Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Fast Anti-Jamming Cognitive Radar Deployment Algorithm Based on Reinforcement Learning

Wencheng Cai, Xuchao Gao, Congying Han, Mingqiang Li, Tiande Guo

TL;DR

The paper tackles fast, anti-jamming deployment of cognitive radars by formulating the problem as a combinatorial optimization and solving it with a deep reinforcement learning framework (FARDA). FARDA introduces problem simplifications, an encoder-based policy network, and a novel CVDP-EXPR reward to enable efficient end-to-end learning, achieving deployment speeds about 7,000× faster than evolutionary baselines while maintaining comparable coverage. Ablation studies verify the necessity of the encoder and the reward design. The work demonstrates the potential for real-time radar deployment in contested electromagnetic environments and lays groundwork for extending to more complex static and dynamic scenarios.

Abstract

The fast deployment of cognitive radar to counter jamming remains a critical challenge in modern warfare, where more efficient deployment leads to quicker detection of targets. Existing methods are primarily based on evolutionary algorithms, which are time-consuming and prone to falling into local optima. We tackle these drawbacks via the efficient inference of neural networks and propose a brand new framework: Fast Anti-Jamming Radar Deployment Algorithm (FARDA). We first model the radar deployment problem as an end-to-end task and design deep reinforcement learning algorithms to solve it, where we develop integrated neural modules to perceive heatmap information and a brand new reward format. Empirical results demonstrate that our method achieves coverage comparable to evolutionary algorithms while deploying radars approximately 7,000 times faster. Further ablation experiments confirm the necessity of each component of FARDA.

A Fast Anti-Jamming Cognitive Radar Deployment Algorithm Based on Reinforcement Learning

TL;DR

The paper tackles fast, anti-jamming deployment of cognitive radars by formulating the problem as a combinatorial optimization and solving it with a deep reinforcement learning framework (FARDA). FARDA introduces problem simplifications, an encoder-based policy network, and a novel CVDP-EXPR reward to enable efficient end-to-end learning, achieving deployment speeds about 7,000× faster than evolutionary baselines while maintaining comparable coverage. Ablation studies verify the necessity of the encoder and the reward design. The work demonstrates the potential for real-time radar deployment in contested electromagnetic environments and lays groundwork for extending to more complex static and dynamic scenarios.

Abstract

The fast deployment of cognitive radar to counter jamming remains a critical challenge in modern warfare, where more efficient deployment leads to quicker detection of targets. Existing methods are primarily based on evolutionary algorithms, which are time-consuming and prone to falling into local optima. We tackle these drawbacks via the efficient inference of neural networks and propose a brand new framework: Fast Anti-Jamming Radar Deployment Algorithm (FARDA). We first model the radar deployment problem as an end-to-end task and design deep reinforcement learning algorithms to solve it, where we develop integrated neural modules to perceive heatmap information and a brand new reward format. Empirical results demonstrate that our method achieves coverage comparable to evolutionary algorithms while deploying radars approximately 7,000 times faster. Further ablation experiments confirm the necessity of each component of FARDA.

Paper Structure

This paper contains 20 sections, 19 equations, 5 figures, 5 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Problem Formulation
  3. Methodology: FARDA
  4. Problem Simplification
  5. Dimension Reduction
  6. Relaxation
  7. Environment Simplification
  8. Formulation of MDP
  9. Policy Network
  10. Encoder Network
  11. The Actor and Critic Network
  12. Reward Based on Constraint Violation Degree Penalty and Exponential Function
  13. Experiments
  14. Experimental Setups
  15. Main Results
  16. ...and 5 more sections

Figures (5)

  • Figure 1: An example figure of the problem.
  • Figure 2: The formulation FARDA, concluding environment shaping, encoder, actor, and critic network, and CVDP-EXPR.
  • Figure 3: When using PSO and GA, the results obtained from deploying within the deploy area $D_R$ versus along its boundary $B_R$. Correspondingly: (a) using PSO and deploy in $D_R$, (b) using GA and deploy in $D_R$, (c) using PSO and deploy in $B_R$, (d) using GA and deploy in $B_R$. The four cyan dots indicate radar positions, while the upper three yellow dots represent jamming node positions. The color of the heatmap ranges from blue, indicating a detection probability greater than 0.5, to red, indicating a detection probability less than 0.5.
  • Figure 4: (a) The heatmap of sampling points in 100 meters. (b) The heatmap of sampling points in 500 meters. The gray rectangle region of (b) denotes the region we will discard. Note that in (b), the positions of the two radars are not shown.
  • Figure 5: (a) An example radar deployment for FARDA, (b) An example of removing CVDP in FARDA. The result of removing CVDP-EXPR is the same as (b).