Traffic Influence in Distributed Networks Using Cooperation and Set-Membership Filtering

This paper investigates the influence of data loss in the performance of cooperative adaptive filters in distributed networks. The algorithms analyzed are those with and without data selection based on innovation. Our simulation results indicate that set-membership adaptation algorithms, which perform some form of innovation check prior to transmission of data to the neighbors, have better performance than their counterparts which flood the network with data at every iteration. Therefore the space-time data selection of set-membership adaptive filters reduces computational complexity and energy consumption, and also improves convergence performance in case of data loss during transmission.