A data structure with predictable performance for real-tiime Java systems

Advancement of real-time features in high-level programming languages is taking place. The Real-Time Specification for Java (RTSJ) describes strong execution constraints for real-time applications developed in Java. RTSJ mainly concentrates on three features, scheduling, memory management, and synchronization, for making it realtime applicable. The run-time library in Java cannot be used for real-time applications. Memory allocations play an important role in the real world in real-time applications. Java has a collections framework in its run-time library that supports dynamic memory allocation. The run-time properties of this framework need to the examined and modified to make it applicable to real-time applications.; In this work, we examine the ArrayList class, which is a part of the collection framework, and describe an approach towards making it real-time applicable. The behavior of the ArrayList was reformulated to obtain predictable performance. Various testing mechanisms were applied to increase confidence in the fidelity of the real-time ArrayList vis-a-vis the regular ArrayList. We present experiments to compare the behavior of the real-time ArrayList to that of the regular ArrayList and show that the modified ArrayList is more predictable and real-time applicable.