Self-Configuring Anomaly Detection IoT Architecture
Self-Configuring Internet of Things Architecture for Context-Aware Anomaly Detection Towards Rapid Deployments
Anomaly detection is the process of identifying unexpected items or events in data sets, which differ from the norm. It is a well-investigated area within research communities, and however, anomaly detection using IoT sensor data is comparatively unexplored. In order to develop, IoT sensor-based anomaly detection solution, engineers require significant technical knowledge (e.g., which algorithms to use, how to set parameters, etc.) and domain knowledge (e.g., agriculture, built environments, usual patterns within a given context, etc.). Recently, some commercial solutions (e.g., Microsoft Anomaly Detector) are being developed to simplify the development process by allowing engineers to use black-boxed anomaly detection algorithms with few configurable parameters (i.e., sensitivity, max window size, max anomaly ratio).
We believe that much more complicated contributing factors need to be considered when deploying anomaly detection systems. Further, even though we may know some of the contributing factors during design time, we may not know how to configure a system until we deploy the anomy detection system in a given context. For example, IoT devices have limited resources (e.g., energy, memory, computing resources) and may have shared responsibilities (i.e., not dedicated to anomaly detection). As a result, which devices would be available to perform anomaly detection may not be known beforehand. Further, the heterogeneity of IoT application scenarios makes it infeasible to find one generalised anomaly detection technique that works for every possible IoT architecture. Additionally, the could be competing requirements such as privacy vs performances that need to be managed. We believe that the best way to handle these challenges is to develop a self-configurable anomaly detection system that can configure the above-mentioned configurable parameters at runtime and adapt to the given context. We propose FedBio-IoT, a federated self-configuring IoT architecture for context-aware anomaly detection in this project. FedBio-IoT is based on nature-inspired algorithms that use the concept of evolutionary algorithms and swarm intelligence to monitor, configure, adapt, and change the federated IoT architecture according to the population’s behaviour and biological evolution from one generation to the next. We aim to investigate how to reduce the technical and domain expertise engineers require and reduce trial-and-error guesswork required during the development stage. This project is composed of several objectives:
- Conduct a literature review on anomaly-detection techniques, their characteristics and configurable properties.
- Study the capabilities of a wide range of swarm-intelligence algorithms that can be used in self-configuring IoT architecture and examine their strengths and weaknesses.
- Evaluate the performance of self-configuring IoT architecture for context-aware anomaly detection based on swarm intelligence through experimental evaluations in different IoT application scenarios.
Exalens protects digital manufacturing against downtime and safety incidents through early warning of both system malfunctions and cyber security breaches. With ground-breaking cyber-physical security analyst AI, manufacturers enhance their operational resilience with automated incident detection and response.
PETRAS National Centre of Excellence for IoT Systems Cybersecurity is a consortium of eleven
leading UK universities which will work together over the next three years to explore
issues in privacy, ethics, trust, reliability, acceptability, and security.
Government Communications Headquarters, commonly known as GCHQ, is an intelligence and security organisation responsible for providing signals intelligence and information assurance to the government and armed forces of the United Kingdom.