Hierarchical diffusion algorithms for distributed estimation

We study the problem of distributed estimation, where a set of nodes are required to collectively estimate some parameter of interest from their measurements. Distributed implementations avoid the use of a fusion center and distribute the processing and communication across the entire network. Among distributed solutions, diffusion algorithms have been shown to achieve good performance, increased robustness and are amenable for ad-hoc implementation. In this work we focus on hierarchical diffusion algorithms, where we allow different nodes to have different responsibilities, as opposed to our previous work where every node performed exactly the same type of operations. Our results are general in the sense that they apply to any diffusion algorithm. We illustrate the concept using diffusion LMS, provide performance analysis for hierarchical collaboration and present simulation results showing improved performance over non-hierarchical methods.