Exploiting relative transfer function for source localization in underwater acoustics

Passive source localization techniques employing data-derived replicas have recently gained attention as an effective means of mitigating well-known mismatch issues associated with matched field processing methods. Conventional approaches typically rely on cross-correlation-based replicas. To enhance replica uniqueness and improve robustness to differences in the spectral content between library and target sources, this Letter investigates the use of the relative transfer function as the localization feature within a data-derived replica-based framework. The method's performance is evaluated through normal mode simulations and compared with that of the conventional approach, and its feasibility is assessed using controlled in situ data.