Two-Timescale Sum-Rate Maximization for Movable Antenna Enhanced Systems

2025-09-04

AI Summary

This paper addresses the downlink multi-user MIMO system enhanced by movable antennas (MAs), aiming to maximize the average achievable sum rate via dual-timescale optimization leveraging both instantaneous and statistical channel state information (CSI). Method: We propose a joint optimization framework for antenna positions and transmit covariance matrices: antenna placement is optimized over the slow timescale using statistical CSI, while precoding is designed over the fast timescale based on instantaneous CSI. A planar-constrained MA mobility pattern is introduced, and a low-complexity primal-dual decomposition-based stochastic successive convex approximation (PDD-SSCA) algorithm is developed, with theoretical guarantees of almost-sure convergence to a KKT point. Results: Simulations demonstrate that the proposed scheme significantly outperforms fixed-antenna baselines and existing MA designs in both sum rate and feasibility robustness.

Technology Category

• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems
• Search and Optimization: Mixed Discrete/Continuous Search
• Multiagent Systems: Multiagent Planning

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Security and Privacy: Large-scale security measurements

Abstract

This paper studies a novel movable antenna (MA)-enhanced multiuser multiple-input multiple-output downlink system designed to improve wireless communication performance. We aim to maximize the average achievable sum rate through two-timescale optimization exploiting instantaneous channel state information at the receiver (I-CSIR) for receive antenna position vector (APV) design and statistical channel state information at the transmitter (S-CSIT) for transmit APV and covariance matrix design. We first decompose the resulting stochastic optimization problem into a series of short-term problems and one long-term problem. Then, a gradient ascent algorithm is proposed to obtain suboptimal receive APVs for the short-term problems for given I-CSIR samples. Based on the output of the gradient ascent algorithm, a series of convex objective/feasibility surrogates for the long-term problem are constructed and solved utilizing the constrained stochastic successive convex approximation (CSSCA) algorithm. Furthermore, we propose a planar movement mode for the receive MAs to facilitate efficient antenna movement and the development of a low-complexity primal-dual decomposition-based stochastic successive convex approximation (PDD-SSCA) algorithm, which finds Karush-Kuhn-Tucker (KKT) solutions almost surely. Our numerical results reveal that, for both the general and the planar movement modes, the proposed two-timescale MA-enhanced system design significantly improves the average achievable sum rate and the feasibility of the formulated problem compared to benchmark schemes.

Problem

• Maximizing sum rate in movable antenna systems
• Optimizing antenna positions using two-timescale information
• Enhancing wireless performance with efficient movement algorithms

Innovation

• Two-timescale optimization for sum-rate maximization
• Gradient ascent and CSSCA algorithms for suboptimal solutions
• Planar movement mode with PDD-SSCA for efficiency