A novel multi-stage attack dataset for smart home intrusion detection

Modern homes are increasingly relying on networks of internet-connected devices for smart home applications. Unlike an enterprise network, such smart home networks typically tend to have lax cybersecurity protocols in place and are prone to cyber-attacks. Attack patterns have also evolved over time, with multi-stage attacks being more difficult to detect and mitigate. This article presents a novel dataset captured under normal operation as well as from seven different multi-stage attack scenarios that can arise in a smart home environment that hosts end devices of heterogeneous nature. Unlike conventional network flow datasets which describe individual single-stage attacks on devices, this dataset aims to provide researchers with valuable insights into the behaviour of network parameters when each attack scenario includes a combination of attacks. This dataset with a total of 178,831 samples, is split into two parts: Training- consisting of 148,959 samples and Testing- consisting of 29,872 samples. The captured dataset can be used to develop intrusion detection systems capable of detecting and classifying such interleaved attack patterns that include multiple sequential attack steps in a smart home environment. The dataset captures the common network flow parameters in the network and hence, the dataset can be used to extend the research to other type of networks as well.