Experimental Validation of Large-Scale Distributed MIMO With Real-Time Adaptive Modulation

This paper presents the design, implementation, and experimental validation of a distributed multiple-input multiple-output (dMIMO) testbed. The testbed comprises 24 spatially distributed transmission point (TP) units, each with 2 antenna ports, centrally coordinated by a baseband unit (BBU). Orthogonal pilot signals are transmitted from all TP units, enabling each user equipment (UE) to perform downlink channel estimation and provide explicit channel state information (CSI) feedback to the BBU. The centralized unit then aggregates these channel vectors into a full CSI matrix and computes a zero-forcing precoder which is applied to both demodulation reference signals and user data. An adaptive modulation algorithm is employed that selects the modulation order for each UE in every subframe based on a pre-defined performance metric, allowing the system to adapt to varying channel conditions. Using this architecture, our testbed spatially multiplexes 17 single-antenna UEs in real time and achieves an average per-user spectral efficiency of 2.94 bits/s/Hz/UE, significantly outperforming existing implementations. All the findings are supported by extensive real-world experimental evaluations conducted using the testbed.