A space model for user-centered building services

Recent advances in computing have the potential to enable user-centered building services (UCBS) in the areas of building automation and ubiquitous computing which are responsive to the comfort or information needs of individual users. For example, agents are conceivable that manage local indoor climate and lighting quality based on individual users` location, activity, and corresponding preferences. This project examines a common characteristic of UCBS, which is a need for detailed and up-to- date information about the configuration of architectural spaces in order to deliver a high quality of service. A dedicated space model is proposed in which to consolidate such information. By contrast, in current building automation systems, space configuration information is often poorly organized, difficult to access and maintain. The primary project goals are the specification, design, prototypical implementation, and validation of a space model for UCBS. The project builds on and extends existing work in the areas of building information models (BIM) and location models. The proposed space model includes a higher level of detail than location models. Moreover, multiple views and representations of space configurations are provided. They enable query operations that let UCBS clients retrieve small subsets from a space model instance based on view, geometric and topological parameters. Queries are the most frequent operations and, in domains such as lighting, may require short response times. These are different requirements compared with design, where BIM tend to be used. A space model schema is implemented in a spatially enabled database environment. Equally important, a set of update operations are developed to automate repetitive update tasks. Geometric reasoning methods are devised and existing geometric modeling libraries reused to implement these operations. The space model database and query operations are modules of a space model server. Its main non-functional design objective is responsiveness under peak query loads that are to be expected in large public or office buildings. Tests and demonstrations of prototypical implementations of the space model structure and operations, the space model server, and an instance of a user- centered lighting service deployed in an office space will provide initial feedback on the feasibility and potential utility of space models for UCBS.

This research project has achieved three main advances in the use of space models towards realizing personalized (user-centered) building services. The project is based on the assumption that such services, if enabled by fine-grained spatial and other context data, may better meet comfort and information needs of individual building users than conventional building services.First, we have developed a space model that meets common spatial context query requirements of personalized building services. An example for such a query is the retrieval of the space in which a user is located together with nearby objects. In our space model, spaces and space elements (windows, doors, furniture, and so on), as well as spatial relations are modeled as a geometric network that may be traversed efficiently by graph algorithms. We have defined a novel schema (data structures) and a set of operations to automatically generate domain-specific space views (e.g. for lighting, heating, ventilation, and air conditioning, or indoor path planning) from unstructured space data that are created in a BIM authoring system. We have designed and implemented a system prototype and validated the space model's data structures and operations with case studies of two existing office buildings.Second, we have developed a space model server that is based on the space model. The server provides an interface for concurrent spatial context queries by personalized building service clients.Fundamental query operations are based on the graph traversal paradigm. They include breadth-first, depth-first, and shortest path search. We have reused and extended algorithms from an existing graph library to process query operations. We have defined, implemented, and tested operations that form the kernel of a spatial context query language.Third, in order to validate the concept of personalized building services enabled by a space model, we have developed a personalized lighting control (PLC) system that queries the space model to customize the lighting environment according to a user's spatial context, activity, and lighting preferences. We have developed a conflict management algorithm which queries the space model to automatically detect and resolve lighting conflicts when multiple users are present in a space.We have deployed prototypes of the space model, the space model server, and the PLC system in a medium-size, multi-person office space. We have conducted an experiment with the system prototype that involved 3 virtual users checking in and out of workplaces in the test office in different sequences. Compared with full lighting, we observed electricity savings between 45% (three person occupancy) and 70% (single person occupancy) due to targeted dimming of individual luminaires.