Setting break-points in distributed time-triggered architecture

Abstract

Setting a brake-point in a local node of a distributed real-time system freezes the execution of the application in the local node, while other nodes in the network continue execution of their part of a distributed application. This will lead to an inconsistent state of nodes in the system, because the physical time cannot be stopped and the system is no more consistent in the temporal domain. We present a new method for setting brake-points in distributed time-triggered systems. After a brake-point is set all nodes in the network freeze the execution of the local application at the same instant and enters a debug operation mode. This method can be applied on target systems, which use the concept of sparse time base, like the time-triggered architecture (TTA). During the debug mode a monitoring (debugger) node is used to exchange debugging commands and data with particular nodes in the network. The debugging process is transparent to distributed time-triggered applications. The proposed debugging technique can be applied also for software fault injection. It circumvents the problem of temporal intrusiveness during the debugging or software fault injection in time-triggered systems.