Distributed Estimation for Adaptive Networks Based on Serial-Inspired Diffusion

Distributed estimation and processing in networks modeled by graphs have received a great deal of interest recently, due to the benefits of decentralised processing in terms of performance and robustness to communications link failure between nodes of the network. Diffusion-based algorithms have been demonstrated to be among the most effective for distributed signal processing problems, through the combination of local node estimate updates and sharing of information with neighbour nodes through diffusion. In this work, we develop a serial-inspired approach based on message-passing strategies that provides a significant improvement in performance over prior art. The concept of serial processing in the graph has been successfully applied in sum-product based algorithms and here provides inspiration for an algorithm which makes use of the most up-to-date information in the graph in combination with the diffusion approach to offer improved performance.