Multi-Sensor Adaptive Birth for Labeled RFS Filters using Bistatic Range-Only Measurements

Recently, a Monte Carlo importance sampling-based approach has been established to achieve scalable, multi-sensor, measurement adaptive track initialization for labeled random finite set filters. However, previously suggested proposal distributions require every sensor's measurement function to have a differentiable inverse in the observable dimensions of the target's state space. This assumption is valid for measurement modalities such as position or angle-range sensors, but not for many common non-invertible measurement modalities such as bistatic range-only, angle-only, and range-only sensors. This paper provides an alternative proposal distribution for Monte Carlo importance sampling-based, multi-sensor, measurement adaptive track initiation that is not restricted to invertible measurement functions. The solution for a bistatic range-only measurement function is provided, and simulation results are shown to verify the efficacy of the solution.