Abstract

Understanding the behavioural aspects of software systems can help in a variety of software engineering tasks such as debugging, feature enhancement, performance analysis, and security.
Software behaviour is typically represented in the form of execution traces. Traces, however, have historically been difficult to analyze due to the overwhelming size of typical traces. Trace analysis, more particularly trace abstraction and simplification, techniques have emerged to overcome the challenges of working with large traces. Existing traces analysis tools rely on some sort of visualization techniques to help software engineers make sense of trace content. Many of these techniques have been studied and found to be limited in many ways.
In this thesis, we present a novel approach for trace analysis inspired by the way the human brain and perception systems operate. The idea is to mimic the psychological processes that have been developed over the years to explain how our perception system deals with huge volume of visual data. We show how similar mechanisms can be applied to the abstraction and simplification of large traces.
As part of this framework, we present a novel trace analysis technique that automatically divides the content of a large trace, generated from execution of a target system, into meaningful segments that correspond to the system’s main execution phases such as initializing variables, performing a specific computation, etc.
We also propose a trace sampling technique that not only reduces the size of a trace but also results in a sampled trace that is representative of the original trace by ensuring that the desired characteristics of an execution are distributed similarly in both the sampled and the original trace. Our approach is based on stratified sampling and uses the concept of execution phases as strata.
Finally, we propose an approach to automatically identify the most relevant trace components of each execution phases. This approach also enables an efficient representation of the flow of phases by detecting redundant phases using a cosine similarity metric.
The techniques presented in this thesis have been validated by applying to a variety of target systems. The obtained results demonstrate the effectiveness and usefulness of our methods.