IoT Service Recommendation for Multi-resident Smart Homes

Abstract

The Internet of Things (IoT) is revolutionizing industries by connecting everyday items. This research aims to develop an innovative IoT service recommendation framework for multi-resident smart homes. Three key challenges are addressed: conflict detection, preference estimation, and conflict resolution, all within a multi-user, multi-application smart home context. The research comprises a literature review, the design of an IoT-based smart home setup, and conflict detection and resolution frameworks. The smart home system includes occupancy estimation using thermal images, resident identification using positional encoding and LSTM models, and habit extraction using unsupervised clustering and pattern-mining techniques. The conflict detection framework introduces a new IoT conflict model, conflict ontology, and a hybrid conflict detection algorithm, along with a fine-grained conflict detection approach using temporal proximity and entropy. The impact conflict detection framework utilizes a preference estimation model and temporal and preferential proximity techniques. Conflict resolution strategies include an adaptive priority-based framework using the analytic hierarchy process and a dynamic framework using a matrix factorization-based approach. Evaluation on real-world datasets demonstrates the effectiveness of the proposed approaches, with the ultimate aim of advancing the deployment of IoT-based service recommendations in smart homes.