Fast Convex Method for Off-grid Millimeter-Wave/Sub-Terahertz Channel Estimation via Exploiting Joint Sparse Structure

In this paper, a fast optimization-based off-grid channel estimation method is proposed for millimeter wave (mmWave) or sub-terahertz(THz) cellular systems. Different from most existing works concerned with compressed sensing (CS) or off-grid methods, we proposed a novel first-order convex off-grid channel estimation approach to relieve the resolution loss from the angle quantization. Moreover, our proposed method is approximate inverse-free without costly matrix inversion. The numerical results demonstrate the effectiveness of the proposed FOG method, which achieves superior performance compared to CS-based approach and existing off-grid algorithm in terms of computational complexity and estimation accuracy.