A penalty ADMM with quantized communication for distributed optimization over multi-agent systems
Chenyang Liu, Xiaohua Dou, Yuan Cheng Fan, Songsong Cheng · Kybernetika · 2023
In this paper, we design a distributed penalty ADMM algorithm with quantized communication to solve distributed convex optimization problems over multi-agent systems.Firstly, we introduce a quantization scheme that reduces the bandwidth limitation of multi-agent systems without requiring an encoder or decoder, unlike existing quantized algorithms.This scheme also minimizes the computation burden.Moreover, with the aid of the quantization design, we propose a quantized penalty ADMM to obtain the suboptimal solution.Furthermore, the proposed algorithm converges to the suboptimal solution with an O( 1 k ) convergence rate for general convex objective functions, and with an R-linear rate for strongly convex objective functions.