A lightweight encrypted packet priority classification framework for Cognitive radio enabled Industrial IoT Networks

Industrial IoT (IIoT) has been adopted due to its significant benefits in industrial operations. However, the rapid proliferation of IIoT devices raises concern about the spectrum shortage. Cognitive radio (CR) technology has been proposed as a potential solution to meet the increasing spectrum demand. CR enables IIoT devices to use the licensed spectrum bands opportunistically. Due this exponential growth of IIoT devices, it has become nearly impossible to accommodate all spectrum demands from IIoT devices. To optimize resource management, it is important to classify the packets into different classes to ensure that at least critical packets can be accommodated within the available spectrum. However, categorizing the encrypted packets based on their information priority is complex. To address this challenge, we introduce a lightweight Packet Prioritization Framework (PPF) for CR enabled IIoT (CR-IIoT) networks. The PPF uses convolutional neural networks (CNNs) to extract complex spatial features from encrypted packets. We acknowledge that packets from the same IIOT device may have varying priorities depending on the sensed data. The model’s effectiveness is verified using the K-fold cross-validation technique. To ensure the model deployable on IIoT devices, we applied model compression techniques to make PPF lightweight. We created a custom dataset in a simulated IIoT network to evaluate the performance of PPF. Further, we conducted an exhaustive simulation to compare PPF with existing models to validate the accuracy, efficiency, and resource utilization of our model.