Power-Efficient Resource Allocation and Quantization for TDMA Using Adaptive Transmission and Limited-Rate Feedback

Abstract

Power-efficient scheduling and resource allocation are critical tasks for wireless sensor networks as well as commercial and tactical radios relying on IEEE access standards for power-limited communications. Tailored for such applications, this paper formulates and solves analytically novel convex optimization problems offering globally optimal user scheduling, as well as rate and power allocation for time-division multiple access (TDMA) in time-division-duplex or frequency-division-duplex operation. Through a limited-rate feedback link the access point provides quantized channel state information to the transmitters (Q-CSIT) based on which users adapt their modulation and code choices to the intended fading channel. When the quantizer needed to form the Q-CSIT is not prescribed, a joint allocation-quantization scheme is devised to minimize average transmit power subject to average rate and bit error rate constraints. The novel design couples adaptive transmission modes with quantization regions which are constructed to attain at least a local minimum of the average transmit power. Fairness in resource allocation is guaranteed by design. Transmit power and quantization region books are efficiently obtained offline while the online Q-CSIT based operation turns out to entail only a few feedback bits. Analysis and simulations include a perfect CSIT benchmark and reveal substantial power savings (as high as 15 dB) with low-overhead feedback.