Skip to Main Content
In a time-division duplexing (TDD)/code-division multiple-access (CDMA) cellular network with asymmetric data traffic, dynamic channel allocation (DCA) enhances resource utilization compared with fixed channel allocation (FCA). However, it also induces base-to-base and mobile-to-mobile crossed-slot intercell interference, which can severely degrade system performance. To deal with this problem, a decentralized scheme is proposed, which combines an interference-aware DCA algorithm with space-time linear minimum-mean-square-error (LMMSE) joint detection at the base and mobile stations. The former assigns active links to timeslots in a way that crossed-slot interference is mitigated, while the latter suppresses the remaining intercell interference (along with intersymbol and intracell interference), exploiting its spatio-temporal autocorrelation statistics. The performance of this scheme is evaluated in terms of downlink and uplink signal-to-interference-plus-noise ratio (SINR) outage and average throughput via analytical approximations and Monte Carlo simulations, and it is compared with that of benchmark random DCA (RDCA) and FCA schemes. The cases of single- and dual-antenna reception with perfect and imperfect channel state information are examined. It is shown that the proposed scheme achieves higher average throughput than FCA (particularly for dual-antenna reception) as well as RDCA (for heavy traffic loads). These throughput gains are more significant in uplink than in downlink.