Obtaining Highly Dependable Communication Protocols

University Of Southern California, Los Angeles CA

Investigators

Abstract

Communication protocols form an integral part of many complex software systems. In many such systems, protocols' impact on system dependability far exceeds the fraction of the total number of lines of code devoted to their implementation. Unfortunately, there exists virtually no systematic approach for design and evaluation of protocols that considers dependability. Previously, we have developed a systematic, semi-automatic approach for validation of communication protocols. Our approach uses semi-formal representations of protocols and efficient search (forward search as well as fault-oriented test generation) to analyze protocols over the entire range of scenarios, including external events, faults (such as loss of packets, communication link failures, loss of state in communication nodes, and so on), as well as network topologies. The effectiveness of our approach has been demonstrated via its application to a varied range of protocols. In this project, we will undertake three main tasks. First, we will extend our previously developed framework to compute values of appropriate dependability metrics for existing protocols of interest to the NASA and the HDCCSR community. Second, we will develop the first systematic approach to design new protocols that satisfy desired dependability criteria. Finally, we will use the testbed provided by NASA in conjunction with the expertise and software made available by NASA and other HDCCSR researchers to fine-tune our approaches and to demonstrate that: (i) accurate values of dependability metrics can indeed be computed for existing communication protocols using a systematic, semi- automatic approach, (ii) new protocols can be designed in a systematic, semi-automated manner to meet desired dependability criteria, and (iii) enhancements in the dependability of communication protocols can indeed significantly enhance the dependabilities of many software-hardware systems of interest to NASA and the HDCCSR community. In addition to above technical contributions, the proposed research will have broader impact on society. First, results of the proposed research will enable improvements in dependability of critical infrastructure, such as the air-traffic control system. Second, the research will develop new approaches and practical case studies that will provide us with a unique opportunity to educate a large and diverse student body in techniques for development of dependable protocols and systems.