Bounded-Latency Change Detection of Channel Covariance Matrix in Real-Time MIMO Communications

The accuracy of the channel covariance matrix (CCM) critically determines MIMO system performance. While CCM typically evolves gradually, abrupt changes may occur in practice, necessitating immediate detection for timely updates. Conventional detectors risk unbounded delays under minor variations, violating protocol latency constraints. To address this challenge, we propose a transient change detection (TCD) framework that integrates fixed-sample-size (FSS) sampling with optimized window configuration, ensuring strict detection deadlines while minimizing missed detection probabilities. Experimental results show significant reduction in detection delay and improved computational efficiency while maintaining high accuracy compared to benchmarks, demonstrating its efficacy for real-time MIMO applications.