Computer Science Technical Report Swarm Synthesis of Convergence for Symmetric Protocols

Due to their increasing complexity, today’s distributed systems are subject to a variety of transient faults (e.g., loss of coordination, soft errors, bad initialization), thereby making self-stabilization a highly important property of such systems. However, designing Self-Stabilizing (SS) network protocols is a complex task in part because a SS protocol should recover to a set of legitimate states from any state in its state space; i.e., convergence. Once converged, a SS protocol should remain in its set of legitimate states as long as no faults occur; i.e., closure. The verification of SS protocols is even harder as developers have to prove the interference-freedom of closure and convergence. To facilitate the design and verification of SS protocols, previous work proposes techniques that take a non-stabilizing protocol and automatically add convergence while guaranteeing interference-freedom. Nonetheless, such algorithmic methods must search an exponential space of candidate sequences of transitions that could be included in a SS protocol. This paper presents a novel method for exploiting the computational resources of computer clusters and search diversification towards increasing the chance of success in automated design of finite-state selfstabilizing symmetric protocols. We have implemented our approach in a software tool that enables an embarrassingly parallel platform for the addition of convergence. Our tool has automatically synthesized several SS protocols that cannot be designed by extant automated techniques.