The System-Level Design Methodology For Smart Space

With the development of pervasive computing, smart space has been widely applied in various domains. Smart space is the high fusion of cyberspace, physical space and social space. It refers to the multi-discipline technologies such as computing, communication, human computer interaction, security, etc. The design of smart space is facing grand challenges due to integrating these technologies. System-level design methodology can transform the high-level design specification into underlying architecture implementations leveraging refining step by step. It provides a novel idea to the design of smart space. This thesis focuses on multiple design issues of smart space including object emplacement, architecture platform design, human computer interaction design and security design, and uses system-level design methodology to examine description, modeling and optimization of design. The main contributions of this thesis are summarized as follows:1) The system-level design framework is proposed. It is a top-down design and can refine the design specification of smart space into underlying architecture implementations. The procedure of design can be seen as searching and synthesizing in given architecture platform library and given constraints.2) An intermediate representation model is presented. It can transform the specification of smart space into intermediate model to be used for the system-level performance estimation. The model describes the control flow of the smart space object based on timed Petri net, meanwhile corresponding physical flow, data flow and human flow are used to indicate the object interactions of cyberspace and physical space, cyberspace internal, and cyberspace and social space.3) The thesis proposes to partition the design flow of smart space into three parts: object emplacement, system synthesis and preference synthesis, where object emplacement aims to nearly optimal assignment for the objects defined in the high-level design specification to actual physical coordinates, system synthesis and preference synthesis are performed to nearly optimal allocation for computing and communication resources and human-computer interface. Then a home health care case is used to validate the feasibility and effectiveness of our proposed approach.4) An improved design framework is proposed to tailor the design of smart space for various users under social scenario. Specifically, a hierarchical Petri net based model is presented to improve the control flow in intermediate representation model, meanwhile the social flow is added into the model to indicate social interactions of multiple users. A smart office case is used to demonstrate the effectiveness of the improved design framework.5) Regarding to the security design, the thesis extends the improved design framework and a general security model is proposed to protect the messages communication of smart space task from the attack by unauthorized users, and also it considers lowering the energy consumption leveraging the dynamic voltage and frequency scaling technologies. Moreover, to mitigate the negative impact caused by DVFS, reliability constraint of system is taken into design consideration. Based on this, the thesis formalizes the security-critical smart space design as a triple-objective combinatorial optimization issue to be solved. Finally a case study is used to verify the efficiency of the proposed optimization approach.