STLGBM-DDS: An Efficient Data Balanced DoS Detection System for Big Data Wireless Sensor Networks

PROJECT TITLE :

STLGBM-DDS: An Efficient Data Balanced DoS Detection System for Wireless Sensor Networks on Big Data Environment

ABSTRACT:

During the data collection and transmission processes, Wireless Sensor Networks (WSNs) are susceptible to a variety of different security risks and threats that are unique to themselves. DoS attacks are among the most common types of attacks carried out against WSNs. These attacks can target any layer of the protocol stack. Within the scope of this investigation, a novel Distributed Denial of Service Intrusion Detection System, or DDS, is proposed as a method for identifying DDoS assaults that are directed toward WSNs. The proposed system is an ensemble intrusion detection system known as STLGBM-DDS. It was developed on the Apache Spark Big Data platform within the Google Colab environment. The system combines the LightGBM Machine Learning algorithm with the data balancing and feature selection procedures. Synthetic Minority Oversampling Technique (SMOTE) and Tomek-Links sampling methods called STL were used in the data imbalance processing that was carried out. This was done in order to mitigate the negative effects that data imbalance has on the performance of the system. In addition, during the data preprocessing stage, the Information Gain Ratio was implemented as a technique for selecting features. Investigation was done into how the system's detection capabilities were affected by the stages of data balancing and feature selection respectively. Accuracy, F-Measure, Precision, Recall, ROC Curve, and Precision-Recall Curve were the evaluation parameters that were utilized in order to assess the results that were obtained. As a consequence of this, the proposed method accomplished an accuracy level of 99.95% overall. Also, it achieved an accuracy performance of 99.99% in the Normal class, 99.96% in the Grayhole class, 99.98% in the Blackhole class, 99.92% in the TDMA class, and 99.87% in the Flooding class, respectively. The results that were obtained indicate that the proposed method has achieved very successful results in the detection of DoS attacks in WSNs in comparison to the methods that are currently in use.

Did you like this research project?

To get this research project Guidelines, Training and Code... Click Here