Rapid Architecture Alternative Modeling (RAAM): a framework for capability-based analysis of system of systems architectures

[摘要] The current national security environment and fiscal tightening makeit necessary for the Department of Defense to transition away from athreat based acquisition mindset towards a capability based approachto acquire portfolios of systems. This requires that groups ofinterdependent systems must regularly interact and work together assystems of systems to deliver desired capabilities.Technologicaladvances, especially in the areas of electronics, computing, andcommunications also means that these systems of systems are tightlyintegrated and more complex to acquire, operate, and manage.Inresponse to this, the Department of Defense has turned to systemarchitecting principles along with capability based analysis. However,because of the diversity of the systems, technologies, andorganizations involved in creating a system of systems, the designspace of architecture alternatives is discrete and highlynon-linear. The design space is also very large due to the hundreds ofsystems that can be used, the numerous variations in the way systemscan be employed and operated, and also the thousands of tasks that areoften required to fulfill a capability. This makes it very difficultto fully explore the design space. As a result, capability basedanalysis of system of systems architectures often only considers asmall number of alternatives. This places a severe limitation on thedevelopment of capabilities that are necessary to address the needs ofthe war fighter.The research objective for this manuscript is to develop a RapidArchitecture Alternative Modeling (RAAM) methodology to enabletraceable Pre-Milestone A decision making during the conceptual phaseof design of a system of systems. Rather than following current trendsthat place an emphasis on adding more analysis which tends to increasethe complexity of the decision making problem, RAAM improves oncurrent methods by reducing both runtime and model creationcomplexity.RAAM draws upon principles from computer science, systemarchitecting, and domain specific languages to enable the automaticgeneration and evaluation of architecture alternatives.For example,both mission dependent and mission independent metrics areconsidered. Mission dependent metrics are determined by theperformance of systems accomplishing a task, such as Probability ofSuccess.In contrast, mission independent metrics, such asacquisition cost, are solely determined and influenced by the othersystems in the portfolio. RAAM also leverages advances in parallelcomputing to significantly reduce runtime by defining executablemodels that are readily amendable to parallelization. This allows theuse of cloud computing infrastructures such as Amazon's ElasticCompute Cloud and the PASTEC cluster operated by the Georgia Instituteof Technology Research Institute (GTRI). Also, the amount of data thatcan be generated when fully exploring the design space can quicklyexceed the typical capacity of computational resources at theanalyst's disposal. To counter this, specific algorithms andtechniques are employed. Streaming algorithms and recursivearchitecture alternative evaluation algorithms are used that reducecomputer memory requirements.Lastly, a domain specific language iscreated to provide a reduction in the computational time of executingthe system of systems models. A domain specific language is a small,usually declarative language that offers expressive power focused on aparticular problem domain by establishing an effective means tocommunicate the semantics from the RAAM framework. These techniquesmake it possible to include diverse multi-metric models within theRAAM framework in addition to system and operational level trades.A canonical example was used to explore the uses of themethodology. The canonical example contains all of the features of afull system of systems architecture analysis study but uses fewertasks and systems. Using RAAM with the canonical example it waspossible to consider both system and operational level trades in thesame analysis. Once the methodology had been tested with the canonicalexample, a Suppression of Enemy Air Defenses (SEAD) capability modelwas developed. Due to the sensitive nature of analyses on thatsubject, notional data was developed. The notional data has similartrends and properties to realistic Suppression of Enemy Air Defensesdata. RAAM was shown to be traceable and provided a mechanism for aunified treatment of a variety of metrics. The SEAD capability modeldemonstrated lower computer runtimes and reduced model creationcomplexity as compared to methods currently in use.To determine theusefulness of the implementation of the methodology on currentcomputing hardware, RAAM was tested with system of system architecturestudies of different sizes. This was necessary since system of systemsmay be called upon to accomplish thousands of tasks. It has beenclearly demonstrated that RAAM is able to enumerate and evaluate thetypes of large, complex design spaces usually encountered incapability based design, oftentimes providing the ability toefficiently search the entire decision space. The core algorithms forgeneration and evaluation of alternatives scale linearly with expectedproblem sizes. The SEAD capability model outputs prompted thediscovery a new issue, the data storage and manipulation requirementsfor an analysis. Two strategies were developed to counter large datasizes, the use of portfolio views and top `n' analysis. This provedthe usefulness of the RAAM framework and methodology duringPre-Milestone A capability based analysis.

[发布机构] University:Georgia Institute of Technology;Department:Aerospace Engineering

[关键词] System of systems