Frequency jamming resource allocation method based on temporal reinforcement learning

doi:10.1016/j.sigpro.2025.110330

Frequency jamming resource allocation method based on temporal reinforcement learning

2025-10-08

0

PRE

AI

DOI：10.1016/j.sigpro.2025.110330

Original

Original request for help

Save

Abstract

En 中文

In order to solve the problem that the resource allocation of jamming frequency domain lags behind the radar frequency change when the jammer is dealing with the frequency agility signal, a frequency domain parameter decision method based on time series reinforcement learning is proposed. Firstly, the frequency agile signals are simulated and modeled as a partially observable Markov decision process based on the problem of incomplete observation caused by decision lag in frequency domain. At the same time, the gated recurrent unit is proposed to infer the hidden state from the observed past information, which makes up for the imperfect observation at the current moment. It improves the cognition of the agent to the environment. Meanwhile, this paper uses the proximal policy optimization algorithm to establish the frequency domain parameter allocation model. The experimental results show the proposed method can achieve frequency coverage comparable to long short-term memory-proximal policy optimization while using narrower bandwidth. Specifically, its frequency coverage exceeds 97% under the three frequency hopping modes. In addition, in the generalization tests, the algorithm can also maintain a frequency coverage rate of over 95%, effectively enhancing jamming frequency decision capability for different agile systems in complex electromagnetic environments.

Frequency jamming resource allocation method based on temporal reinforcement learning

Abstract

Journal

Organization