Toward Effective Multi-capacity Resource Allocation in Distributed Real-time and Embedded Systems

Standards-based quality of service (QoS)-enabled component middleware is increasingly being used as a platform for developing distributed real-time embedded (DRE) systems because it provides features and capabilities that enhance the design, development, evolution, and maintenance of these systems. Performing effective resource management is an important and challenging issue for DRE systems due to their unique characteristics, including (1) constraints in multiple resources (e.g., limited computing power and network bandwidth) and (2) highly fluctuating resource availability and input workload. These systems can benefit from the integrated operation of resource management algorithms and the Resource Allocation and Control Engine (RACE), which is a middleware framework that enables adaptive resource management capabilities in DRE systems to ensure application QoS requirements are met. This paper identifies key challenges in designing and extending resource allocation algorithms in the context of DRE systems. We present an empirical study of popular bin-packing algorithms enhanced to meet these challenges. Our analysis identifies patterns in the applicability of these algorithms—based on the input application characteristics—that aid in generating appropriate heuristics for use in production DRE systems.