A smart reactive jamming approach to counter reinforcement learning-based antijamming strategies in GEO SATCOM scenario,International Journal of Satellite Communications and Networking

当前位置： X-MOL 学术 › Int. J. Satell. Commun. Netw. › 论文详情

Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)

A smart reactive jamming approach to counter reinforcement learning-based antijamming strategies in GEO SATCOM scenario
International Journal of Satellite Communications and Networking ( IF 0.9 ) Pub Date : 2021-07-27 , DOI: 10.1002/sat.1418
Shahzad Arif _{1,

2} , Ali Javed Hashmi ₁ , Waseem Khan ₃ , Rizwana Kausar ₄

Affiliation

Reinforcement learning (RL) is being considered for future SATCOM systems due to its inherent capability to self-learn the optimum decision-making policy under different scenarios. This capability enables SATCOM systems to manage their resources judiciously and mitigate jamming attacks autonomously without prior jammer type classification. We propose a novel smart reactive SATCOM jamming approach that would not only counter these RL based anti-jamming strategies but would also be effective against conventional anti-jamming schemes, that is, FHSS and DSSS. The proposed jamming approach exploits the limitations in learning patterns of Q-learning-based RL agent and achieves effective jamming while conserving considerable amount of jamming power. To achieve this, we propose an intelligent jamming engine (IJE) along with few potent jamming algorithms and evaluate their performance in terms of throughput degradation of victim SATCOM link, jamming power conservation, and design complexity of the jammer. Software simulations successfully demonstrate the effectiveness of our proposed smart reactive jamming approach which outperforms the standard reactive jammer against RL-based antijamming schemes.

中文翻译：

GEO SATCOM场景中基于强化学习的抗干扰策略的智能反应干扰方法

强化学习 (RL) 正在考虑用于未来的 SATCOM 系统，因为它具有在不同场景下自学习最优决策策略的内在能力。这种能力使卫星通信系统能够明智地管理其资源并自主减轻干扰攻击，而无需事先进行干扰类型分类。我们提出了一种新颖的智能反应卫星通信干扰方法，不仅可以对抗这些基于 RL 的抗干扰策略，而且还可以有效对抗传统的抗干扰方案，即 FHSS 和 DSSS。所提出的干扰方法利用了基于 Q-learning 的 RL 代理的学习模式的局限性，并在节省大量干扰功率的同时实现了有效的干扰。为达到这个，我们提出了一种智能干扰引擎（IJE）以及一些有效的干扰算法，并在受害卫星通信链路的吞吐量下降、干扰功率节约和干扰机的设计复杂性方面评估它们的性能。软件仿真成功地证明了我们提出的智能反应性干扰方法的有效性，该方法在基于 RL 的抗干扰方案中优于标准反应性干扰器。

更新日期：2021-07-27

点击分享查看原文

点击收藏

阅读更多本刊新发论文本刊介绍/投稿指南