A linear programming approach to general dataflow process network verification and dimensioning

TL;DR

This paper introduces a linear programming method to verify liveness and boundedness in dataflow process networks, providing efficient conditions and bounds for network safety and resource dimensioning.

Contribution

It presents novel linear programming-based sufficient conditions, including polynomial-time checks, for ensuring liveness, boundedness, and buffer size estimation in general dataflow process networks.

Findings

Linear programming conditions can verify network liveness and boundedness.

The approach provides safe upper bounds on channel buffer sizes.

Some conditions are polynomial-time computable.

Abstract

In this paper, we present linear programming-based sufficient conditions, some of them polynomial-time, to establish the liveness and memory boundedness of general dataflow process networks. Furthermore, this approach can be used to obtain safe upper bounds on the size of the channel buffers of such a network.