Explicit Storage and Analysis of Billions of States using Commodity Computers

https://doi.org/10.3182/20121003-3-MX-4033.00058Get rights and content

Abstract

The objective of this paper is to develop a framework and associated algorithms for explicit state space exploration of discrete event systems that can scale to very large state spaces. We consider classes of resource allocation systems (RAS), where a set of resources are shared by concurrent processes. In particular, we focus on Gadara RAS, whose Petri net representations have recently been used for liveness enforcement in multithreaded software. We present a framework where each reachable state of the RAS is represented by a single bit. We show how single-bit representations can lead to efficient implementations of supervisory control algorithms. In order to support single-bit state representations, we develop two indexing functions that map each state to a unique integer that serves as the corresponding index of the state in the large bit array. These functions exploit the invariants of the given RAS. Experimental results show that our techniques scale up to exploration and analysis of billions of states on commodity computers.

Keywords

Resource Allocation Systems
Petri nets
Supervisory Control
State Space Exploration

Cited by (0)

This work was partially supported by NSF grant CCF-0819882 and an award from HP Labs Innovation Research Program.

View Abstract