Private LAN Architecture : Reflective Architecture for Home LAN Systems

Consumer electronics such as TV, VCR, Camcoder and STB, have been connected to networks, and home local area network (LAN) systems have been organized. As shown in Figure 1, a home LAN system consists of various I/O devices, thus, new application systems, e.g. the music downloading service via internet, which were not supported by existing distributed systems with only PCs have been proposed and get used on business. In future, home LAN systems will get more useful and convenient at home. For efficient execution of home LAN systems, system characteristics according to home should be considered well. One of system characteristics is the amount of computational resources, such as memory space, CPU speed and disk size installed on the devices. For example, an internet-TV with the small memory size and low CPU speed can only execute application programs which require the small amount of resources. If computational resources belonging a home server with the large memory space and a powerful CPU can be shared with such an internet-TV, a large-scaled application programs can be executed as if they are executed on the TV. User requirements are also system characteristics according to home. For example, a user who uses a VCR frequently wants to give higher priority to procedures for communication from/to the VCR than those between other devices. A user using a mobile device wants the system to load updated device driver programs for the mobile device while it is disconnected from the home LAN. Let us consider a middleware system which provides the programming framework for optimizing application program execution according to system characteristics for home LAN systems. Optimizing procedures based on system characteristics are independent on basic algorithms of application programs, because it is impossible to take optimization into account by estimating all system characteristics before those are executed. From the viewpoint of programming, the middleware needs to support the framework which enables optimization procedures to be separated from basic algorithms. Separation of concerns is one of the most important aspects provided by meta-level architectures, which make optimization code separated from basic functional code [1]. Thus, the middleware for home LAN systems should have meta-level architectures. Papers such as [2] describe benefits of reflective middleware systems. The middleware for home LAN systems should support the programming framework where the basic functional algorithms are described as base-level programs while optimization algorithms are described by meta-level programs according to computational …