Performance comparison of different methods.

Against the backdrop of the rapid development of wireless communication technology, the complex signal interference issues in the electromagnetic spectrum environment have become a key factor affecting the quality and reliability of signal transmission. Existing solutions, such as traditional interference suppression techniques that rely on static spectrum allocation and fixed interference patterns, are no longer able to adapt to the rapidly changing electromagnetic environment and face computational complexity challenges when processing large amounts of real-time data. This study proposes an intelligent anti-interference algorithm that combines deep neural networks and game theory, and constructs a model based on near-end strategy optimization. By extracting and processing signal features through deep neural networks, and dynamically adjusting communication strategies with near-end optimization, the model effectively addresses the recognition and prediction of signal transmission feature parameters in target communication systems, generates interference signals with the same feature parameters, and achieves effective interference suppression. Experiments show that the proposed model achieves an accuracy rate of 95.23% in identifying interference signals and an anti-interference accuracy rate of 85.47%, significantly outperforming random forest and deep Q-network models. The study not only clarifies the limitations of existing solutions but also precisely defines the goals of the new model, which are to reduce error rates and improve adaptability in dynamic environments. The results further explain the significance of the used metrics and test conditions, providing new means and strategies for the development of anti-interference communication technology, especially in dealing with new complex electromagnetic spectrum interference.

在无线通信技术高速发展的背景下，电磁频谱环境中的复杂信号干扰问题已成为影响信号传输质量与可靠性的关键因素。现有解决方案，如依赖静态频谱分配与固定干扰模式的传统干扰抑制技术，已无法适配快速变化的电磁环境，且在处理海量实时数据时面临计算复杂度难题。本研究提出一种融合深度神经网络与博弈论的智能抗干扰算法，并构建基于近端策略优化（near-end strategy optimization）的模型。该模型通过深度神经网络提取并处理信号特征，结合近端策略优化动态调整通信策略，可有效实现目标通信系统中信号传输特征参数的识别与预测，生成具有相同特征参数的干扰信号，并达成高效的干扰抑制效果。实验结果表明，所提模型在干扰信号识别任务中准确率达95.23%，抗干扰准确率达85.47%，性能显著优于随机森林（Random Forest）与深度Q网络（Deep Q-Network）模型。本研究不仅阐明了现有解决方案的局限性，还明确界定了新型模型的优化目标：降低错误率并提升动态环境下的适配性。本研究结果进一步阐释了所采用评价指标与测试条件的重要意义，为抗干扰通信技术的发展提供了全新的手段与策略，尤其在应对新型复杂电磁频谱干扰场景中具备应用价值。