A Model for Reasoning About the Privacy Impact of Composite Service Execution in Pervasive Computing

摘要

Service composition is a fundamental feature of pervasive computing middleware. It enables users to leverage available computing power by using existing services as building blocks for creating new composite services. In open and dynamic environments, service composition must be flexible enough to admit realization by different executable workflows that have similar functionalities but that present different partitions of tasks among available services. This flexibility, however, raises new privacy issues e.g., a single service performing all tasks of a workflow has access to more data than different services executing parts of the workflow. In this paper we propose a model that enables users to reason about the impact on privacy of executing a composite service. The model is based on an extension of Fuzzy Cognitive Maps, and considers the impact of the composition as a whole according to the partition of tasks. We introduce our extension called Fuzzy Cognitive Maps with Causality Feedback, describe how they can be used to model the relationship among different personal data and the privacy impact of their disclosure, and give an example of how the model can be applied to a composition scenario.