Toggle Main Menu Toggle Search

Open Access padlockePrints

ADaCS: A Tool for Analysing Data Collection Strategies

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.

Publication metadata

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


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