A Particle Swarm Optimization Approach to Surveillance Resource Management

Multi-function radar systems utilize electronically steered phased arrays and arbitrary waveform generation to perform tasks such as surveillance, tracking, and imaging on a single aperture. To take full advantage of these capabilities, new algorithms must be developed to efficiently distribute the finite time, energy, and processing budget of the radar between competing tasks. Much of the existing literature on resource management focuses on the tracking task, applying a fixed strategy for surveillance (e.g., a raster scan). This paper presents a novel representation of the surveillance environment that can be used to manage resources in phased array radar systems, using particle swarm optimization (PSO) to estimate the density of untracked targets in a search volume. The proposed algorithm, known as Surveillance PSO (SPSO), is computationally efficient for real-time operation, makes few assumptions about the environment or the underlying radar system, can easily be extended to simultaneous control of multiple beams, and provides an output representation that is a constant-length tensor that can be used directly as an input to a neural network. The efficacy of this approach is demonstrated by a series of simulated surveillance environments.