Sparse Channel Estimation From Discrete-Time Fourier Transform Beam Measurements

In this paper, we study channel estimation at a uniform linear array (ULA) with N antennas, where the ULA is composed of xmlns:xlink="http://www.w3.org/1999/xlink">L paths different angles arrival (AoAs). It assumed that Discrete-Time Fourier Transform (DTFT) beams (also known as xmlns:xlink="http://www.w3.org/1999/xlink">analog DFT special cases) are applied to project incoming signal onto single (or multiple) RF chain(s), after which sampled and measured in baseband domain; underlying be constant during projections. This measurement procedure arises various communication systems, such receive beam sweeping phase 5G NR, DTFT used due their simple implementation shifts on analog antennas. A fundamental question about number measurements xmlns:xlink="http://www.w3.org/1999/xlink">K needed recover AoAs. Previous work problem showed (by applying compressed sensing theory) ≈ xmlns:xlink="http://www.w3.org/1999/xlink">LO (log( xmlns:xlink="http://www.w3.org/1999/xlink">N/L )) sufficient for recovering AoAs, grows . First, show necessary conditions recovery ≥ 2 Second, by using properties projections, able if then 3 xmlns:xlink="http://www.w3.org/1999/xlink">arbitrary suffice; hence, dependency completely removed. Furthermore, chosen equal period , enough, achieving sufficiency conditions. With these results, an AoA algorithm formulated has enormous complexity savings compared -dimensional search maximum likelihood (ML) estimation. Numerical simulations demonstrate algorithm’s improved performance over conventional algorithms beamspace ESPRIT sensing.