Fast Resource Management Algorithm for Passive Surveillance Systems

TL;DR

The paper tackles real-time resource management for passive surveillance systems with limited receivers observing multiple frequency bands. It proposes ResourceTune, a method that combines a left-right heuristic for generating candidate receiver configurations with linear programming to allocate configurations over time, thereby optimizing task observability. The approach yields a polynomial-time, real-time capable solution and demonstrates strong performance gains over greedy and GA-based baselines, including substantial improvements in most experimental scenarios. The work offers practical implications for PSS deployment and motivates future theoretical and priority-based extensions.

Abstract

Passive surveillance systems (PSS) detect and track objects that emit electromagnetic signals from hundreds of kilometers away. These systems have a limited number of receivers and can only observe a fraction of the frequencies of interest simultaneously. To improve its behavior, we propose the ResourceTune algorithm, which iteratively constructs optimized schedules to determine which frequencies each receiver should observe at a given time step. The algorithm's main component is the optimization of receiver configurations using a left-right heuristic combined with linear programming. Our approach is unique because, unlike others, we focus on optimizing available resources and observed frequencies, which was never done before. We experimentally compared the proposed algorithm with a greedy and the state-of-the-art method for construction of PSS schedules. In most of the considered scenarios, ResourceTune outperformed both algorithms, and in the most extreme case, its objective value was more than 2.7 times better than the values reached by other methods.

Figures (4)

• Figure 1: Illustration of the working principles of PSS. Image taken from era.
• Figure 2: The left part of the figure shows the frequency bands of the individual tasks. Their complete properties are described in Table \ref{['tab:tasks']}. The right side of the figure shows tuning plan $\mathit{T\!P}^1$.
• Figure 3: Configurations (blue rectangles) that are generated for parent $x$ (orange rectangle), $\omega_x = [500, 525]$, and selected shape $d = (70)$.
• Figure 4: Configurations (blue rectangles) that are generated for parent $x$ (orange rectangle), $\omega_x = [500, 525]$, and selected shape $d = (50, 100, 100)$.