An Empirically Developed Fourier Series Model for Describing Software Failures
Abstract
In this paper, the authors draw attention to the fact the times between failures of a large software system, between any two program fixes, need not be independently and exponentially distributed, as has been often assumed in the past. It is argued that, in several instances, such times between failures occur in clusters, often systematically, and present some data to substantiate this claim. Then proposed is an empirically developed Fourier series model, which can provide an adequate description of our data, and which under certain circumstances can be used to predict future failures. Much of the analysis here is informal, and the key tool used to develop our approach is a spectrogram of the data.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 01, 1983
- Accession Number
- ADA132739
Entities
People
- Larry H. Crow
- Nozer D. Singpurwalla
Organizations
- United States Army Materiel Systems Analysis Activity