الفهرس 
Chapter 1 Introduction to Software Defined NetworksOnly 14 pages are availabe for public view
 |  | from | 94 |  |  |
| | | from | 94 | | |
Abstract
With the increase in smart devices, performance of traditional networks is limited by this huge amount of generated traffic flows. A scalable and programmable networking solution can be achieved in software defined network (SDNs) through the separation between control plane and forwarding plane. This advantage can allow machine learning (ML) applications to control and automate networks. Concurrently, network slicing (NS) is a promising technology, it is necessary to meet the variety of service needs and bandwidth requirements. It provides network as a service (NaaS). So, combining NS and ML in SDNs can achieve good resource management. This thesis focuses on applying real-time network traffic analysis with its suitable network slice according to traffic flows classification. There are two suggested models, the first model is a network traffic slicing based on traffic classification, robust scale was used to scale the features instead of max/min normalization. Also, k-means clustering algorithm was used to separate the dataset into optimum number of different clusters (slices). Five different supervised models were applied to achieve high classification accuracy. A standard dataset was used for training and testing the models. The highest accuracy obtained from artificial neural network and it was 98.2%, while support vector machine (SVM) with linear function gave an accuracy of 96.7%. The second one is a real-time traffic classification. Five classes of traffic (streaming, VoIP, telnet, DNS and ping) were generated and captured directly from the controller through SDN environment. Recursive features elimination (RFE) was used to select the best features which used to train six supervised ML models for classification. Stratified K-folds with k=4 was used to split the generated dataset into train and test. The highest average classification accuracy was 98.1% from naïve bayes and this model was used for real-time test. The challenges faced are collecting data from SDN’s controller to apply real-time traffic flow classification, which is a primary step to assign each flow to its suitable network slice (Bandwidth). Mapping between forward and backward flows was a big challenge also.