On optimization of the Measurement Matrix for Distributed Compressed Estimation

This paper proposes an efficient universal scheme based on the combination of compressed sensing and distributed cooperative estimation. This scheme is designed according to the application background of actual distributed estimation. An optimized measurement matrix is also presented, which can further improve the performance of the proposed actual distributed compressed estimation scheme. Simulations for a distributed sensor network based on the adapt-then-combine diffusion NLMS algorithm illustrate that the obtained measurement matrix effectively helps the proposed scheme achieving a better mean-squaredeviation performance and significantly improves convergence rate compared with other existing algorithms.