Distributed Denial of Service (DDoS) attacks can put the communication networks in instability by throwing malicious traffic and requests in bulk over the network. Computer networks form a complex chain of nodes resulting in a formation of vigorous structure. Thus, in this scenario, it becomes a challenging task to provide an efficient and secure environment for the user. Numerous approaches have been adopted in the past to detect and prevent DDoS attacks but lack in providing efficient and reliable attack detection. As a result, there is still notable room for improvement in providing security against DDoS attacks. To overcome the problem of DDoS attacks detection, in this paper, a novel high-efficient approach is proposed named DIDDOS to protect against real-world new type DDoS attacks using Gated Recurrent Unit (GRU) a type of Recurrent Neural Network (RNN). For effective performance results different classification algorithms are applied Gated Recurrent Units (GRU), Recurrent Neural Networks (RNN), Naive Bayes (NB), and Sequential Minimal Optimization (SMO) are utilized to detect and identify DDoS attacks. For the performance evaluation metrics like accuracy, recall, f1-score, precision are used to evaluate the efficiency of the machine and deep learning classifiers. Experimental results yield the highest accuracy of 99.69% for DDoS classification in case of reflection attacks and 99.94% for DDoS classification in case of exploitation attacks using GRU.

分布式拒绝服务（DDoS）攻击可以通过在网络上大量散布恶意流量和请求，使通信网络不稳定。计算机网络形成了复杂的节点链，从而形成了有力的结构。因此，在这种情况下，为用户提供有效和安全的环境成为一项具有挑战性的任务。过去已采用多种方法来检测和预防DDoS攻击，但缺乏提供有效且可靠的攻击检测的能力。结果，在提供针对DDoS攻击的安全性方面仍存在显着的改进空间。为了克服DDoS攻击检测的问题，本文提出了一种新的高效方法DIDDOS使用门控循环单元（GRU）（一种循环神经网络（RNN））来防御现实世界中的新型DDoS攻击。为了获得有效的性能结果，应用了不同的分类算法，分别使用门控循环单元（GRU），循环神经网络（RNN），朴素贝叶斯（NB）和顺序最小优化（SMO）来检测和识别DDoS攻击。对于性能评估指标，如准确性，召回率，f1分数，精度，用于评估机器和深度学习分类器的效率。实验结果得出，对于反射攻击，DDoS分类的准确性最高，为99.69％；对于使用GRU进行的攻击，DDoS分类的准确性最高，为99.94％。