Satisfying multiple constraints in real-time database systems

Real-time database systems, such as military command, control and communication, avionics, radar tracking, and managing automated factories, have timing constraints and temporal consistency constraints. The timing constraint requires a transaction to be completed by a specified deadline. The temporal consistency constraints require that data read by a transaction be up-to-date. In such environments, multiple users share the same database, and some of the users may have restricted access to information from the database Hence, In addition to maintaining timing and temporal consistency constraints, it is also necessary to provide security for real-time databases. Satisfying more than one constraint in a real-time database system is challenging, because maintaining one constraint often means the sacrifice of another. In this work, we examine multiple constraints in real-time database system, specifically, maintaining security and temporal consistency.; The objective of this work is twofold. Our first goal is to incorporate security constraints without degrading real-time performance significantly. We presented two new concurrency control algorithms for maintaining security in a real-time database system. These algorithms specially aim at removing covert channels, which facilitate an indirect flow of information between users at two different security levels. We implemented our algorithms using a real-time database system simulation model and studied the performance results. Results show clear improvement of security when the algorithms are used. We also have shown that achieving security does not necessarily mean a great deal of sacrifice in maintaining real-time constraints.; Our second objective is to maintain temporal consistency of real-time database systems. We introduced a novel concept known as “triggered updates” for the improvement of temporal consistency. We identified properties of triggered updates and explained how they worked by giving both an intuitive and a probabilistic analysis. We also presented four strategies for generating triggered updates and discussed their suitability in various contexts. The results show that, with an appropriate choice of strategies it is possible to improve temporal consistency without losing much of performance in terms of timeliness.