Skill-Based Multi-Agent Systems for Building Control Using KNX IoT and Web of Things

Modern building management systems (BMS) require scalable and flexible solutions to integrate new devices and functionalities with minimal manual effort. This paper addresses this challenge by proposing a multi-agent system (MAS) architecture that leverages the KNX IoT and Web of Things (WoT) specification. The proposed skill-based MAS approach enables agents to interact autonomously through a semantically enriched communication layer. This reduces configuration efforts and commissioning, which typically pose a great challenge for the deployment and integration of new devices within a BMS. The MAS is virtually deployed and evaluated for a single-family residence. Two scenarios are analyzed: (1) Addition of new equipment to an existing setup, and (2) extending functionality of an existing setup. The results demonstrate the seamless device integration process and validate the applicability of KNX IoT and WoT for the developed MAS. Nevertheless, the study also identified limitations in current KNX modeling capabilities, that could be improved in future versions.