Optimization of link adaptation with a practical power consumption model

Link adaptation (LA) is an effective technique to significantly boost the spectral efficiency of current wireless systems, so as to meet the increasing demand for high-speed data service at low cost. LA achieves high spectral efficiency by minimizing power of maximizing throughput under a constant power budget. In order to make its application meaningful in practical systems, we have to take into account the actual battery power consumption instead of just the transmitted power. First, the power amplifier is not 100% efficient and the relationship between transmitted and consumed power is nonlinear. Second, there is the processing power in addition to the transmitted power. This is overlooked in previous work in LA. Thus, we investigate the optimal LA scheme within the real battery power consumption context and present some preliminary results. This paper first identifies the problem with the "conventional" power concept, and gives an abstract nonlinear model for the battery power consumption, then it proposes an optimization method for LA. Finally, it demonstrates this method, using adaptive modulation as an example. The resulting adaptation scheme is quite different from the previous optimal one (both in the modes used and switching thresholds). The numerical results show at least 1.2 dB gain over the previous optimal scheme derived with the "conventional" power concept in the Rayleigh channel. Therefore we have to use this alternative scheme to fully utilize the benefit of LA in real systems. The proposed power model and the analysis approach also are generic and widely applicable.