Linear inequality LTL (iLTL): A model checker for discrete time Markov chains

Abstract

We develop a way of analyzing the behavior of systems modeled using Discrete Time Markov Chains (DTMC). Specifically, we define iLTL, an LTL with linear inequalities on the pmf vectors as atomic propositions. iLTL allows us to express not only properties such as the expected number of jobs or the expected energy consumption of a protocol during a time interval, but also inequalities over such values. We present an algorithm for model checking properties of DTMCs expressed in iLTL. Our model checker differs from existing probabilistic ones in that the latter do not check properties of the transitions on the probability mass function (pmf) itself. Thus, iLTLChecker can check, given an interval estimate of current pmf, whether future pmfs will always satisfy a specification. We believe such properties often arise in distributed systems and networks and may, in particular, be useful in specifying requirements for routing or load balancing protocols. Our algorithm has been implemented in a tool called iLTLChecker and we illustrate the use of the tool by means of some examples. © Springer-Verlag Berlin Heidelberg 2004.