Active Consensus over Sensor Networks via Randomized Communication

TL;DR

This paper introduces an energy-efficient distributed consensus method for sensor networks that adaptively selects links to balance convergence speed and energy consumption, enhancing robustness to link failures.

Contribution

It proposes a novel randomized link selection algorithm that optimizes consensus efficiency and energy use through local approximations and quadratic programming relaxation.

Findings

Reduces communication energy costs significantly

Maintains convergence rate despite link failures

Demonstrates robustness across different network types

Abstract

Distributed consensus has been widely studied for sensor network applications. Whereas the asymptotic convergence rate has been extensively explored in prior work, other important and practical issues, including energy efficiency and link reliability, have received relatively little attention. In this paper, we present a distributed consensus approach that can achieve a good balance between convergence rate and energy efficiency. The approach selects a subset of links that significantly contribute to the formation of consensus at each iteration, thus adapting the network's topology dynamically to the changes of the sensor states. A global optimization problem is formulated for optimal link selection, which is subsequently factorized into sub-problems that can be solved locally, and practically via approximation. An algorithm is derived to solve the approximation efficiently, using quadratic programming (QP) relaxation and random sampling. Simulations on networks of different types demonstrate that the proposed method reduces the communication energy costs without significantly impacting the convergence rate and that the approach is robust to link failures.