Low complexity bit and power allocation for MIMO-OFDM systems using space-frequency beamforming

This paper presents a new bit and power allocation algorithm for multiple-input multiple-output orthogonal frequency division multiplexing (MIMO-OFDM) systems. The optimal power allocation algorithm in terms of maximum capacity in MIMO-OFDM systems is given by the joint space-frequency water-filling (JSF-WF)¹ algorithm. This algorithm involves multiple singular value decompositions and an iterative water-filling calculation over the subchannels, and therefore, has a high computational complexity. We present an algorithm with reduced bit and power allocation complexity. The proposed algorithm is based on a geometric channel model consisting of several clusters of subpaths, characterized by a set of physical parameters. It performs beamforming with null-steering towards the clusters' directions-of-departure, such that the frequency-selective MIMO channel is transformed into a flat fading MIMO channel. Thus, constant bit and power allocation over the frequency domain can be performed, which is simple to compute and implement. The performance of the proposed algorithm is evaluated and compared to the JSF-WF algorithm in terms of bit-error-rate, for known and misspecified channel model parameters. It is demonstrated that the proposed algorithm performance is slightly lower than the JSF-WF performance, while it significantly reduces the complexity of the allocation algorithm.