A New Method of Distributed Estimation

This paper describes a novel method of hierarchical asynchronous distributed filtering called the Net Information Approach (NIA). The NIA is a Kalman filter based estimation scheme for spatially distributed sensors which must retain their local optimality yet require a nearly optimal global estimate. The key idea of the NIA is that each local sensor dedicated filter tells the global filter "what I\´ve learned since the last local-to-global transmission," whereas in other estimation architectures the local-to-global transmission consists of "what I think now." An algorithm based on this idea has been demonstrated in a small-scale test case, with many encouraging result.