DCRNN: Deep Convolution Reinforcement Neural Network-Based Cyber-Attacks Detection In Iot Data

Abstract

Honeyed framework is a system environment for defending legitimate network resources against attack. The Honeyed framework fosters resource-stealing behavior by encouraging attackers to use it. This is a procedure for detecting an attack using an attack detection procedure. To recognize denial of service (DoS) threats, we employ the Honeyed framework system in this research. Primary security devices to prevent your network by enabling the identification of attacks in the face of network attacks are NIDS (Network Intrusion Detection Systems). We propose a system that exposes an attack and verifies a defense mechanism against the same attack in this paper. For the new cyber security benchmark IoT dataset, this white paper tests the recent machine learning (ML) approaches. The primary purpose of this study is to develop a system that can forecast also secure malware, botnets, and DDOS attacks using Honeyed styles. The goal of this architecture is to accurately represent the data and create an effective cybersecurity predictive agent. Deep Convolution Reinforcement Neural Networks (DCRNN) are used to monitor networks and classify network users as attackers. This proposed method uses a two-step network understanding skills to increase its functionality. For feature engineering issues, the first step, data preprocessing, employs DSAE (Deep Sparse Auto Encoder). The Deep Convolution Reinforcement Neural Network learning approach is used in the second step for classification. The Honeyed framework is then installed, which includes the honeyed firewall and web server. The DCRNN deployment is full, and network users can now be monitored and analyzed. The impact of the published method was evaluated using data collected in a loT environment, specifically heterogeneous datasets such as 'LITNET-2020,' 'NetML-2020, and 'IoT-23,'. Considering the statistical significance of the outcomes of this approach's assessment will be tested using state-of-the-art network detection approaches.