Abstract : In this paper, we investigate the power allocation problem in downlink orthogonal frequency-division multiple access (OFDMA) networks. Different from previous researches on power allocation, we take into account various practical factors, such as the stochastic traffic arrival, the time-varying channel, the queue stability requirements of all users, the channel estimation cost and the corresponding effect of imperfect channel state information (CSI) on data transmission rate. The power allocation problem is formulated as maximizing the time-averaged data transmission rate by optimizing pilot and data power allocation subject to the queue stability and the maximum transmit power constraints. The data transmission rate is defined in terms of the pilot transmit power, the data transmit power and the channel estimation error, which is non-concave. To solve the non-concave and stochastic optimization problem, a dynamic pilot and data power allocation (DPDPA) algorithm is proposed with the aids of approximate transformation, Lyapunov optimization and Lagrange dual formulation. Moreover, we derive the bounds of performances, in terms of the time-averaged data transmission rate and queue length.
Index terms : Channel estimation, dynamic power control, imperfect CSI, queue stability