Optimization techniques for parallel protocol implementation

We propose a method for deriving parallel, scheduling-optimized protocol implementations from sequential protocol specifications. We start with an SDL specification, identify a common path for optimization, and perform a data dependency analysis. The resulting common path graph is parallelized as far as permitted by the data dependency graph. The degree of parallelism is extended even further by deferring data operations to the exit nodes of the common path graph. The resulting parallel operation model is then submitted to a scheduling algorithm, yielding an optimized compile-time schedule. An IP-based protocol stack with TCP and FTP as upper layers serves as an example