Low feedback downlink SDMA using single antenna MIMO-like detection

Space division multiple access (SDMA) is considered one of the key techniques enabling 4G spectral efficiency and capacity, and has been adopted by various advanced communications standards such as LTE and IEEE802.11ac. One of the challenges in downlink (DL) SDMA is obtaining adequate channel state information at the base-station (BS) transmitter to minimize multi user interference (MUI). MUI minimization requires the BS to acquire accurate channel state information respective to the mobiles, achieved in the form of feedback. Digital feedback is usually made of a quantized version of either the channel matrix or the singular vectors. Naturally, finer quantization leads to more accurate channel state information and less MUI, but also to higher feedback capacity. Minimizing MUI is especially important when the number of transmitted SDMA streams is larger than the number of receive antennas at the mobiles, for example when a BS transmits M SDMA streams to M mobiles, each endowed with a single receive antenna. Here the mobiles do not have enough degrees of freedom to suppress the MUI, for example through an MMSE receiver (which would be possible if each mobile had at least M receive antennas). In this paper we propose MIMO-like ML detection of all M SDMA information streams at each of the mobiles (each is endowed with a single antenna). Here each mobile discards all the irrelevant streams, targeting other mobiles, post ML detection. Although counter-intuitive, when the optimal nonlinear ML detector is employed, the standard requirement of N ≥ M receive antennas for spatial multiplexing detection is essentially eliminated. This approach is immediately extended to the case of N <; M receive antennas at the mobile, where N is not necessarily 1. The proposed method significantly outperforms standard detectors in the regime of high MUI due to insufficient feedback quantization. Thus, it paves the way to a significant decrease in feedback capacity and relaxed SD- A group selection.