Tuning Development of Distributed Real-Time Systems with SDL and MSC -- Current Experience and Future Issues

SDL provides powerful capabilities for verification1 and validation2 of a system's behaviour and for automated code generation. This allows to perform system validation at a higher level of abstraction and earlier in the development life cycle. However, one needs to be carefully to really gain advantage of such capabilities, especially when applying SDL to a broader class of applications which may be called "decision-making, distributed systems". Firstly, state explosion may prevent to get any benefit from exhaustive simulation or much effort is required to limit the number of states thereby loosing most of the advantages of automated testing. Secondly, the current means of SDL and of SDL tools may not be sufficient to identify all bugs of a system's specification and design. Even when exhaustive simulation does not report any error, the system may not run correctly on the target, or vice versa, the optimum practical solution may be rejected as erroneous. This paper will analyse the situation, provide with a solution for tuning of system development which is based on an additional layer, called EaSySimII, on top of the ObjectGEODE tool, and will identify future issues. 1