An Approach for Semi-Supervised Machine Learning-Based Mobile Network Anomaly Detection With Tagging

Abstract

The world economy has been stable by emerging into online business and activity with increased online users. There is likelihood to escalate the fraud activity and misuse the corporation's network. Hence, strengthening of network security is necessary to prevent such unwanted activities. In this work, Anomaly Detection System (ADS) is proposed to detect the anomalous activities in the network. Firstly, network packets with the tagging are trained with the k-nearest neighbor algorithm (KNN) and Kohonen’s Self-Organizing Maps (KSOM) algorithm clusters the network packets. Initially, the Tagging Application (TA) dataset is created that contains network packets with the labelling of applications by extracting captured live packets using high computing server that is configured in data center which are used for the proposed Fix Weight Kohonen's Self-Organizing Maps (FW-KSOM) to cluster different activities in the network. Implementation of the proposed ADS model for labelling and clustering is carried out in real time networking scenario to identify the applications for anomaly detection.