Multicast On-Route cluster propagation to detect network intrusion detection systems on MANET using Deep Operator Neural networks

作     者：Saminathan, Karunakaran Perumal, Latha Shajin, Francis H. Shakya, Rajeev Kumar 

作者机构：Kongu Engn Coll Dept Comp Technol Erode 638060 Tamil Nadu India St Josephs Coll Engn Dept Elect & Commun Engn Chennai India Xpertmindz Innovat Solut Pvt Ltd Dept Elect & Commun Engn Res & Dev Kuzhithurai Tamil Nadu India Adama Sci & Technol Univ Dept Elect & Commun Engn Adama Ethiopia 

出 版 物：《EXPERT SYSTEMS WITH APPLICATIONS》 (Expert Sys Appl)

年 卷 期：2025年第271卷

核心收录：

学科分类：1201[管理学-管理科学与工程(可授管理学、工学学位)] 0808[工学-电气工程] 08[工学] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

主　　题：Artificial Rabbits Optimization Cluster Head Deep Operator Neural Networks Internet Of Things Mobile Ad Hoc Network Simple Contrastive Graph Clustering Wavelet Transform 

摘      要：Mobile Ad-hoc Networks (MANETs) face significant challenges related to security threats and energy efficiency due to their dynamic nature. Traditional approaches have struggled with optimizing energy consumption while ensuring robust security. This paper introduces a novel multipath routing protocol for MANETs that integrates Simple Contrastive Graph Clustering (SCGC) with Deep Operator Neural Networks (DONN) to enhance intrusion detection for various attack types including DoS, R2L, U2R, and Probe. The protocol utilizes Artificial Rabbits Optimization (ARO) to prioritize secure and energy-efficient nodes during multipath routing, selecting optimal paths based on metrics such as throughput, energy efficiency, trust, and network connectivity. The proposed method, implemented in Python, is evaluated using metrics such as packet receiving percentage ratio (PRPR), detection rate, throughput, energy consumption, precision, link failure rate, and network lifetime. Comparative analysis demonstrates that the proposed MRCP-DNID-MANET-DONN approach significantly outperforms existing methods, with improvements in throughput by 25.26 %, 16.22 %, and 26.27 %, and detection rate by 18.29 %, 24.31 %, and 23.26 % when compared to AE-IDS-MANET, MSAGCN-MIS-MANET, and DBF-MID-MANET, *** proposed MRCP-DNID-MANET-DONN approach significantly improves throughput and detection rate outperforming existing methods in both security and energy efficiency in MANETs.