Abstract: Aimed at the problem of model deviation caused by factors such as legal punishment and information asymmetry in the current research on network defense strategy, a network defense strategy based on taboo random game is proposed. By discretizing the offensive and defensive process, a taboo random game model was constructed. The tabu search method with memory function was used to conduct bounded rational random game analysis, and realized the memory function through the taboo table data structure. The game model and data-driven memory were combined to obtain the best network attack defense strategy, and the simulation experiment method was used to verify the effectiveness of the proposed network defense strategy. The experimental results show that the quantification accuracy of network offensive and defense benefits of the network defense strategy in this paper has been effectively improved. Compared with the current existing network defense strategies, the method proposed in this paper has lower space complexity and achieves rapid convergence of the algorithm, which reduces the computational resource consumption of the algorithm and improves the accuracy of network defense revenue.