An OMNeT++-Based Approach to Narrowband-IoT Traffic Generation for Machine Learning-Based Anomaly Detection

The importance of security in smart city IoT applications has continued to grow in recent years, especially when critical infrastructure is involved. State-of-the-art deep intrusion detection systems (Deep IDS) help distinguish normal traffic from traffic originating from potential attackers. In this paper, we aim to describe and evaluate a pipeline to simulate a smart city NB-IoT network, generate traffic and subsequently build from it a synthetic dataset using the OMNeT++ simulator. This dataset can then be used to train different ML-algorithms for anomaly detection in deep IDS. The main goal of the present paper is to showcase a proof of concept, examples are kept simple with the possibility of a more complex application at a later point. The research forms the basis for the development of an efficient Deep IDS to be integrated into an urban IoT network in the form of a middlebox. While previous research has relied on specific use cases and mostly on computer architectures with large cpu clusters and memory capabilities, the approach proposed by us offers a simple and straight forward way to generate synthetic traffic that is detailed and closely modelled to the respective use case as well as it can be created quickly and with minimal resources, e.g. a standard laptop.