Scholarlink

<p>The mobile ad hoc networks (MANETs) are scattered, decentralized networks with portable nodes that are connected without the need for a fixed base station. The MANET’s nodes travel continuously in arbitrary directions and behave arbitrarily, which presents several difficulties for these networks and increases their vulnerability to various security risks. Furthermore, diverse existing schemes susceptible to insider attacks and the absence of a central authority also complicate the deployment and management of intrusion detection systems across the network. To tackle these complications, this research is to introduce the Fractional Secretary Bird Red Panda Optimization-based Deep Maxout Network (FSBRPO_DMN) for intrusion detection. Initially, MANET nodes are simulated that incorporate both energy and trust models. Subsequently, multipath routing is performed by Fractional Secretary Bird Red Panda Optimization (FSBRPO) based on certain objective functions. Moreover, FSBRPO is designed by combining Secretary Bird Red Panda Optimization (SBRPO) and Fractional Calculus (FC). At Base Station (BS), Quantile Normalization is employed to normalize input log data. After that, feature fusion is conducted by Clark Distance with ResNeSt. Finally, intrusion detection is performed using a Deep Maxout Network (DMN). In addition, the FSBRPO gained delay, energy and trust as 0.548 ms, 3.350 J, and 90.291, as well as FSBRPO_ DMN accomplished True positive Rate (TPR), True Negative Rate (TNR), and accuracy of 92.310%, 92.540%, and 92.754%.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Multipath routing and intrusion detection in mobile ad hoc network using hybrid optimization-based deep maxout network

  • 期刊论文

  • 发表日期：

    2026年3月4日

查看全文
  • T. Jaya,
  • N. Maradona

摘要

The mobile ad hoc networks (MANETs) are scattered, decentralized networks with portable nodes that are connected without the need for a fixed base station. The MANET’s nodes travel continuously in arbitrary directions and behave arbitrarily, which presents several difficulties for these networks and increases their vulnerability to various security risks. Furthermore, diverse existing schemes susceptible to insider attacks and the absence of a central authority also complicate the deployment and management of intrusion detection systems across the network. To tackle these complications, this research is to introduce the Fractional Secretary Bird Red Panda Optimization-based Deep Maxout Network (FSBRPO_DMN) for intrusion detection. Initially, MANET nodes are simulated that incorporate both energy and trust models. Subsequently, multipath routing is performed by Fractional Secretary Bird Red Panda Optimization (FSBRPO) based on certain objective functions. Moreover, FSBRPO is designed by combining Secretary Bird Red Panda Optimization (SBRPO) and Fractional Calculus (FC). At Base Station (BS), Quantile Normalization is employed to normalize input log data. After that, feature fusion is conducted by Clark Distance with ResNeSt. Finally, intrusion detection is performed using a Deep Maxout Network (DMN). In addition, the FSBRPO gained delay, energy and trust as 0.548 ms, 3.350 J, and 90.291, as well as FSBRPO_ DMN accomplished True positive Rate (TPR), True Negative Rate (TNR), and accuracy of 92.310%, 92.540%, and 92.754%.