Log Analysis of Cyber Physical Systems

Dealing with Cyber Physical Systems, this project takes two directions:
First, system engineers at Auxon (our industrial partner) are interested in understanding the reason(s) why the system fails and, more precisely, why its behavior deviates from previously observed one(s), showing anomalies. Furthermore, to help Auxon diagnose the problem, we also provide them with root cause(s) of the anomalous behaviors of the system. We are then interested in i) developing an effective and scalable sequential anomaly detection procedure and ii) designing an effective and scalable anomaly diagnosis approach.
Second, Auxon is interested in knowing the exact conditions under which a complex component (e.g., Machine Learning Component MLC) contributes to system failure(s). We then develop an effective and efficient automated approach to identify such conditions, considering the black-box nature of MLCs and addressing the interactions among the components of the system, which might cause system failure(s) regardless of whether component(s) have failed.

Faculty Supervisor:

Lionel Briand

Student:

Partner:

Auxon Technologies Inc.

Discipline:

Computer science

Sector:

Automotive; Mining; Technology

University:

University of Ottawa

Program: