Research On Software Architecture Of Wearable Computer

Wearable Computer is a class of new conceptual personal mobile computing system,which spring up and develop rapidly recently. Wearable Computer's emergence is due to the fact that computer has increasingly become super-micromation, and is the inevitable outcome of the concept-"computer should fit humans". Wearable Computer will lead to a great change of human-computer relationship and will put forward new issues and challenge in computer science and technology field. Traditional paradigm of desk computing and mobile computing cannot support the characteristics and functions of Wearable Computer properly. It is of great necessary to explore a brand-new conceptual operation paradigm and software architecture for Wearable Computer. The research of software architecture of Wearable Computer is significant. In this paper, we have designed a service based software architecture model of Wearable Computer and put forward our own innovation. The main works and conclusions are shown as follows:(1) We analyze the special functions and requirements of Wearable Computer software detailedly, also summarize the difficulties and main matters of current research on software of Wearable Computer.(2) We amply investigate several existing software architectures of Wearable Computer and analyze their advantages and disadvantages.(3) We design a service based software architecture model for Wearable Computer with dynamic, adaptive and open properties. This software architecture supports dynamic load and export of service, context-aware interaction, customization and security & privacy protection etc.(4) We provide the architecture and flow of dynamic load of service, and design the CERP replacement algorithm for service cache. The result of experiment shows CERP is a good algorithm which can evidently promote the general efficiency and performance of service cache of Wearable Computer.(5) We also provide a prefetch algorithm for service cache, which is based on prediction model and prefetch threshold. We provide two prediction models: LINK model and Matrix model(SCM+SPM).