Decoupled Adapt-then-Combine diffusion networks with adaptive combiners

In this paper we analyze a novel diffusion strategy for adaptive networks called Decoupled Adapt-then-Combine, which keeps a fully local estimate of the solution for the adaptation step. Our strategy, which is specially convenient for heterogeneous networks, is compared with the standard Adapt-then-Combine scheme and theoretically analyzed using energy conservation arguments. Such comparison shows the need of implementing adaptive combiners for both schemes to obtain a good performance in case of heterogeneous networks. Therefore, we propose two adaptive rules to learn the combination coefficients that are useful for our diffusion strategy. Several experiments simulating both stationary estimation and tracking problems show that our method outperforms state-of-the-art techniques, becoming a competitive approach in different scenarios.