LALP: Experimental Analysis of an Intrusion Detection System over Wireless Sensor Network using Learning Assisted Logical Protocols

Intrusion Detection Systems (IDS) play a crucial role in safeguarding networks against malicious activities. This study proposes an innovative approach LALP to intrusion detection over Wireless Sensor Networks (WSN) using a hybrid artificial neural network (ANN) architecture integrated with GoogleNet, a state-of-the-art convolutional neural network (CNN). The model is trained and evaluated on the CICIDS 2017 dataset, sourced from Kaggle, which provides a diverse collection of network traffic data encompassing both benign and malicious activities. The study encompasses various phases including data preprocessing, data sampling, feature extraction, classification, and parameter tuning, all orchestrated within a Python environment using Jupyter Notebooks. Preprocessing involves handling missing values, normalization, and categorical encoding, followed by data sampling techniques such as SMOTE and random undersampling to address class imbalance. Feature extraction techniques are employed to capture informative features from raw network traffic data, and the hybrid ANN -GoogleN et architecture is utilized for learning complex patterns and hierarchical representations. Training utilizes backpropagation for parameter updates, and parameter tuning is conducted using techniques like grid search or Bayesian optimization.