Optimal Delay-Power Tradeoff in Wireless Transmission with Fixed Modulation

Cross-layer scheduling is an emerging and powerful solution that can significantly improve the quality of service (QoS) and power efficiency in wireless networks by jointly using channel and buffer states to adjust transmission power and rate. In conventional scheduling methods, adaptive modulation and coding (AMQ) is adopted to realize rate adaptation. However, this may greatly increase the transceiver and protocol complexity. To overcome this, scheduling methods based on fixed modulation are studied in this paper. In this case, the scheduler only needs to determine whether the transmitter is active at given channel and buffer state. When the transmitter is active, it will transmit in a fixed data rate, by using power adaptation to achieve a required SNR at the receiver. Our aim is to minimize the average packet delay given constraint on average power. To do that, we shall derive the analytical results of delay, packet-loss rate, and power consumption of cross-layer scheduling. It will then be shown that there exists a fundamental tradeoff between delay and power, which characterize the performance limit of cross-layer scheduling. In addition, a delay-optimal scheduling method will also be presented.

I. Introduction

Packet-switched infrastructure is expected to provide ubiquitous multimedia services for wireless terminals because it holds the promise of multiplexing various traffic flows in an efficient and flexible way. In future wireless systems, it is a challenging work to guarantee QoS parameters such as delay and packet-loss rate, not only because the energy of mobile terminalsis highly limited, but also because the power efficiency may degrade greatly due to the fading and jamming nature of wireless communications. To achieve better QoS and higher resource efficiency, scheduling is extensively studied over the last decade. Most recently, it was found that various scheduling algorithms that utilize the interconnections among different layers have emerged to significantly improve of overall system performance [1].