Energy-efficient resource allocation of wireless systems with statistical QoS requirement

This paper strives to improve the energy efficiency (EE) of wireless systems by resource allocation for delay sensitive traffic with the statistical quality of service (QoS) requirement, where a delay bound and a small violation probability is ensured. To avoid wasting energy when serving random traffic, we turn the QoS exponent, a key parameter to characterize statistical QoS guarantee under the framework of effective bandwidth and effective capacity, into a QoS function dependent on the queue length. A method to design queue length based resource allocation strategy is proposed, which maximizes the EE under the constraint on the QoS function. By taking an OFDM system serving Poisson source as an example, a closed form subcarrier allocation strategy with constant transmit power on each subcarrier is derived. Simulation and numerical results show that the EE achieved by the proposed strategy is much higher than the relevant strategies in the literature considering statistical QoS provision, and is very close to the EE upper bound achieved when the delay bound is infinite.