A Novel Approach Based on Federated Learning for the Identification of Man in the Middle Attacks in IOT Networks Using Blockchain

Federated Learning (FL) enables cooperative model training across dispersed edge devices while protecting data privacy and providing localized insights without the need for centralized data aggregation. In the Internet of Things (IoT), federated learning enables cooperative model training among dispersed edge devices while protecting data privacy and providing localized insights without the need for centralized data aggregation. Nevertheless, Federated Learning’s local model sharing approach makes it susceptible to Man-in-the-Middle (MITM) attacks. In FL, attackers have the ability to manipulate local models. As a result, a global model produced from the altered local models may be inaccurate. In this paper, we suggest a blockchain-based FL architecture to prevent intermediaries from readily altering the FL parameters throughout the transmission process. All of the clients' parameters are combined by a cloud server, which is acting as the federated parameter server. By integrating blockchain technology into the overall architecture, we link all cloud and edge servers. This paper uses a PoC algorithm based on the SHA-256 hashing function in order to validate the data. The results and comparison analysis show that the suggested framework has a low false-positive rate and a high accuracy in detecting MITM attacks in their early stages, with a detection rate ranging from 98 to 100%.