Towards 5G High Mobility: A Fairness-Adjustable Time-Domain Power Allocation Approach
Ke Xiong,Pingyi Fan,Yu Zhang,Khaled Ben Letaief
DOI: https://doi.org/10.1109/access.2017.2712710
IF: 3.9
2017-01-01
IEEE Access
Abstract:How to provide efficient information service for high-mobility scenarios, including high-speed railway communications (HSRCs), is one of the most important requirements in future 5G communication networks. In HSRCs, due to pass-loss effect, the information rate between roadside base station (BS) and the moving train closely depends on the distance between the BS and the train, and the high-mobility speed of trains makes the path loss vary fast with time. It is essential to implement time-domain power allocation to mitigate the near-far effect. To explore the information transmission capacity of HSRCs, some power allocation schemes were developed. With the water-filling method, the maximum amount of information (mobile service amount) can be delivered, where, however, most power is allocated to the time interval when the train is nearest to the BS, which causes great unfairness with respect to time. Although the proportional power allocation can achieve much better fairness along the time, it causes a relatively big loss in mobile service amount, resulting in low utilization efficiency of HSRC channels. Enlightened by the definition of Rényi entropy, this paper proposes an novel power allocation scheme called β-fairness power allocation, which is able to achieve relatively high mobile service amount with fairness between the water-filling and proportional power allocation. It is a generalized fairness power allocation, by which the tradeoff between the mobile service amount and fairness can be easily controlled by adjusting the value of β. Particularly, as β = 0, it becomes a traditional water-filling method, while β = 1, it becomes the existing proportional power allocation. To achieve a more general power adjustment with user QoS requirement, the rate-constrained β-fairness power allocation is also investigated, where a closed form of the rough optimal result is first derived and then the precise power allocation is obtained by an efficient algorithm. Besides, we also discuss the impact of different parameters on system performance, which may provide some useful insight for an HSRC system design.