STAR: secure real-time transaction processing with timeliness guarantees

Abstract

Real-time databases are needed in security-critical applications, e.g., e-commerce, agile manufacturing, and military applications. In these applications, transactions and data items can be classified into several security levels according to their clearance and sensitivity levels. It is essential for real-time databases to prevent illegal direct/indirect transfer of sensitive data, e.g., trade secret, manufacturing, or operational data, between transactions belonging to different security levels. Further transactions should be committed within their deadlines, i.e., before the market, manufacturing, or battlefield status changes. In this paper we present a novel real-time database architecture, in which illegal direct/indirect inter-level information flows are prevented while controlling the deadline miss ratio for admitted transactions to remain below a certain threshold. In our approach, mandatory access control mechanisms are applied for security purposes. QoS management, admission control, and feedback control schemes are applied to support certain guarantees on the miss ratio against potential overload and data conflicts. A detailed simulation study shows that our approach can support the specified miss ratio preventing illegal information flows even in the presence of unpredictable workloads and varying degrees of data contention, whereas baseline approaches fail.