Toward adaptable distributed real-time computer systems

A distributed real-time computer system consists of several processing nodes which are interconnected by communication channel. For a safety critical application the real-time computer system should maintains timely and dependable services despite component failures or environmental changes. In this paper we present a framework of an adaptable distributed real-time computer system which can adjusts its operating strategies in response to changing application environments or variation in the fault patterns. The proposed adaptable real-time system (ARTS) is based on the integration of the scheduling strategy using imprecise technique (IT) with the fault tolerance mechanism. In our simulation study the performance of an ARTS of a controller area network (CAN) based distributed system is evaluated as to whether the timing constraints of the automotive benchmark signals can be satisfied for normal, overload, and degraded operational modes. Our investigation also shows that the quality of service maintained by the ARTS can be further improve by varying the workload