The rapid growth of technology has led to the creation of computing networks. The applications of the Internet of Things are becoming more and more visible with the expansion and development of sensors and the use of a series of equipment to connect to the Internet. Of course, the growth of any network will also provide some challenges. The main challenge of IoT like any other network is its security. In the field of security, there are issues such as attack detection, authentication, encryption and the so on. One of the most important attack is cyber-attacks that disrupt the network usage. One of the most important attacks on the IoT is BotNet attack. The most important challenges of this topic include very high computational complexity, lack of comparison with previous methods, lack of scalability, high execution time, lack of review of the proposed approach in terms of accuracy to detect and classify attacks and intrusions. Using intrusion detection systems for the IoT is an important step in identifying and detecting various attacks. Therefore, an algorithm that can solve these challenges has provided a near-optimal method. Using training-based models and algorithms such as Deep Dearning-Reinforcement Learning and XGBoost learning in combination (DRL-XGBoost) models can be an interesting approach to overcoming previous weaknesses. The data of this research is Bot-IoT-2018.
翻译:技术的迅速发展导致了计算机网络的建立。随着传感器的扩大和开发以及一系列连接互联网的设备的使用,“物联网”的应用越来越明显。当然,任何网络的增长也将带来一些挑战。象任何其他网络一样,IoT的主要挑战在于其安全。在安全领域,有一些问题,如攻击探测、认证、加密等。最重要的攻击之一是破坏网络使用的网络攻击。对IoT攻击的最重要攻击之一是BotNet攻击。这个专题的最重要挑战包括极复杂的计算、与以往方法缺乏比较、缺乏可缩放性、执行时间长、在探测和分类攻击和入侵的准确性方面缺乏对拟议方法的审查。在IoT使用入侵探测系统是查明和发现各种攻击的一个重要步骤。因此,能够解决这些挑战的算法提供了一种近乎最佳的方法。在深地Dest-Regain-Regure-Regure-Regurement中采用的培训模型和算法方法,例如深地DAR-Regure-Regence-Regregreat-BGBX) 学习和XOGBX研究中的前一个令人感兴趣的组合。