Implementing Access Control Markov Decision Processes with GLPK/GMPL

TL;DR

This paper demonstrates how to implement access control Markov Decision Processes using the open-source solver GLPK and GMPL language, enabling decision-making and value calculation for access control policies.

Contribution

It introduces an implementation approach for access control MDPs with GLPK/GMPL, including modeling, decision-making, and value computation, with illustrative examples.

Findings

The approach successfully models access control decisions as MDPs.

Variation of parameters affects the final access control decisions.

The method computes decision values alongside decisions.

Abstract

In a recent approach, we proposed to model an access control mechanism as a Markov Decision Process, thus claiming that in order to make an access control decision, one can use well-defined mechanisms from decision theory. We present in this paper an implementation of such mechanism, using the open-source solver GLPK, and we model the problem in the GMPL language. We illustrate our approach with a simple, yet expressive example, and we show how the variation of some parameters can change the final outcome. In particular, we show that in addition to returning a decision, we can also calculate the value of each decision.