Modeling and Compensation of Transceiver Non-Reciprocity in TDD Multi-Antenna Base-Station

Abstract

Due to the increasing demands for higher system capacity, higher data rates and better quality of service in wireless networks, advanced techniques that improve wireless link reliability and spectral efficiency are introduced. This includes different multi-antenna technologies, in particular multi-user (MU) MIMO-OFDM. In MU MIMO-OFDM systems, base-station with multiple antennas communicates simultaneously with multiple users over a given time-frequency resource. In downlink transmission, base-station transmits multiple data streams through its antennas towards the user devices. In uplink transmission, the user equipment send in parallel multiple data streams towards the base-station. In general, channel non-reciprocity is a very important factor in cellular communications, in particular in precoded MU MIMO-OFDM systems adopting time division duplexing (TDD). Based on the channel reciprocity principle, the channel state information at base-station for the downlink transmission can be determined through estimating the uplink channels. In practice, however, there are always unavoidable frequency mismatch characteristics between transmitter and receiver. Frequency response mismatch can thus change the reciprocal nature of downlink and uplink channels. The impact of transceiver non-reciprocity at equipment on user side causes inter-stream interference which can be compensated using detection processing. The impact of transceiver non-reciprocity at base-station causes inter-user interference and degrades the system performance of MU MIMO-OFDM systems. To ensure the system reliability and high performance in case of transceiver non-reciprocity, some non-reciprocity estimation and compensation methods are required. The previous work has proposed the estimation-compensation framework that gives a flexible solution to restore the channel reciprocity. But there is a need to validate the findings and performance of the proposed estimation-compensation framework. The modeling of transceiver frequency response mismatch characteristics using actual measurement data has been carried out in this thesis research work. The actual measurement data comprises of one base-station with two antennas and two user equipment devices with single antenna. The estimated uplink and downlink channels from measurement data are used to compute the non-reciprocity matrix at base-station and at the equipment on user side after mathematical calculations. The normalized parameters for transceiver non-reciprocity matrices are extracted subcarrier-wise. The frequency-domain normalized non-reciprocity parameters are modeled as a FIR filter in the time-domain and the most energy concentrates then on few time-domain taps. The extracted parameters are mildly frequency-selective. The impact of extracted transceiver non-reciprocity is then analyzed by implementing a simulator of TDD precoded MU MIMO-OFDM system. In general, the frequency-selectivity implies that the reciprocity estimation and compensation is needed subcarrier-wise. The pilot-based estimation of non-reciprocity parameters at base-station is carried out in order to enhance the system performance. To estimate channel non-reciprocity parameters, a link between base-station and one of user equipment devices is assumed. The right choice of selecting the user is also important for noise reduction in estimation. For estimation, the DL transmission channel is modeled as a Rayleigh fading multipath channel with a given 7-tap channel power delay profile. The downlink data including sparsely located pilots at selected subcarriers is transmitted to the user through downlink channel without precoding. The downlink channel is then estimated at the user equipment side. This provides estimates only at the pilot subcarriers. Therefore, linear interpolation is used to obtain channel response estimates at the actual data subcarriers. The uplink pilot data is transmitted to base-station from user equipment through uplink channel. The uplink channel is obtained by estimated downlink channel in case of non-reciprocity parameters. Then, estimate of non-reciprocity at base-station is computed by using inverse processing and an interpolator. The estimated parameters are used as a compensator filter in order to compensate the channel non-reciprocity in the system. The simulated results show that the performance deviates from the ideal linear precoded MU MIMO-OFDM system because of non-reciprocity in case of both error control coded and uncoded channels. The compensated results in terms of coded and uncoded channel schemes have been evaluated which are closer to ideal linear precoded MU-MIMO OFDM system. These results show that the impact of non-reciprocity on system performance is less severe when a coded channel is deployed as compared to uncoded channel. The modeling of transceiver frequency response mismatch characteristics using actual measurement data proves that the proposed non-reciprocity model in the previous research work is close to reality.