Many-antenna MU-MIMO, or "massive" MIMO at large scale, is a key candidate technology for next-generation wireless systems. However, from a practical design perspective scaling up MU-MIMO presents a number of unique challenges and opportunities.

To efficiently utilize computational, power, and channel resources requires a complete redesign of many aspects of traditional MIMO systems, particularly the base station architecture and control channel. Furthermore, mobility fundamentally limits the capacity of many-antenna MU-MIMO systems, thus to efficiently realize practical many-antenna MU-MIMO systems requires performance modeling and testing in real-world environments.

This thesis presents a novel scalable many-antenna base station architecture and control channel design, which are implemented and tested on three generations of large-scale custom built hardware platforms. We derive a theoretical model of many-antenna MU-MIMO system performance in real-world environments, accounting for hardware capabilities and channel estimation overhead. Leveraging these many-antenna MU-MIMO platforms, we conducted a comprehensive channel measurement campaign, spanning the UHF, 2.4 GHz, and 5 GHz bands, with varying degrees of mobility. Based on these measurements, we implement a mobility-aware MU-MIMO channel sounding system, which is able to virtually eliminate overhead from unnecessary channel sounding. Combined, these innovations enable many-antenna MU-MIMO systems to be efficiently implemented in real-world environments with mobility.