A distributed context aggregation scheme for personal smart spaces using aggregation task decomposition

Abstract

In context-aware computing, the context aggregation is an important part of the context management. It aggregates low-level contexts captured and supplied by mobile devices into high-level ones to perceive the situation of users. In an infrastructure-based smart space (i.e., immobile and battery-free), a centralized context management system needs not concern about its resource consumption for intensive context aggregation. However, in a personal smart space which consists of resource-constrained mobile devices, minimizing resource consumption of the mobile devices by minimizing the amount of context aggregation task is a very important issue. In this thesis, we propose a noble scheme of aggregation task decomposition in which heavy context aggregation tasks are decomposed and distributed to all the participating mobile devices as clients. The proposed idea would be useful for an application situation where mobile devices are context-intensive generating a large amount of frequent low-level contexts, such as biological, environmental and geographical information. By decomposing and distributing the heavy context aggregation operations to the client devices the processing overhead upon the coordinating device can be minimized without having the total amount of context processing significantly increased. The proposed task decomposition of context aggregation introduces a context aggregation coordinator and distributed local context aggregation operations among the participating devices. We implemented the proposed context aggregation scheme on participant devices which together form a personal smart space. We measured the efficiency of the proposed distributed aggregation scheme by task decomposition and thus verified its feasibility. This measurement shows that our scheme achieves to reduce the amount of energy usages and distribute the concentrated processing overhead on a coordinator.