A Radar Anti-Jamming Strategy Based on Game Theory With Temporal Constraints

Abstract

The confrontation between radar and jammer is increasingly competitive in electromagnetic spectrum warfare. The current radar anti-jamming methods are constrained to some extent in complex electromagnetic environment. To improve the anti-jamming capacity of radar system, a game theory-based optimization method is proposed to enhance the decision-making of anti-jamming strategy in this paper. First, a temporal sequence interaction based dynamic game model between radar and jammer is constructed. Then, Q-learning is performed to optimize the radar anti-jamming strategies. The simulation experimental results show that the proposed method can significantly improve the radar winning probability in the confrontation.

A radar anti‐jamming strategy optimisation based on Stackelberg game

“…The game is composed of multiple rounds whose number is determined by the total game interval, the recognition interval and preparation interval in each round. This process between radar and naval vessel is called temporal sequence interaction model [21].…”

“…In the reinforcement learning optimisation method proposed in Ref. [21], the radar utility when successfully recognising the action of naval vessel is given as follows: $$${G}_{RL0}=P(i)\ast \left({T}_{R}(a)+{T}_{P}(s)-{T}_{R}(s)-{T}_{P}(a)\right),$$$ where G RL 0 is the utility when radar recognises the action of naval vessel accurately.…”

“…It can be seen that in case of inaccurate recognition, the radar utility of reinforcement learning model [21] and radar in Stackelberg game model are actually the same, as shown in Equation (20), which means under inaccurate recognition the radar utility will be negative in the whole round. In contrast, when the recognition is successful, as shown in Equation (19), the radar gains much more from the Stackelberg game model than from the reinforcement learning optimisation.…”

“…Ref. [21] put forward a radar anti‐jamming game strategy optimisation method based on the non‐real‐time characteristic of radar and naval vessel actions, and improved the radar winning probability by Q‐learning under the condition of temporal constraints.…”

“…In the model built in Ref. [21], if the radar and naval vessel do not perceive the opponent's action changing in the game process, they tend to keep their actions unchanged, resulting in their own actions always taking effect later than the opponent's. As each recognition interval and preparation interval are constant during the game, there is an upper winning probability limit after optimisation.…”