An Authorization System for Temporal Data Avigdor Gal Technion — Israel Institute of Technology Faculty of Industiral Engineering & Management Technion City, Haifa 32000, Israel and Rutgers University Department of Management Science & Information Systems 94 Rockafeller Road, Piscataway, NJ 08854-8054
[email protected] Vijayalakshmi Atluri Rutgers University Department of Management Science & Information Systems 180 University Avenue, Newark, NJ 07102
[email protected] Gang Xu Rutgers University Division of Computer & Information Sciences 110 Frelinghuysen Road Piscataway, NJ 08854-8019
[email protected]
Abstract We present a system, called Temporal Data Authorization Model (TDAM), for managing authorizations for temporal data. TDAM is capable of expressing access control policies based on the temporal characteristics of data. TDAM extends existing authorization models to allow the specifications of temporal constraints on data, based on data validity, data capture time, and replication time, using either absolute or relative time references. The ability to specify access control based on such temporal aspects were not supported before. The formulae are evaluated with respect to various temporal assignments to ensure the correctness of access control.
1 Introduction Temporal datum is a data value associated with information such as the time at which data has been captured and the time interval during which a data value is valid. Given
Proceedings of the 18th International Conference on Data Engineering (ICDE’02) 1063-6382/02 $17.00 © 2002 IEEE
a specific data model, temporal data can be collected into versions of objects, distinguished from one another on the basis of their temporal specifications. The use of temporal data has become widespread in recent years, within applications such as data warehouses (e.g., daily sales and seasonal catalogs), spatiotemporal databases, and electronic commerce (with applications in finance and digital libraries). For example, a digital library should be able to enforce access control based on the time at which a document has been “checked-out,” (referred to as replication time) to allow flexible digital library policies for restricting access to a digital book after the expiry of the temporal-based authorization. In this demo, we present a system, implementing aspects of TDAM (Temporal Data Authorization Model) [1], for managing authorizations for temporal data. TDAM is capable of expressing access control policies based on the temporal characteristics of data. TDAM extends existing authorization models to allow the specifications of temporal constraints on data, based on data validity, data capture time, and replication time, using either absolute or relative time references. We introduce language, which allows
Figure 1. Demo architecture TDAM to express security policies such as “a subject is allowed to read financial data 15 minutes after it has been captured,” “only managers are allowed to read prospective product plans,” and “a digital book can be read for 21 days from time of download.” The ability to specify access control based on such temporal aspects were not supported before. The formulae are evaluated with respect to various temporal assignments to ensure the correctness of access control.
2 Details The
demo
is
available
at
http://itec4.rutgers.edu/TempAuth/servlet/TempAuth.TempSecurity.TempSysLogin.
Figure 1 provides a three-tier model, consisting of a client, a communication server, and a database server. The communication between the client and the server is based on Java Servlet Technology. All in all, we have built 31 classes, grouped into 7 packages. The core of the system lies at the database server, which implements the temporal authorization model. In short, it appends users ACL to each query request and puts it into its WHERE clause. Standard JDBC-ODBC bridge is used to isolate the database from external SQL operations. In order to achieve high-performance, a pool of connections are built into the server. Furthermore, the server will remember each established http session and when the user comes back with another request, the server will reuse the open connection. We use as an example scenario the financial information, as provided to the readers of a fictitious journal, the E-WS Journal. The demo has three types of subjects, namely publisher, subscriber, and user. The publisher is the sysadmin of the site, and therefore can insert new articles and set authorizations. A publisher is allowed to add news items
Proceedings of the 18th International Conference on Data Engineering (ICDE’02) 1063-6382/02 $17.00 © 2002 IEEE
through this screen. For the purpose of the demo, we have also added a utility that generates random “news items” and publishes them authomatically. All authorizations in the system are of the form (pg; o; m; sign; ), where pg is a privileged group (e.g., publisher, subscriber, and user), o is an object (e.g., news article), m is a privilege mode (e.g., read), sign 2 f+; g indicating access or denial, and is a well-formed formula, which is our novel addition. In this demo instance, we have assumed three privileged groups, one type of object (news article), and one privilege mode (read). In this instance, a subscriber is allowed to see all news items that are more than three minutes old. In the T DAM language, such an authorization is stated using = tr > ts + 3, where tr stands for a request time and ts stands for the valid time starting time point. A user is allowed to see 25 minutes delayed information only . In the T DAM language, such an authorization is stated as = tr > ts + 25. Throughout the demo, we shall demonstrate the capabilities of the system in changing the user temporal authorizations.
Acknowledgment The system is supported by ITECC: Information Technologies and Electronic Commerce Clinic at the Rutgers School of Business.
References [1] V. Atluri and A. Gal. An authorization model for temporal and derived data: Securing information portals. ACM Transactions on Information and System Security (TISSEC), 5(1), 2002. to appear.
