Fog massive MIMO with on-the-fly pilot contamination control
Giuseppe Caire (TU Berlin)
Massive MIMO presents several attractive features for very low latency, high reliability, random access communications. In particular, due to the large number of antennas, the wireless fading channel behaves almost deterministically, such that complicated adaptive rate schemes are not needed. Nevertheless, in a multi-cell dense deployment, frequent handovers, with per-cell pilot re-assignment, may still incur significant protocol overhead and latency. In this talk, we present a novel “Fog” massive MIMO architecture where users seamlessly and implicitly associate to the most convenient multiantenna Remote Radio Head (RRH) in a completely autonomous manner. Each user is associated with a unique uplink pilot sequence, and pilot contamination is mitigated by a novel “on-the-fly” pilot contamination control mechanism. Our scheme preserves the advantages of Cloud-RAN processing (in particular, the notion of “cell” is blurred and no association between users and RRHs needs to be explicitly negotiated), without incurring in the latency of fully joint processing of the RRH signals at a common cloud center. Furthermore, we can analyze the spectral efficiency of the resulting scheme via stochastic geometry, using some recent results on “unique coverage in Boolean models”, which were developed specifically to analyze our proposed system. We compare our Fog massive MIMO system with a baseline massive MIMO cellular system and with the recently proposed “cell-free” architecture, and show the superiority of the proposed scheme through analysis and simulation.