This paper introduces an innovative approach to enhance the security infrastructure of Internet of Things (IoT) environments through the implementation of a decentralized intrusion-detection framework. The inherent challenges arising from the distributed and resource-constrained nature of IoT systems render traditional centralized intrusion-detection methodologies impractical, primarily due to scalability constraints and privacy vulnerabilities. In response, our work proposes a decentralized security architecture that employs federated machine learning techniques tailored for intrusion detection. Additionally, we integrate distributed ledger technologies to strengthen authentication and authorization mechanisms. The incorporation of edge computing addresses data privacy concerns, providing an effective resolution to security vulnerabilities in IoT systems. The prototype’s implementation was evaluated using a variety of machine learning models, including ANN, RNN, LSTM, and GRU, across two datasets, Bot-IoT and N-BaIoT. The empirical results indicate that DCD-FL achieves significant performance metrics, with an F1 Score of up to 98.20%, achieved with less than six aggregation rounds.