POMDP-Driven Cognitive Massive MIMO Radar: Joint Target Detection-Tracking In Unknown Disturbances

TL;DR

This work tackles joint detection and tracking of a moving target in unknown disturbances using a POMDP framework for a Massive MIMO radar. It combines a disturbance-agnostic, robust Wald-type detector with an online planning method, POMCP, to maximize the detection probability $P_D$ while maintaining a fixed $P_{FA}$. The authors introduce a cognitive radar design that uses an unweighted particle filter and a generator-based POMCP simulator to handle non-Gaussian disturbances and to adapt waveform focus across multiple angle bins. Key findings include sustained high $P_D$ and accurate Cartesian state estimates in slow and fast target scenarios, outperforming SARSA-based benchmarks and traditional particle filtering, with a clear path to extensions to multi-target scenarios.

Abstract

The joint detection and tracking of a moving target embedded in an unknown disturbance represents a key feature that motivates the development of the cognitive radar paradigm. Building upon recent advancements in robust target detection with multiple-input multiple-output (MIMO) radars, this work explores the application of a Partially Observable Markov Decision Process (POMDP) framework to enhance the tracking and detection tasks in a statistically unknown environment. In the POMDP setup, the radar system is considered as an intelligent agent that continuously senses the surrounding environment, optimizing its actions to maximize the probability of detection $(P_D)$ and improve the target position and velocity estimation, all this while keeping a constant probability of false alarm $(P_{FA})$. The proposed approach employs an online algorithm that does not require any apriori knowledge of the noise statistics, and it relies on a much more general observation model than the traditional range-azimuth-elevation model employed by conventional tracking algorithms. Simulation results clearly show substantial performance improvement of the POMDP-based algorithm compared to the State-Action-Reward-State-Action (SARSA)-based one that has been recently investigated in the context of massive MIMO (MMIMO) radar systems.

Figures (11)

• Figure 1: POMCP Tree illustration with two actions and two observations.
• Figure 2: Study case 1: Potential trajectories with $\mathbf{s}_0 = (60\text{km}, 0.2\text{km/s}, -60\text{km}, 0.2\text{km/s})^T$ and noise $\sigma_s = 0.03$. The inner figure represents the average of SNR trajectories.
• Figure 3: Study case 1: RMSE between the estimated and true coordinates of the target for each algorithm.
• Figure 4: Study case 1: RMSE between the estimated and true velocities of the target for each algorithm.
• Figure 5: Study case 1: Probability of detecting a moving target for different approaches.