Given the critical dependence of broadcast channels by the accuracy of channel state information at the transmitter (CSIT), we develop a general downlink model with zero-forcing precoding, applied in realistic heterogeneous cellular systems with multiple-Antenna base stations (BSS). Specifically, we take into consideration imperfect CSIT due to pilot contamination, channel aging due to users relative movement, and unavoidable residual additive transceiver hardware impairments (RATHIs). Assuming that the BSS are Poisson distributed, the main contributions focus on the derivations of the upper bound of the coverage probability and the achievable user rate for this general model. We show that both the coverage probability and the user rate are dependent on the imperfect CSIT and RATHIs. More concretely, we quantify the resultant performance loss of the network due to these effects. We depict that the uplink RATHIs have equal impact, but the downlink transmit BS distortion has a greater impact than the receive hardware impairment of the user. Thus, the transmit BS hardware should be of better quality than user's receive hardware. Furthermore, we characterise both the coverage probability and user rate in terms of the time variation of the channel. It is shown that both of them decrease with increasing user mobility, but after a specific value of the normalized Doppler shift, they increase again. Actually, the time variation, following the Jakes autocorrelation function, mirrors this effect on coverage probability and user rate. Finally, we consider space-division multiple access (SDMA), single-user beamforming (SU-BF), and baseline single-input single-output transmission. A comparison among these schemes reveals that the coverage by means of SU-BF outperforms SDMA in terms of coverage.
- Additive hardware impairments
- channel aging
- channel estimation
- coverage probability
- multiple antenna heterogeneous cellular networks