2016/08/08 by Saeid Haghighatshoar, Giuseppe Caire, Haghighatshoar, Saeid +1 · 1 citation
Computer Science · Engineering · #Direction-of-Arrival Estimation Techniques #Advanced MIMO Systems Optimization #Advanced Adaptive Filtering Techniques
paper · pdf · doi:10.48550/arxiv.1608.02477
Massive MIMO is a variant of multiuser MIMO, where the number of antennas M\nat the base-station is large, and generally much larger than the number of\nspatially multiplexed data streams to/from the users. It has been observed that\nin many realistic propagation scenarios as well as in spatially correlated\nchannel models used in standardizations, although the user channel vectors have\na very high-dim M, they lie on low-dim subspaces due to their limited angular\nspread. This low-dim subspace structure remains stable across many coherence\nblocks and can be exploited in several ways to improve the system performance.\nA main challenge, however, is to estimate this signal subspace from samples of\nusers' channel vectors as fast and efficiently as possible. In a recent work,\nwe addressed this problem and proposed a very effective novel algorithm\nreferred to as Approximate Maximum-Likelihood (AML), which was formulated as a\nsemi-definite program (SDP). In this paper, we address two problems left open\nin our previous work: computational complexity and tracking. The algorithm\nproposed in this paper is reminiscent of Multiple Measurement Vectors (MMV)\nproblem in Compressed Sensing and is proved to be equivalent to the AML\nAlgorithm for sufficiently dense angular grids. It has also a very low\ncomputational complexity and is able to track sharp transitions in the channel\nstatistics very quickly. Although mainly motivated by massive MIMO\napplications, our proposed algorithm is of independent interest in other\nrelated subspace estimation applications. We assess the estimation/tracking\nperformance of our proposed algorithm empirically via numerical simulations,\nespecially in practically relevant situations where a direct implementation of\nthe SDP would be infeasible in real-time. We also compare the performance of\nour algorithm with other related subspace estimation algorithms in the\nliterature.\n