GA-Dueling DQN Jamming Decision-Making Method for Intra-Pulse Frequency Agile Radar.

Area of Science:

• Electronic Warfare
• Cognitive Jamming Systems
• Deep Reinforcement Learning

Background:

• Optimizing jamming strategies is critical for cognitive jamming systems operating in dynamic electromagnetic environments.
• Frequency-agile radars, with their ability to rapidly change carrier frequencies, present significant challenges for intelligent jamming decision-making.
• Traditional methods struggle to adapt to the rapid frequency changes at the sub-pulse level characteristic of Intra-Pulse Frequency Agile Radar.

Purpose of the Study:

• To develop intelligent jamming decision-making algorithms for Intra-Pulse Frequency Agile Radar.
• To enhance the performance of cognitive jamming systems against advanced radar threats.
• To address the limitations of traditional jamming methods in dynamic frequency environments.

Main Methods:

• Utilized deep reinforcement learning for intelligent jamming frequency selection.
• Proposed a novel GA-Dueling DQN (GRU-Attention-based Dueling Deep Q Network) method.
• Employed Gated Recurrent Units (GRU) and an attention mechanism to capture long-term dependencies and sequence data.

Main Results:

• The GA-Dueling DQN method demonstrated superior jamming effectiveness compared to traditional Q-learning, DQN, and Dueling DQN.
• Achieved faster convergence speeds in learning optimal jamming strategies.
• Showcased reduced reliance on prior knowledge, indicating greater adaptability.

Conclusions:

• The proposed GA-Dueling DQN offers significant advantages for jamming sub-pulse-level frequency-agile radar.
• Deep reinforcement learning, enhanced with GRU and attention, provides an effective solution for dynamic electromagnetic environments.
• This approach enhances the capability of cognitive jamming systems in complex threat scenarios.