Robust Sensor Geometry Design via Sensitivity-Based Information Analysis

This letter develops an information-geometric framework for robust sensor geometry design in multi-sensor localization with heterogeneous measurements. Bearing, range, and range-difference measurements are represented in a unified Fisher information matrix (FIM) form as tangential, radial, and differential-radial information components, respectively. Based on this directional decomposition, the sensing geometry is mapped to a D-optimality surface, and an error-weighted sensitivity measure is introduced to quantify the local variation of information under deployment perturbations. A hierarchical selection criterion is then proposed to choose a low-sensitivity geometry within a prescribed information-retention region, instead of directly pursuing the classical D-optimal peak. Numerical results in a three-sensor scenario show that the selected geometry preserves about$90\%$of the maximum information while reducing sensitivity and information variance under perturbations. These results demonstrate that sensitivity-aware geometry selection can improve deployment robustness with only a small loss of nominal localization performance.