Dependence Based Model-Healing

Abstract

With the intensifying requirements of intelligent software, self-healing system becomes increasingly imperative. Such systems attempt to heal themselves in the sense of recovering from faults. The system is usually represented at a higher-level of abstraction than a program, such as Extended Finite State Machines (EFSMs), in which a fault could be regarded as a restricted operational environment for the EFSM. In general, the parts of model affected or potential affected by the environment changes are simply removed, which may lead the loss of system functions, and even system sinking or breakdown. In this paper, we propose a model-healing technique with the ability of retaining model functions. Dependence analysis and model-based repair rules are used to reserve model functions as many as possible to strengthen the model's capacity of self-adaptive in a restricted environment. A case study is conducted to illustrate that the functions affected in a restricted environment can be healed to some extent.