Gridless Channel Estimation for Millimeter Wave MIMO-OFDM Systems with Beam Squint

Obtaining accurate channel state information is crucial to fully harness the considerable bandwidth advantages of millimeter wave (mmWave) communications. However, as the bandwidth and the size of antenna array increase, the spatial-wideband effect causes the array response vector to be frequency-dependent, also known as the beam squint effect, making channel estimation more challenging. To address this issue, in this paper, we introduce the Jacobi-Anger expansion to extract common angular information from the frequency-dependent array response vectors and formulate the channel estimation as an atomic norm minimization problem. To circumvent the prohibitively high computational complexity, we propose a decoupled-atomic norm minimization (D-ANM) algorithm by decoupling the original problem into three sub-problems for sequentially estimating the angles of arrival, angles of departure, and path delays. Simulation results show that our proposed D-ANM algorithm achieves significant performance improvement in the channel estimation accuracy over the benchmark algorithms for the mmWave systems with the beam squint effect.