Sum-rate optimized power allocation for the OFDM multiple access relay channel

We¹ consider transmission over a set of parallel MARCs (multiple-access relay channel) based on the OFDM (Orthogonal Frequency Division Multiplexing) modulation. Two users communicate to a common destination (D) with the help of a relay (R) over a set of parallel channels. The relay operates in decode and forward (repetition coding type) and half duplex modes. The problem considered is the maximization of the total sum rate at the destination by optimal allocation of the power to the different carriers for the users and the relay. The problem is investigated for a sum power constraint. We show that for each carrier only one user should be relayed at most, and the other user should communicate over the direct link. Unlike the classical MAC, we show that there exist cases where both users should share the same carrier to transmit symbols and this increases the rate. For these carriers, there exist, considering both the relay and the destination, four possible decoding orders. As the problem is non-convex and non-linear, it is not easy to a-priori select the best decoding order (at the relay and the destination), and which user should be relayed. An algorithm is proposed to solve the global optimization problem. The theoretical results are illustrated by computational results.