Revisiting Randomized Gossip Algorithms: General Framework, Convergence Rates and Novel Block and Accelerated Protocols

In this work we present a new framework for the analysis and design of randomized gossip algorithms solving average consensus problem. We show how classical iterative methods linear systems can be interpreted as when applied to special encoding underlying network explain in detail their decentralized nature. Our general recovers comprehensive array well-known cases, including pairwise algorithm path averaging gossip, allows development provably faster variants. The flexibility approach enables number specific methods. For instance, propose analyze novel block first xmlns:xlink="http://www.w3.org/1999/xlink">accelerated protocols, xmlns:xlink="http://www.w3.org/1999/xlink">dual algorithms. From numerical viewpoint, our is that explores depth nature proposes them evaluate performance proposed protocols by performing extensive experimental testing on typical wireless topologies.