Robust and Low-Overhead Resource Allocation for Latency-Sensitive Cell-Free Downlink

The growing demand for latency-sensitive applications such as industrial control and real-time monitoring calls for transmission strategies that ensure both low latency and high reliability in cell-free networks. However, this goal is hindered by two main challenges: uncertainty in channel state information (CSI) and significant interaction overhead between access points (APs) and the central processing unit (CPU). To address these challenges, this paper proposes a robust and low-overhead transmission framework that guarantees user quality-of-service (QoS) requirements with minimal bandwidth consumption. The proposed scheme leverages over-the-air superposition of locally computed channel Gram matrices from APs, which enables the CPU to reconstruct the global matrix for robust beamforming and joint power-bandwidth allocation. Simulations show that the proposed low interaction resource allocation (LIRA) algorithm achieves bandwidth efficiency comparable to traditional semidefinite relaxation (SDR) methods, exhibits strong robustness against channel estimation errors, and substantially reduces computation time and AP–CPU interaction overhead, making it a compelling solution for latency-sensitive cell-free networks.