Unintended yaw introduces structured phase errors that blur coherent gain and elevate sidelobes in synthetic aperture sonar. This work presents a compact compensation scheme that couples a non-orientable Mobiusstrip hydrophone array with twenty elements and a single 180-degree twist to a maximum-likelihood receiver-selection rule. At each ping the method selects the best receiver, interpreted as the element whose phase history most closely matches the pre-motion reference, and then re-anchors phase through a closed-form correction before coherent image formation. The geometry’s controlled vertical excursion and near-uniform azimuthal coverage reduce both the number and the strength of yaw-induced phase modes, enabling the selector to operate in a low-rank feature space derived from phase differences and short-term coherence. Evaluation uses a three-by-three point-target grid T1–T9 under a yaw profile of plus-minus 0.8 deg with a 0.5 hertz component and mild colored noise. Metric windows, side-lobe masks, color scales, and dynamic ranges are held fixed across conditions to ensure fair comparison. Image quality is reported with PSLR, ISLR, PSNR, and RMSE averaged over the nine targets. Along the yaw axis the compensated imagery consistently deepens the sidelobe floor and raises fidelity without broadening the main lobe. Averaged over nine targets, PSLR improves by 19.50 decibels, ISLR improves by 18.85 decibels, PSNR increases by 32.76 decibels, and RMSE falls from 0.19588 to 0.00451, demonstrating that maximum-likelihood receiver selection on a Mobius layout is a practical route to yaw-robust synthetic aperture sonar.

Keywords