Synthesizing Locally Symmetric Parameterized Protocols from Temporal Specifications

Scalable protocols and web services are typically parameterized: that is, each instance of the system is formed by linking together isomorphic copies of a representative process.Verification of such systems is difficult due to state explosion for large instances and the undecidability of verifying properties over all instances at once.This work turns instead to the derivation of a parameterized protocol from its specification.We exploit a reduction theorem showing that it suffices to construct a representative process P that meets a local specification under interference by neighboring copies of P .Every instance of the parameterized protocol is built by deploying replicated instances of P .While the reduction from the original to a local specification is done by hand, the construction of P is fully automated.This is a new and challenging synthesis question, as one must synthesize an unknown process P while simultaneously considering interference by copies of this unknown process.We present two algorithms: an eager reduction to the synthesis of a transformed specification, and a lazy, iterative, tableau construction which incorporates fresh interference at each step.The tableau method has worst-case complexity that is exponential in the length of the local specification.We have implemented the tableau construction and show that it is capable of synthesizing parameterized protocols for mutual exclusion, leader election, and dining philosophers.