Engineering dynamic real-time distributed systems: architecture,system description language, and middleware

摘要

The paper presents an architectural framework and algorithms fornengineering dynamic real-time distributed systems using commercialnoff-the-shelf technologies. In the proposed architecture, a real-timensystem application is developed in a general-purpose programmingnlanguage. Further, the architectural-level description of the systemnsuch as composition and interconnections of application software andnhardware, and the operational requirements of the system such asntimeliness and survivability are specified in a system descriptionnlanguage. The specification of the system is automatically translatedninto an intermediate representation (IR) that models the system in anplatform-independent manner. The IR is augmented with dynamicnmeasurements of the system by a language runtime system to produce andynamic system model. The dynamic model is used by resource managementnmiddleware strategies to perform resource management that achievesntimeliness and survivability requirements. We present two classes ofnalgorithms: predictive and availability-based, for performing resourcenallocation. To validate the viability of the approach, we use anreal-time benchmark application that functionally approximates dynamicnreal-time command and control systems. The benchmark results illustratenthat the middleware is able to achieve the desired timelinessnrequirements during a number of load situations. Furthermore,navailability-based allocation algorithms perform resource allocationnless frequently, whereas predictive algorithms give a better steadynstate performance for the application