A Statistical Approach to Model-Based Robustness Testing

Recent massive research in the area of robustness testing of communication protocols have been driven by the fact that embedded systems are becoming more and more complex every day. Providing correct implementations of communication protocols is a key to successful interconnection of embedded system constituents. Test and verification is the last and the most critical phase in the production of the corresponding software artifacts. This paper contributes to the overall research effort in the area by proposing an original approach to model-based robustness testing of communication protocols. In the paper we introduce the notion of hidden FSM states and state transitions. We also show how to calculate the probabilities of detecting the existence of the hidden state transitions. Then we present our methodology of model-based robustness testing of communication protocols. The methodology is based on the special operational profiles, which are referred to as the stress operational profiles.

1. Introduction

The communication protocol engineering [1] is an engineering discipline that deals with the engineering of communication protocols. Although frequently addressed in the literature, the methodology that governs the production process of the corresponding software artifacts remains work in progress. Particularly, the test & verification phase of the process is still the Achilles heel of the communication protocol engineering. UML 2.0 has improved some of these short-comes, but there is still room for further improvements in that area. For example, there are a number of formal verification methods (e.g. [3]) and statistical usage testing methods [4] –[7] at our disposal, which may be used to improve the current practice.