Binary Consensus With Soft Information Processing in Cooperative Networks

In this paper, the authors consider reaching binary consensus over a network with AWGN channels. They consider the case where knowledge of the corresponding link qualities is available at every receiving node. They propose novel soft information processing approaches to improve the performance in the presence of noisy links. They characterize the performance and derive an expression for the second largest eigenvalue. They show that soft information processing can improve the performance drastically. They furthermore show that, by statistically learning the voting patterns, they can solve the undesirable asymptotic behavior of binary consensus.