaning of a system's intrinsic a priori constraints is always neglected. And we assume that in practical instances, the whole system is supposed to take on a consistent result, or “consensus” decision. This paper, taking constraints into account, presents a new strategy to help minimize the filtering error. Furthermore, based on consensus policy, the approach is applied to local strongly coupled systems, especially systems with packet loss. Effectiveness and practicability of all the proposed algorithms are shown through simulations.