Reciprocity Calibration of TDD MIMO Channel for Interference Alignment

Abstract

Interference alignment (IA) is a promising technique to boost the Multiple-input Multiple-output (MIMO) network throughput. Concurrent data transmission links between multiple pairs of transmitters (TX) and receivers (RX) can be established in the same spectrum via cooperation. To make the channel state information (CSI) available on the transmitter side, channel reciprocity in time division duplex (TDD) networks is assumed by a large number of distributed IA algorithms. The physical channel is reciprocal, but nevertheless the channel considered by IA is not, because it also involves the radio frequency (RF) front ends of the transceiver, which are different circuits for transmit and receive modes individually. The difference between the two circuits yields mismatch between the forward and backward channels. Given the mutual coupling effect of the antenna array, the channel's non-symmetry is further complicated. In this paper, we provide the modeling of the non-reciprocity for the TDD MIMO channels and propose a corresponding calibration method to minimize the mismatch. Experimental validation has been conducted with classic distributed IA algorithms. The results highlight the indispensability of the reciprocity calibration for approaching the network capacity gain promised by IA.