VAIDS: A Hybrid Deep Learning Model to Detect Intrusions in MQTT Protocol Enabled Networks

The adoption of 5G networks is seen as an enabler for IoT. With low latency and increased reach, many more IoT devices can now be connected and controlled. An increase in the number of IoT devices since its inception brought forth the need for a lightweight communication protocol. The protocol in question should be able to cater to a large number of IoT devices. These requirements were the basis of the MQTT protocol. Attackers can utilize the protocol to target a network. Furthermore it is a herculean task to manually identify an attack in a huge network. Artificial Intelligence can be used to efficiently detect such attacks with a high degree of accuracy. In this research we propose a hybrid deep learning mult-iclass classification model VAIDS. VAIDS utilizes the CNN algorithm to extract features from the dataset. These features are then utilized as inputs for the LSTM algorithm. The proposed model can detect five types of anomalies within an IoT network that uses the MQTT protocol. The proposed model is trained, tested and validated against the MQTT-IoT-IDS2020 dataset and classifies a given input into one of five attack classes. The model showcases a high degree of accuracy of 99.97%.