A significant amount of effort and time is devoted to debugging in all phases of software development cycle. In fact, debugging is one of the sources for the software crisis the computer industry is facing today. To diagnose and fix an error, it is necessary to reproduce the error. Generally, reproduction of an error is not possible in real-time software systems because of the timing constraints involved and their non-deterministic characters. The unpredictable sequence in which the inputs arrive at a certain point in time will not exhibit the same behavior of the system upon repeated execution of the program. This project will develop a methodology for knowledge-based debugging of real-time software systems. A program execution monitoring method is first used to record a system execution history in a non-interfering manner--that is without corrupting the critical timing requirements of the target system. The traces, thus collected are analyzed to isolate the bug in an off-line mode. The examination of voluminous traces is a laborious and boring task that requires a high degree of expertise. This expertise is encapsulated into a knowledge-based system and the examination process automated to at least localize the fault and answer questions about its reasoning. Such an approach will definitely shorten the overall debugging time and contribute towards software productivity and reliability.
