Adaptive power control for data traffic in a cellular DS-CDMA system.

摘要

Among various multiple-access techniques, DS-CDMA has shown great promise for the third generation wireless cellular systems. In this work, we consider adaptive power control schemes of data mobiles to maximize the system capacity while maintaining the quality of service in a cellular DS-CDMA uplink system. By exploiting the characteristics of data traffic, power control can be done adaptively to achieve better performance. This issue is particularly important because the radio spectrum is a scarce resource.; We first study an adaptive uplink power control scheme based on truncated channel inversion. The basic idea is to suspend mobile transmission when the wireless channel is in a bad state to reduce interference to other cells. We develop a queueing model for performance evaluation. We compare the truncated and the traditional scheme, and observe significant improvement in both the throughput and the power consumption at the cost of extra queueing delay. We then generalize the idea to two-level channel inversion, to alleviate the problem of large queueing delay and to improve the overall performance. We find that the two-level scheme significantly improves the delay performance while achieving substantial improvement on the system capacity. We further develop a numerical algorithm to optimize the overall performance. Finally, we investigate a mixed voice and data system, where the reservation based admission control scheme is used to prioritize voice mobiles and the truncated power control scheme is used by data mobiles to improve the system capacity. The performance of the system is measured by the Erlang capacity. Two methods of determining the Erlang capacity are presented. The first method decouples the analysis of the blocking and outage performance, thus simplifying numerical search. The second method takes into account the impact of mobile traffic fluctuation on interference statistics, and provides more accurate results. Our numerical results show that the difference between the two methods is quite significant.