Submitted to IEEE Transactions on Wireless Communication, December 8, 2005.
This paper proposes a novel multi-dimensional power control technique for the uplink of cellular MIMO systems. Similar to prior power control work for single-antenna systems, we employ a fixed SINR target that allows guaranteed quality-of-service for delay-sensitive data applications. We propose two approaches, the first where each user's power is equally allocated to its antenna array, in the second the allocation is adaptive. The optimal solution with full channel knowledge, and a practical near-optimal solution requiring only partial channel knowledge, are both derived. Numerical results show that power control, compared to supposedly optimal water-filling strategies, actually achieves substantially higher throughput at the low SINRs typical in cellular systems, with significantly lower overhead and complexity. Specifically, our results show substantially better throughput improvement at low-to-moderate SINRs compared to the well-known iterative water-filling approach. In the higher SINR ranges, iterative water-filling indeed slightly outperforms adaptive power allocation. Due to its better exploitation of spatial diversity and reduced transmit power (and hence reduced other-cell interference), adaptive power allocation increases the achievable SINR by an order of magnitude over equal power allocation, resulting in far better coverage.
This preprint is available via email.