Traffic classification in SDN based Iot network using two level fused network with self adaptive manta ray foraging

Project Code :TCMAPY2247

Objective

Software Defined Networking (SDN)–based IoT networks generate massive and diverse traffic, making accurate classification essential for detecting benign and malicious flows. Traditional machine learning models such as Bi-LSTM, Decision Tree, and Random Forest often struggle with high-dimensional feature spaces, limiting their performance. To address this challenge, the project introduces a two-level fused model combined with a Self-Adaptive Manta Ray Foraging Optimization (SMRFO) algorithm for efficient feature selection. SMRFO identifies the most relevant features by maximizing classification accuracy while minimizing redundancy. The selected features are then fed into advanced classifiers including XGBoost, AdaBoost, and a Stacking Classifier to enhance detection capability. This fusion strategy improves overall performance while reducing computational complexity. The system is implemented using Python and the Flask framework, incorporating modules for user authentication and traffic classification. Trained and evaluated on the CICDarknet2020 dataset, the proposed model achieves superior accuracy, precision, recall, and F1-score compared to traditional machine learning approaches.

Abstract

Software Defined Networking (SDN) based IoT networks generate large volumes of traffic with different behavior patterns. Accurate classification of this traffic is necessary to identify benign and malignant flows. Traditional machine learning models such as Bi-LSTM, Decision Tree, and Random Forest often face limitations in achieving high classification performance when feature dimensions are large. To overcome this issue, this project proposes a two-level fused model integrated with a Self-Adaptive Manta Ray Foraging Optimization (SMRFO) algorithm for feature selection.The SMRFO algorithm selects the most relevant features by optimizing classification accuracy while reducing feature redundancy. The optimized features are then used with advanced classifiers including XGBoost, AdaBoost, and Stacking Classifier. This fusion improves detection performance and reduces computational complexity. The system is implemented using Python and Flask framework, providing modules for user authentication and traffic classification. The model is trained and tested using the CICDarknet2020 dataset. Experimental evaluation demonstrates improved accuracy, precision, recall, and F1-score compared to traditional models.

Keywords: SDN, IoT Traffic Classification, SMRFO, Feature Selection, XGBoost, AdaBoost, Stacking Classifier, Darknet Detection, Ensemble Learning, Network Security.

NOTE: Without the concern of our team, please don't submit to the college. This Abstract varies based on student requirements.