Browse by author
Lookup NU author(s): Dr John Mace, Nipun Thekkummal, Dr Charles Morisset, Professor Aad van Moorsel
Full text for this publication is not currently held within this repository. Alternative links are provided below where available.
© Springer International Publishing AG 2017. Given a model with multiple input parameters, and multiple possible sources for collecting data for those parameters, a data collection strategy is a way of deciding from which sources to sample data, in order to reduce the variance on the output of the model. Cain and Van Moorsel have previously formulated the problem of optimal data collection strategy, when each parameter can be associated with a prior normal distribution, and when sampling is associated with a cost. In this paper, we present ADaCS, a new tool built as an extension of PRISM, which automatically analyses all possible data collection strategies for a model, and selects the optimal one. We illustrate ADaCS on attack trees, which are a structured approach to analyse the impact and the likelihood of success of attacks and defenses on computer and socio-technical systems. Furthermore, we introduce a new strategy exploration heuristic that significantly improves on a brute force approach.
Author(s): Mace JC, Thekkummal N, Morisset C, Van Moorsel A
Publication type: Conference Proceedings (inc. Abstract)
Publication status: Published
Conference Name: 14th European Workshop on Performance Engineering (EPEW 2017)
Year of Conference: 2017
Pages: 230-245
Online publication date: 13/08/2017
Acceptance date: 02/04/2016
Publisher: Springer Verlag
URL: https://doi.org/10.1007/978-3-319-66583-2_15
DOI: 10.1007/978-3-319-66583-2_15
Library holdings: Search Newcastle University Library for this item
Series Title: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
ISBN: 9783319665825