Efficient Load Balancing Schemes for Large-Scale Real-Time HLA/RTI Based Distributed Simulations

The real-time extension of high level architecture (HLA) is very essential and useful for many large-scale distributed simulation systems. Most previous attempts to design the real-time run time infrastructure (RT-RTI) have enabled the usage of supported scheduling and prioritization services from underlying real-time operating systems (RTOSs) augmented by communication QoS mechanisms. In this paper, we wish to build on this functionality by proposing an algorithm that differentiates services processing within the RTI itself by incorporating resources load balancing mechanisms with several scheduling and allocation policies. We focus our efforts on making the RTI's internal operations organized and well suited to the tasks and services it will be providing throughout the lifetime of HLA-compliant federations. We discuss our load balancing strategies and demonstrate through our analytical performance evaluation and simulation experiments that our proposed real-time RTI framework exhibits a better performance in terms of the number of tasks served within deadlines compared with existing real-time RTI frameworks.