Efficient Parameter Determination for BPDN in Underwater Acoustic MIMO-OSDM Communications Using Stein's Unbiased Risk Estimate

Underwater acoustic (UWA) communication schemes including multiple-input-multiple-output orthogonal signal division multiplexing (MIMO-OSDM) are crucial for the wireless operation of autonomous unmanned vehicles and remotely operated vehicles. This paper presents an efficient regularization parameter determination for MIMO-OSDM to design a reliable communication system. Recent studies have revealed that utilizing sparsity, a physical nature of the UWA channel can improve communication quality. While the combination of MIMO-OSDM and basis pursuit denoising (BPDN) enables the UWA communication that accounts for channel sparsity, this approach requires the prior simulation to obtain a database to fine-tune linear coefficients which decide the regularization parameter of BPDN. To reduce computational cost without sacrificing communication quality, we propose the statistical solution: an application of Stein's unbiased risk estimate (SURE) for MIMO-OSDM using BPDN aiming for parameter determination through a minimization of the channel estimation error. Our simulation results demonstrated that the proposed parameter determinator serves as an estimator of channel estimation error. Our experimental results demonstrated that the proposed channel estimator using SURE achieved almost the same communication quality and demodulation performance as the existing method without prior database creation.