by Tracy Zafian, Research Fellow
A research team led by Dr. Lance Fiondella, UMTC Research Affiliate, and professor at UMass Dartmouth, studies software reliability related to various transportation systems; and has developed a quantitative approach to assess the dynamic vulnerability of transportation networks.
The role of computers in operating, optimizing, and securing transportation networks and systems has grown tremendously in the last few decades. Onboard computers help run, sometimes autonomously, individual vehicles for land, sea, and air. On a larger scale, computers assist with the monitoring of traffic and transportation infrastructure to help transportation networks operate safely and efficiently.
Unfortunately, the computers performing such tasks do not always run as intended. For example, in September 2017, the New York Times described how air travel was temporarily delayed at airports on at least four continents when airline software that manages customer reservations and check-ins for close to 200 airlines worldwide experienced network glitches. In another example, in February 2018, traffic signal lights for an estimated 600 intersections in New York City failed to work properly after a “routine software upgrade” was carried out overnight. CBS news reported that some signal lights became flashing red lights and others went completely dark. Drivers, pedestrians, and others traveling through those intersections needed to exercise extra caution until the impacted traffic lights were fixed; over 99% of the lights were fixed within 18 hours.
Dr. Lance Fiondella, UMTC Research Affiliate and a professor at UMass Dartmouth in the Department of Electrical and Computer Engineering, conducts research on software reliability engineering, and transportation. The American National Standards Institute defines software reliability as the probability of failure-free operation for a certain amount of time in a certain environment. Unlike hardware, software does not fail due to physical flaws and wear, but due to design flaws, which as described in the Handbook of Software Reliability Engineering (Michael Lyu, editor), can be harder than hardware flaws to visualize, detect, and correct.
Dr. Fiondella and his Ph.D. student Vidhyashree Nagaraju authored a chapter in the recently published Handbook of RAMS in Railway Systems: Theory and Practice. RAMS stands for Reliability, Availability, Maintainability, and Safety, and Fiondella and their chapter focused on software reliability in RAMS management. As the chapter describes, “In the context of railway systems, software reliability is important in critical applications such as dynamic control of safe separation between trains in railway signaling, railway interlocking systems, monitoring and real-time control software, and hardware control software.” The chapter presents different software reliability models, discussing their mathematical formulation, the underlying assumptions, and procedures to fit these models to failure data obtained during testing with examples from the research literature. The chapter also provides a web link to an open source software failure and reliability tool created at Fiondella’s lab that implements many of the concepts discussed in the chapter.
Dr. Fiondella and his team have also developed a quantitative approach to assess the dynamic vulnerability of transportation networks. The approach uses methods from traffic simulation and incorporates traffic demand and congestion changes as a function of time, including the peaks in traffic volumes by time of day and for community activities such as large sporting events or other gatherings. Fiondella and his students, including lead author Ph.D. student Venkateswaran Shekar, presented a research paper describing this approach at the 2017 IEEE (Institute of Electrical and Electronics Engineers) Symposium on Technologies for Homeland Security, along with a series of examples. The paper stated that “this approach can quantify the time-varying criticality of [network] links, which can inform network defense and resilience planning. Because pervasive deployment of defenses is prohibitively expensive, identifying how the vulnerability of links changes over time will provide greater insight, enabling quantitative assessment of competing for defense strategies to preserve continuity of travel time reliability within a transportation network despite disruptions.”
Fiondella is currently part of a MassDOT sponsored project, multi-campus (Lowell, Dartmouth, and Amherst) UMass research team studying the use of unmanned aerial systems (UASs, commonly referred to as drones) for surface transportation applications. This study is being sponsored by MassDOT. Fiondella’s project task includes researchers from UMass Dartmouth and UMass Lowell and focuses on the security of UAS systems and data. As described in the study proposal, “there is an urgent need to ensure that UASs are engineered with sufficient security to withstand and recover from inevitable cyber-attacks.” The researchers are conducting a detailed literature search and synthesis on state-of-the-art secure UAS engineering techniques including compliance with national and international standards. The results from this work and the other project tasks will be used to help develop a pilot program for using UASs at MassDOT.
Fiondella recently received a prestigious National Science Foundation CAREER Award and multi-year grant for his research on Software Reliability and Security Assessment: Modeling and Algorithms.