Estimation of BS Transceiver Non-Reciprocity in Multi-User Massive MIMO Systems

This paper proposes efficient algorithms for estimating the base-station transceiver non-reciprocity characteristics in precoded TDD multi-user MIMO-OFDM networks, such that the estimation processing is carried out on the user equipment (UE) side. First, detailed signal models are derived to characterize the effects of transceiver non-reciprocity on both base-station and UE sides, in case of zero-forcing precoded multiuser MIMO-OFDM downlink transmission. The developed signal models show that transceiver non-reciprocity on the base-station side is the main cause of downlink performance degradation, resulting in general to inter-user interference. Then, based on closed-form analysis of the overall interference or distortion level on the receiver side, two efficient UE-based algorithms are proposed, depending on the availability of downlink pilot symbols, to estimate the level of base-station transceiver nonreciprocity without any specific calibration procedure. The performance of the proposed estimation methods are evaluated using extensive computer simulations. The obtained results show that even without any downlink pilot symbols, the estimation of the base-station non-reciprocity characteristics is still feasible, while further improved estimation accuracy can be obtained if downlink pilots are available in the system. In general, by being able to estimate the level of base-station transceiver nonreciprocity on the UE side, dedicated calibration procedures and the associated system overhead can be avoided, thus enabling the base-stations to carry out calibration only when truly needed.