Constructive formal methods and protocol standardization

This research is part of the NWO project "Improving the Quality of Protocol Standards". In this project we have cooperated with industrial standardization committees that are developing protocol standards. Thus we have contributed to these international standards, and we have generated relevant research questions in the field of formal methods. The first part of this thesis is related to the ISO/IEEE 1073.2 standard, which addresses medical device communication. The protocols in this standard were developed from a couple of MSC scenarios that describe typical intended behavior. Upon synthesizing a protocol from such scenarios, interference between these scenarios may be introduced, which leads to undesired behaviors. This is called the realizability problem. To address the realizability problem, we have introduced a formal framework that is based on partial orders. In this way the problem that causes the interference can be clearly pointed out. We have provided a complete characterization of realizability criteria that can be used to determine whether interference problems are to be expected. Moreover, we have provided a new constructive approach to solve the undesired interference in practical situations. These techniques have been used to improve the protocol standard under consideration. The second part of this thesis is related to the IEEE 1394.1-2004 standard, which addresses High Performance Serial Bus Bridges. This is an extension of the IEEE 1394-1995 standard, also known as FireWire. The development of the distributed spanning tree algorithm turned out to be a serious problem. To address this problem, we have first developed and proposed a much simpler algorithm. We have also studied the algorithm proposed by the developers of the standard, namely by formally reconstructing a version of it, starting from the specification. Such a constructive approach to verification and analysis uses mathematical techniques, or formal methods, to reveal the essential mechanisms that play a role in the algorithm. We have shown the need for different levels of abstraction, and we have illustrated that the algorithm is in fact distributed at two levels. These techniques are usually applied manually, but we have also developed an approach to automate parts of it using state-of-the-art theorem provers.