Multi-Cell Massive MIMO Uplink With Underlay Spectrum Sharing

• Published in: IEEE transactions on cognitive communications and networking 2019-03, Vol.5 (1), p.119-137
• Main Authors: Al-Hraishawi, Hayder, Aruma Baduge, Gayan Amarasuriya, Ngo, Hien Quoc, Larsson, Erik G.
• Format: Article
• Language: English
• Subjects: Antennas; Asymptotic properties; Cochannel interference; Cognitive radio; Contamination; Interference; Massive MIMO; MIMO (control systems); MIMO communication; Performance measurement; Pilots; Power control; underlay spectrum-sharing; Uplink
• ISSN: 2332-7731; 2332-7731
• DOI: 10.1109/TCCN.2018.2877998

The achievable rates are investigated for multi-cell multi-user massive multiple-input multiple-output (MIMO) systems with underlay spectrum sharing. A general pilot sharing scheme and two pilot sequence designs (PSDs) are investigated via fully shared (PSD-1) and partially shared (PSD-2) uplink pilots. The number of simultaneously served primary users and secondary users (SUs) in the same time-frequency resource block by the PSD-1 is higher than that of PSD-2. The transmit power constraints for the SUs are derived to mitigate the secondary co-channel interference (CCI) inflicted at the primary base-station (PBS) subject to a predefined primary interference temperature (PIT). The optimal transmit power control coefficients for the SUs with max-min fairness and the common achievable rates are derived. The cumulative detrimental effects of channel estimation errors, CCI and intra-cell/inter-cell pilot contamination are investigated. The secondary transmit power constraint and the achievable rates for the perfect channel state information (CSI) case become independent of the PIT when the number of PBS antennas grows unbounded. Therefore, the primary and secondary systems can be operated independent of each other as both intra-cell and inter-cell interference can be asymptotically mitigated at the massive MIMO PBS and secondary base-station. Nevertheless, the achievable rates and secondary power constraints for the imperfect CSI case with PSD-1 are severely degraded due to the presence of intra-cell and inter-cell pilot contamination. These performance metrics depend on the PIT even in the asymptotic PBS antenna regime. Hence, the primary and secondary systems can no longer be operated independently for imperfect CSI with PSD-1. However, PSD-2 provides an achievable rate gain over PSD-1 despite the requirement of lengthier pilot sequences of the former than that of the latter.