QoS-Driven Power-Allocation Game over Fading Multiple-Access Channels

We integrate the effective capacity theory and game-theoretic approach to develop quality-of-service (QoS) driven power-allocation schemes in fading multiple-access channels (MAC). The effective capacity characterizes the capability of the wireless channels to support data transmission subject to the statistical delay QoS constraints, and the game-theoretic approach can efficiently handle the selfish behaviors of game users. In particular, each user allocates power in a selfish and noncooperative yet rational way, and its transmit power will be treated as the background noise to the other peer users. Under the above setup, we formulate the noncooperative power-allocation game, where each game user attempts to maximize its own effective capacity under the average power constraint. We focus on the two-user case and derive the Nash equilibrium and the corresponding power-allocation policy for this game. Furthermore, we show that as the delay QoS constraint becomes extremely loose, our power-allocation game reduces to the existing water-filling game. Also conducted are numerical and simulation analyses which compare the performance between our QoS-driven game based scheme and the water-filling game based scheme.