Dijkstra’s concept of self-stabilization assumes that faults can only affect the variables of a program. We study the notion of selfstabilization if faults can also affect (i.e., augment) the program code of a system. A code stabilizing system automatically recovers from (almost) arbitrary perturbations of its program code. We prove some lower bounds for code stabilizing systems and argue that code stabilization has many resemblances to the area of integrity management in the domain of security.
Felix C. Freiling, Sukumar Ghosh