Hybrid based traffic classification of online internet application

Also available in printed version

التفاصيل البيبلوغرافية
المؤلف الرئيسي: Elkarim, Hamza Awad Hamza Ibrahim
مؤلفون آخرون: Sulaiman Mohd. Noor, supervisor
التنسيق: Doctoral thesis
اللغة:الإنجليزية
منشور في: Universiti Teknologi Malaysia 2025
الموضوعات:
الوصول للمادة أونلاين:https://utmik.utm.my/handle/123456789/60131
Abstract Abstract here
_version_ 1854975100679356416
author Elkarim, Hamza Awad Hamza Ibrahim
author2 Sulaiman Mohd. Noor, supervisor
author_facet Sulaiman Mohd. Noor, supervisor
Elkarim, Hamza Awad Hamza Ibrahim
author_sort Elkarim, Hamza Awad Hamza Ibrahim
description Also available in printed version
format Doctoral thesis
id utm-123456789-60131
institution Universiti Teknologi Malaysia
language English
publishDate 2025
publisher Universiti Teknologi Malaysia
record_format dspace
record_pdf Abstract
spelling utm-123456789-601312025-08-21T09:34:42Z Hybrid based traffic classification of online internet application Elkarim, Hamza Awad Hamza Ibrahim Sulaiman Mohd. Noor, supervisor Electrical engineering Also available in printed version Internet traffic classification has gained significant attention in the last few years. Even though several classification approaches were proposed by the research community, there still exist many open problems on Internet traffic classification. The unreliability of port and payload based classification methods due to unknown ports and encrypted traffic motivates researchers towards adopting Machine Learning (ML) approach. However, the effect on ML performance when training and testing datasets are from different network environments has not been formally addressed. Hybrid classifier was proposed to overcome the limitation of individual port, statistical, and signatures classifiers. But, these hybrid classifiers final results are only based on any one of the individual classifier results. The issue of live online traffic is also not considered. This research is divided into two main parts. The first part considers the effect of three training and testing dataset scenarios obtained from different network segment on ML performance. The second part proposes hybrid Signature Statistical Port Classifiers (SSPC) that not only integrate, but also synergize the advantages of port, signature and statistical classifier. Both parts use real datasets collected from our campus network. The ML datasets were found to influence the traffic features, causing the classification accuracy to enhance if the training and testing dataset collected from the same network segment. Besides, the false positive will decrease if the training and testing datasets were collected from the same network level. The proposed hybrid classifier was used to classify four Internet application classes i.e. web, FTP, Skype, and online game in two stages, initially offline and later online. In the offline stage, SSPC produces more than 95% classification accuracy which is higher when compared with other individual classifiers. As demonstrated in live online experiments, SSPC achieve more than 91% accuracy; therefore, it is suitable to be used for online classification atiff UTM 229 p. Thesis (Ph.D (Kejuruteraan Elektrik)) - Universiti Teknologi Malaysia, 2014 2025-03-17T06:30:27Z 2025-03-17T06:30:27Z 2014 Doctoral thesis https://utmik.utm.my/handle/123456789/60131 valet-20170430-18147 vital:98479 ENG Closed Access UTM Complete Unpublished application/pdf Universiti Teknologi Malaysia
spellingShingle Electrical engineering
Elkarim, Hamza Awad Hamza Ibrahim
Hybrid based traffic classification of online internet application
thesis_level PhD
title Hybrid based traffic classification of online internet application
title_full Hybrid based traffic classification of online internet application
title_fullStr Hybrid based traffic classification of online internet application
title_full_unstemmed Hybrid based traffic classification of online internet application
title_short Hybrid based traffic classification of online internet application
title_sort hybrid based traffic classification of online internet application
topic Electrical engineering
url https://utmik.utm.my/handle/123456789/60131
work_keys_str_mv AT elkarimhamzaawadhamzaibrahim hybridbasedtrafficclassificationofonlineinternetapplication