Title: Optimal SVM classifier-based cross-layer design in an ad hoc wireless network
Authors: Ridhima Mehta
Addresses: School of Computer and Systems Sciences, Jawaharlal Nehru University, New Delhi – 110067, India
Abstract: The rapid advancement of wireless technology and routing devices has led to the expeditious evolution of the ad hoc type of networking. The infrastructure-less dynamic network with the error-prone wireless medium in the resource-constrained ad hoc communication system poses several challenges for efficient routing and design optimisation. In this paper, an optimal cross-layer design architecture for an ad hoc wireless network is developed based on the supervised categorisation algorithm. Specifically, the support vector machine (SVM) classification scheme is employed to evaluate the margin and error associated with the disparate features of a wireless communication system. This technique ensures that the margin obtained with the computed linear separating plane is maximum from the labelled training samples belonging to two different categories of a two-class problem. The contemplated networking attributes considered for the integrated application of cross-layer information exchange and binary SVM models include the throughput, persistence probability, and transmit power associated with the directed wireless links.
Keywords: ad hoc network; cross-layer design; persistence probability; power; SVM; support vector machine; throughput.
DOI: 10.1504/IJAACS.2024.142195
International Journal of Autonomous and Adaptive Communications Systems, 2024 Vol.17 No.5, pp.383 - 405
Received: 22 Nov 2021
Accepted: 11 Apr 2022
Published online: 14 Oct 2024 *