Abstract
This paper investigates power allocation for a downlink non-orthogonal multiple access (NOMA) system with a base station and two users under imperfect channel estimation, where the variance of the channel estimation error for each transmission link is assumed to be known. We first present approximated user capacity expressions according to the error variance information, and then formulate a power allocation optimization problem between the two users for maximizing the minimum approximated user capacity under a total power constraint. Since such a max-min problem is equivalent to maximizing the minimum approximated received signal-to-interference-plusnoise ratio (SINR), a closed-form power allocation solution can be derived based on SINR balancing between the two users. The proposed method involves solving a quadratic equation, and only low implementation complexity is required for practical applications. Computer simulation results show that the proposed max-min power allocation scheme is more robust against channel estimation errors and achieves better worst bit-error-rate performance for the two users, as compared with the conventional fixed power allocation approach and the related max-min method derived according to perfect channel state information.