Mid Sweden University

This paper deals with the problems of maintaining a long lifetime embedded system, including obsolescence, function change requirement or technology migration etc. The aim of the presented work is to analyze the maintainability of long lifetime embedded systems for different design technologies. FPGA platform solutions are proposed in order to ease the system maintenance. Different platform cases are evaluated by analyzing the essence of each case and the consequences of different risk scenarios during system maintenance. Finally, the conclusion is drawn that the FPGA platform with vendor and device independent soft IP is the best choice.

The reuse of predefined Intellectual Property (IP) can lead to great success in system design and help the designer to meet time-to-market requirements. A soft IP usually needs some customization and integration efforts rather than plug-and-play. Communication interface mismatch is one of the problems that integrators often meet. This paper suggests a soft-IP interface modification methodology (SIPIMM) for systems on Field Programmable Gate Array (FPGA) SIPIMM targets an interface-based soft IP model which is introduced to ease the interface modification and interface reuse. A case study of an open-source IP is presented using SIPIMM for system integration.

The reuse of predefined Intellectual Property (IP) can shorten development times and help the designer to meet time-to-market requirements for embedded systems. Using FPGA IP in a proper way can also mitigate the component obsolescence problem. System migration between devices is unavoidable, especially for long lifetime embedded systems, so IP portability becomes an important issue for system maintenance. This paper presents a case study analyzing the portability of an FPGA-based M-JPEG decoder IP. The lack of any clear separation between computation and communication is shown to limit the decoder's portability with respect to different communication interfaces. Technology and tool dependent firmware IP components are often supplied by FPGA vendors. It is possible for these firm IP components to reduce development time. However, the use of these technology and tool dependent firmware specifications within the M-JPEG decoder is shown to limit the decoder's portability with respect to development tools and FPGA vendors.

This paper discusses the component obsolescence problem and presents a mathematic model for life cycle analysis of long life cycle embedded system maintenance. This model can estimate minimized management costs for different system architecture. Matlab is used to generate a graph and Lingo is used for linear programming. A simple CAN controller system case study is shown to apply this model. A minimized management cost and an optimized management time schedule are given as the result. The responses from the experiments of the model meet our expectation. Although the model has lots of simplifications and limitations, it can give management strategy guidance to the designers who suffer from component obsolescence problems.