Asymmetric average consensus under SINR-based interference

Consensus algorithms are a family of distributed processes that are based on exchanging local information in order to obtain some particular global information. An example of these algorithms is the average consensus, in which the value to be obtained is the average of some initial data. Most of the existing consensus techniques assume unrealistic models of communications that require complex control mechanisms in practice. In contrast, we consider the average consensus algorithm under a realistic asynchronous and asymmetric scheme of communications, where the interferences constrain the information exchanged among the nodes. To ensure a correct operation in this scheme, we propose a link scheduling protocol that satisfies certain convergence conditions and maximizes the number of simultaneous links in each iteration of the consensus algorithm. This increase in the number of communications per iteration improves the performance of the consensus algorithm. Simulation results are presented to verify and clearly show the efficiency of our approach.