Joint Detection and DOA Tracking with a Bernoulli Filter Based on Information Theoretic Criteria

Area of Science:

• Signal Processing
• Estimation Theory
• Target Tracking

Background:

• Accurate tracking of moving targets with random appearances/disappearances is challenging.
• Low signal-to-noise ratio (SNR) conditions degrade detection and direction-of-arrival (DOA) performance.

Purpose of the Study:

• To develop a robust method for joint detection and DOA tracking of a single moving source with random state changes.
• To improve tracking accuracy in low SNR environments using multi-snapshot sensor array data.

Main Methods:

• Utilized Bernoulli random finite set (RFS) to model source dynamics and appearance/disappearance.
• Proposed a track-before-detect (TBD) Bernoulli filter for randomly switching systems.
• Marginalized unknown noise variances using uninformative priors and information theoretic criteria for likelihood compensation.

Main Results:

• The proposed TBD Bernoulli filter demonstrates effective joint detection and DOA tracking.
• The filter shows improved performance in low SNR scenarios.
• The method successfully handles randomly appearing and disappearing targets.

Conclusions:

• The developed Bernoulli filter provides a robust solution for tracking dynamic targets with uncertain states.
• The approach is effective for improving detection and DOA tracking in challenging low SNR environments.
• The filter's ability to handle random target on/off switching is validated by simulations.