Life-cycle privacy policies for the ambient intelligence
Harold van Heerde · 2006
A smart, anticipating and learning environment will have a great impact on privacy. Ambient Intelligence will be everywhere, is invisible, has powerful sensing capabilities and most of all has a memory [2]. This does introduce a privacy risk, because context histories are vulnerable for attacks (hacking the context database, corrupt database administrators, human mistakes, et cetera), so storing of privacy sensitive data is not desirable in privacy perspective. However, high quality and quantity of data improves smartness for the Ambient Intelligence, while less and degraded data improves privacy. Therefore the problem to be solved is how to balance smartness and privacy. We propose to give the control over the life cycle of their data to users, who themselves can balance their needs and wishes in terms of smartness and privacy. Balancing the control and the information flow between owner and collector of data decreases the asymmetry of information and the chance of privacy violations [1]. We let users (the owners of the collected data) specify Life-Cycle Policies which will be bound to the acquired data. These data is stored in a privacy aware context database which degrades the data progressively according to the policy. Data is modeled as triplets (time, person, context). Triplets can take values in context states, exhibiting a certain level of accuracy specified in a domain generalization graph of that attribute. The generalization graphs form together a cube, in which each dimension