Adaptive space-time compression for efficient massive MIMO fronthauling

One of the key features of the forthcoming fifth-generation (5G) communications is the deployment of massive multiple-input-multiple-output (MIMO) antennas to support ultra-high mobile traffic density. This scenario will pose a serious challenge on the capacity of mobile fronthaul in the centralized/cloud radio access network (C-RAN) since the required fronthaul bandwidth would linearly increase with number of antennas if conventional fronthaul interfaces (e.g., CPRI) are used. In this paper, we propose an adaptive space-time compression technique to significantly improve bandwidth efficiency of fronthaul. The technique incorporates an adaptive spatial filter to track the signal subspace and reduce the number of spatial channels, followed by adaptive quantizers to compress bandwidth of each channel in time domain. Enabled by the technique, the required fronthaul bandwidth becomes only dependent on the number of users, which is no longer proportional to the number of antennas. Moreover, compared with traditional fronthaul compression schemes in only the time domain, the flexibility of the compressor increases, and joint space-time optimization becomes feasible. On the other hand, optical fronthaul bandwidth is usually limited by cost-effective optical and electronic components. Moreover, increased reach would limit the bandwidth of IM-DD-based fronthaul (due to chromatic dispersion) as well as the received optical power. We experimentally investigate the combined optimization of a proposed space-time compressor with an optical fronthaul link. Experimental results of uplink 256-antenna fronthaul (259.5-Gb/s CPRI-equivalent rate) show that 32 users with 20MHz (30.72MSa/s) OFDM signal with lower-than-1% EVM are supported by 10GBd PAM4 optical interface.