1. Machine learning for qos optimization in early attack detection networks
Authors : Dharmaveer P. Choudhari, Sanjay. S. Dorle
Pages : 26-30
DOI : http://dx.doi.org/10.21172/1.143.05
Keywords : Early attack detectionQoSmachine learningefficiency Abstract :Early attack detection in wireless networks generally requires a lot of computational steps, and pre-emptive calculations, which in turn reduce the quality of service (QoS) parameters like end to end delay of communication, the packet delivery ratio and most importantly the lifetime of the network. Such networks usually have high security performance, but compromise on crucial and time sensitive parameters, thereby making the networks inefficient for real time applications. In this paper, we propose a machine learning (ML) layer for optimization of the QoS parameters for early stage attack detection networks. The proposed protocol uses multiple one-time computational parameters which can be evaluated before network deployment, and then used recursively throughout the network’s lifetime without much modifications. Evaluation of the proposed protocol was done on various network scenarios, which indicated a 9% performance improvement in terms of end-to-end communication delay, and more than 15% improvement in terms of network lifetime, while it also shows about3% improvement in network throughput and packet delivery ratio, which is due to a trade-off in the fitness function of the proposed machine learning algorithm, and can be fine tuned as per the application’s requirements. These optimizations are achieved without compromising on the network’s attack detection probability.
Citing this Journal Article :Dharmaveer P. Choudhari, Sanjay. S. Dorle, "Machine learning for qos optimization in early attack detection networks", Volume 14 Issue 3 - September 2019, 26-30
Click here to Submit Copyright Takedown Notice for this article.