A Distributed Cybersecurity Solution in an IoMT Network Using a Multi-target Federated Learning

Abstract

Most cybersecurity solutions are based on building a single logical entity to tackle one or multiple cyber threats. Such solutions have proven effective but can be less advantageous in a situation where cyber threats are so many, such that building a detection mechanism will imply writing a lot of logical expressions. Such solutions can be less effective on a specific threat and resource exhaustive on multiple threats. This can be risky to implement in delicate environments such as the medical IoT environment, where an optimal threat detection precision is required. In this paper, we introduce a distributed cybersecurity solution where the logical entities (agents) are distributed in federation along a threat profile, and their individual security operations are merged to produce a globalized security solution over a larger threat profile. We did experiment on this proposed solution and compare the results with those of conventional cybersecurity solution.