Decentralized Alternating Direction Method of Multipliers for Optimal Analog Transceiver Performance

Abstract

This research introduces a Decentralized Alternating Direction Method of Multipliers (D-ADMM) model for optimizing analog transceivers with modulus constraints. Addressing the limitations of centralized and distributed ADMM approaches, the proposed D-ADMM model achieves a balanced trade-off among transceiver performance and hardware complexity. The optimization process involves iterative updates at each node and local fusion centers, ensuring consensus in a decentralized network. Unit modulus constraints are imposed on analog transceivers to enhance hardware feasibility. Results are discussed using spectral efficiency, bit error rate, and error rate. Simulation results demonstrate the D-ADMM's effectiveness in large-scale networks without a global fusion center. The decentralized optimization, through an iterative approach, proves its capability to handle non-convex problems and attain optimal solutions. Comparative results illustrate the D-ADMM's superiority over traditional methods, showcasing its potential for advancing analog transceiver optimization in communication networks.