A DBO-Based Improved 2D-MUSIC Algorithm for Localization Using OFDM

This paper considers a single input multiple output (SIMO) integrated sensing and communication (ISAC) system, where orthogonal frequency division multiplexing (OFDM) communication signals are multiplexed to detect target locations, including the range and angle parameters. The two-dimensional multiple signal classification (2D-MUSIC) algorithm is applied to process such signals with a format of multiple subcarriers related to large arrays. However, the standard 2D-MUSIC suffers from high computational complexity and its estimation accuracy is limited by the two-dimensional grid-based exhaustive search step size. To overcome these issues, we propose a two-step improved 2D-MUSIC (I2D-MUSIC) algorithm, which replaces the original grid-based exhaustive search and performs dung beetle optimization (DBO) algorithm to get a coarse estimation of the parameters. Subsequently, a stochastic gradient descent (SGD) based method is derived to obtain fine estimation of the range and angle parameters. Simulation results and hardware-based experiments demonstrate that the proposed algorithm significantly reduces computational complexity while maintaining comparable estimation accuracy to the standard 2D-MUSIC algorithm, and its accuracy is not constrained by the search step size.