Complex Adaptive System Composition And Design Environment (CASCADE)The Defense Sciences Office at the Defense Advanced Research Projects Agency (DARPA) is soliciting innovative research proposals that will enable and demonstrate new design capabilities for complex adaptive systems. Proposed research should develop and/or exploit innovative approaches in mathematical abstraction and composition for the design of dynamic, adaptive and resilient systems with unified understanding of system structures, behaviors and interactions across multiple spatiotemporal scales.
Specifically excluded is research that results in evolutionary improvements to the existing state of practice, such as federated models for military systems, single- or limited-domain network, discrete event or Monte Carlo models, Systems Modeling Language (SySML), or abstraction and composition models with primary application to software and/or electronics.Performance in the CASCADE program will occur in two technical areas (TAs):
* TA1: Mathematical foundations for unified abstraction, composition and adaptive
behavior; and,
* TA2: Domain applications: design knowledge, problems and data in the areas of military
SoS or resilient urban infrastructure.
DARPA strongly encourages fully integrated TA1 and TA2 teams...Defense Advanced Research Projects AgencyDARPA_46ad30a4-ad5d-11df-9c96-10167a64ea2aDefense Sciences Office (DSO)John S. PaschkewitzTechnical POC: Program Manager, DARPA/DSODepartment of Defense... dynamic, adaptive and resilient systems_816e1f4b-97e6-11e5-8f8e-68bf856e8e00To enable and demonstrate new design capabilities for complex adaptive systems._816e1f4c-97e6-11e5-8f8e-68bf856e8e00HeterogeneityDepartment of Defense (DoD) and urban infrastructure capabilities are increasingly based on the integration and coordination of heterogeneous systems using System-of-Systems (SoS) architectures.ResiliencyHowever, it is difficult to model and currently impossible to systematically design such complex systems using state-of-the-art tools, leading to inferior performance, unexpected problems, and weak resilience. ScalabilityThis inadequacy in design capability results from the complexity of interactions between system structures and behaviors across multiple time and length scales that cannot be adequately modeled using conventional approaches.FederationAs an example, the typical design approach for a military SoS architecture is to use a federated modeling infrastructure that couples campaign- or mission-level models via requirements to physics-based platform models, with the resulting behaviors evaluated using engagement models.ModularityTo manage complexity, interfaces and interactions are controlled as much as possible with modularity being a preferred outcome.NetworkingWhile these federated models work for traditional monolithic platforms, they fail to capture the networked, coordinated effects that cross layers of abstraction characteristic of SoS architectures. These networked effects are therefore unanticipated and are described as emergent behaviors.CoordinationSustainabilityThese models also do not include logistics or sustainment constraints explicitly in the design space, leading to challenges in addressing attrition and overall systems cost.AdaptabilitySystem adaptability is limited to scenarios contemplated in the campaign- or mission-level models and is manifested as "playbooks" of behaviors in response to known scenarios. System resilience is weak because of the optimization to known threats and the limited number of reconfiguration options imposed by modularity and interfacial control.DynamicsThe DARPA Complex Adaptive System Composition And Design Environment (CASCADE) program seeks to address these shortcomings and fundamentally change how systems are designed for real-time resilient response to dynamic, unexpected environments.UnificationThe goal of CASCADE is to provide both a unified view of system behavior, allowing understanding and exploitation of these complex interactions, and a formal language for complex adaptive system composition and design.UnderstandingExploitationMathematicsThis unified view of system behavior, enabled by appropriate mathematical foundations, will also enable adaptation to unanticipated environments using arbitrary system components by providing a framework to dynamically identify and correct deficient system capabilities.System DesignIf successful, CASCADE will fundamentally change paradigms of system design both by elucidating underlying design principles for emergent behavior in complex systems and by highlighting counterintuitive new design strategies to achieve resilience.ArchitectureThe program will impact the DoD by enabling the design of dynamic, adaptive and resilient SoS architectures that go beyond static "playbook" architectures as well as the development of dynamic adaptive infrastructure and plans for systematic establishment of resilient community functions (e.g., hospitals or public safety) in response to adverse events.InfrastructureCommunity FunctionsMathematical FoundationsMathematical foundations for unified abstraction, composition and adaptive
behavior_816e1f4d-97e6-11e5-8f8e-68bf856e8e00TA1Mathematical foundations for abstraction, composition and adaptive behavior
The objective of this area is to provide a unified formal mathematical foundation for complex adaptive system design. TA1 proposers must provide an integrated solution that addresses all of the following [objectives]...
TA2 systems will use the abstraction and composition frameworks along with environmental data to construct and act on a model of the system state in a time-dynamic manner. Since the resulting design and configuration spaces will be high-dimensional, formal methods for approximate reasoning over potentially stochastic state data will be required. Elements of theory from areas such as temporal logic, policy-based planning algorithms, agent models, or model predictive control will need radical modification to take full advantage of the new abstraction and composition frameworks. Proposers must identify the limitations of their starting approaches as well as a strategy for computationally scalable implementation in the new abstraction and composition framework. Neither the state of the environment nor that of the responding system will be known precisely. There will also be sources of error due to imperfect data transmission, sensing limitations, etc. These stochastic and/or probabilistic quantities must be accounted for in a rigorous manner. Approaches that include adding white noise to deterministic data are explicitly not of interest. Additionally, proposed robust and resilient adaptive system designs and design frameworks must incorporate stochasticity in an intrinsic (and decidedly non-ad hoc) fashion. Proposers must address how their frameworks will enable a capability for "reasoning over uncertainty," so that proffered actions may be assessed relative to potential costs (e.g., resources, time, etc.) TA1 performers must also define a notional Application Program Interface (API) to facilitate integration by TA2 performers. At a minimum, this API definition must include basic definitions of data structures, functions, routines, protocols and tools that build on the core mathematical foundations and algorithms.AbstractionDescribe system structures, behaviors, constraints and events in a way that exposes only the information required to define the resulting function._816e203a-97e6-11e5-8f8e-68bf856e8e00TA1.1The goal of abstraction is describing system structures, behaviors,
constraints and events in a way that exposes only the information required to define the resulting function. In complex systems, standard reductionist abstraction approaches fail to correctly capture dynamic system structures or behaviors resulting from interactions at lower levels of abstraction and across layers of abstraction. Therefore, DARPA is explicitly not interested in approaches such as detailed single-domain graph theoretic models, hierarchical multi-scale decomposition, UML/SySML and simplified metamodels; frameworks such as contract- and platform-based design are not sufficient and also not of interest. Proposers must demonstrate that their alternative frameworks address the complexity of interfaces, multiscale interactions, and dynamics in a computable and computationally tractable manner. Proposers must also highlight how their frameworks can enable design across multiple scales and domains of abstraction. Proposers must discuss how their abstraction frameworks unify stochastic, discrete and continuous dynamics. Areas of mathematics that may yield such a framework include but are not limited to category theory and model theory.Composition Synthesize functions from structures and behaviors in response to events._816e272e-97e6-11e5-8f8e-68bf856e8e00TA1.2New frameworks are required to synthesize functions from structures and behaviors, subject to constraints, in response to events. Existing frameworks for composition of structures and behaviors are either entirely ad hoc or are limited to domain-specific abstractions. Proposers must demonstrate that their framework enables both compositionality (e.g., system functional properties can be computed from local properties of structures and behaviors) and composability (e.g., system functional properties do not change with interactions between structures, behaviors and constraints) across multiple levels of abstraction. Successful composition frameworks are likely to utilize new abstraction frameworks in concert with currently unexploited mathematical techniques to facilitate composition. For example, mathematical techniques such as algebraic topology and geometry, sheaf theory, cochain and process algebra, polytopes, triangulations and cluster algebras may be appropriate formalisms. Combinations of abstractions coupled to compositional mathematics may include concepts such as dynamical system template-anchor formalisms for behavioral composition in robotics or so-called "equation-free" methods.Adaptation Sense, reason and act on the design composition space._816e279c-97e6-11e5-8f8e-68bf856e8e00TA1.3Dynamic adaptation of the complex system is enabled by a unified view of and a formal language for functional composition across all possible structures and behaviors. However, to achieve adaptation, the capability to sense, reason and act on the design composition space is also required. "Sensing" in this context means that the complex system has a means to determine its state in an a priori unknown environment with an unpredictable mix of sensing and communication modalities (i.e., how does this system know that an adaptation is required?). Proposers must explain how their frameworks will address distributed, asynchronous assessment of, and uncertainty in, determining the environmental state. Concepts that may partially address this issue include but are not limited to concepts of dynamic heterarchy, information theoretic frameworks, and primaldual optimization.Domain ApplicationsProvide a unified formal mathematical foundation for complex adaptive system design._816e279d-97e6-11e5-8f8e-68bf856e8e00TA2Domain application in the areas of military SoS or resilient urban infrastructure -- TA2 performers will integrate deep knowledge of domain challenges and the novel mathematical foundations of TA1 in powerful new domain-specific modeling and design frameworks. Proposers are expected to leverage and share available data sets or results from modeling capabilities for baselining and validation of the capabilities to be developed. TA2 performers are also expected to work closely with the Challenge/IV&V partners to define specific quantitative figures of merit for system performance as well as customizing the challenge problems to the domain of interest.
Proposers to the CASCADE program can choose to pursue either of two domains of interest: military system of systems architectures or resilient urban infrastructure. Although other domains are possible that would demonstrate the capabilities of the techniques being developed in the CASCADE program, only the aforementioned two areas will be considered for funding. Further definition of the two domains of interest are as follows:Military System of Systems_816e279e-97e6-11e5-8f8e-68bf856e8e00TA2.1Proposers in this space should note that while CASCADE is an unclassified program, it is anticipated that later phases of the program data and scenarios may potentially be subject to International Traffic in Arms Regulations (ITAR) restrictions (See Section III.D.). Proposers for military SoS must ensure that their proposed scenarios, system definitions, and data are unclassified. Proposals must not include any classified information either in the proposal itself or in addenda to the proposal. Some example military scenarios that meet these requirements are [as follows] ...
Proposers are strongly encouraged to consider these scenarios but may propose challenge problems from other military domains that entail significant complexity, involve constraints that are not represented in state-of-the-art design tools, and include multiple structural and behavioral domain areas not adequately and formally modeled in current frameworks. Single- or limited-domain systems, such as electronic warfare (EW) or communications networks are explicitly not of interest.Battlefield MedicineProvide adaptive battlefield medicine capability._816e279f-97e6-11e5-8f8e-68bf856e8e00TA2.1.1Adaptive battlefield medicine capability for forward resuscitative surgery, with
options including differing novel autonomous system, teleoperation, medevac and field hospital architectures, incorporating engineering system and subsystem architectures with detailed (e.g., requiring solution of partial differential equations) time-dependent physics, communications requirements, blood transport, supply chain management, and surgical capabilities as a function of terrain, operational environment and casualty rates and types.LogisticsProvide adaptive logistics capability._816e27a0-97e6-11e5-8f8e-68bf856e8e00TA2.1.2Adaptive logistics capability, with options including novel autonomous and/or
manned land, sea or air assets and distribution center architectures, incorporating details of engineering system and subsystem designs for delivery and distribution platforms with detailed time-dependent physics, communications and coordination modalities, infrastructure for platform learning algorithms and data sustainment, and support for both short-term "pop-up" tactical logistics and long-term sustaining support as a function of terrain, operational conditions and supply requirements, and potential adversary actions (e.g., road closures, communications jamming).Search & RescueSearch for and rescue downed airmen or forward operators in hostile environments._816e27a1-97e6-11e5-8f8e-68bf856e8e00TA2.1.3Adaptive search and rescue capability for downed airmen or forward operators in hostile environments, with options including novel autonomous and/or manned land, sea or air assets with coordination using land, sea, air, space and cyber systems, incorporating engineering system and subsystem architectures with detailed, time-dependent physics, communications and control, and sensing as a function of terrain, operational environment, and adaptive adversary capabilities.Resilient Urban Infrastructure System_816e27a6-97e6-11e5-8f8e-68bf856e8e00TA2.2An example system of interest is a time-dependent and dynamic community function such as health care or public safety as a composition of power, communications, transportation, and logistics in response to changing environmental conditions. Single- or limited-domain systems such as smart grid, communications or transportation networks, or device-centric "smart city" systems are explicitly not of interest. TA2 proposals must discuss in detail the chosen military SoS or resilient urban infrastructure system that will be assessed and redesigned for adaptive resilience. In addition, the suitability of the challenge problem for the mathematical foundations of TA1 needs to be addressed. Using TA1 frameworks as a "backplane" or a "conversion" tool between existing modeling tools is explicitly not of interest. TA2 performers must outline an integration strategy for inclusion of TA1 frameworks into a modeling and design environment. Specifically, TA2 proposers must address the following areas ...
To facilitate successful adoption of the tools and designs developed in CASCADE by the broader application communities, proposers are expected to clearly identify a transition strategy appropriate to the domain of interest (military SoS or resilient urban infrastructure). As stated above, both the tools and the design libraries developed by CASCADE performer teams are expected to be available to the respective design communities after program completion using an open data and model infrastructure. The details of this infrastructure are context-specific, but proposers must provide a plan for a persistent development framework after the end of the program that provides open access to data and design capability.System DefinitionDefine and highlight the complexity of the candidate system._816e28d2-97e6-11e5-8f8e-68bf856e8e00TA2.2.1Proposers must define and highlight the complexity of the candidate system (or SoS) to be assessed and redesigned for maximum adaptive resilience. The system must be sufficiently complex such that existing tools are incapable of providing meaningful design insight. This implies that at a minimum there are multiple coordinated functions, with multiple levels of abstraction required to describe the structures or behaviors, non-linear coupling of constraints to structures or behaviors, and dynamic behaviors with characteristic times spanning multiple orders of magnitude.Design Framework Assessment & Strategy Articulate the limitations or inability of current modeling frameworks to correctly capture dynamic and/or emergent behaviors of the candidate system._816e29c2-97e6-11e5-8f8e-68bf856e8e00TA2.2.2Proposers must articulate the limitations or inability of current modeling frameworks to correctly capture dynamic and/or emergent behaviors of the candidate system. Proposers must highlight specific difficulties in using the modeling frameworks to achieve design -- what is difficult or impossible to capture in the translation of functional needs to structural and behavioral composition? Proposers must InsertionExplain in detail where the new mathematical foundations of TA1 will be inserted into the modeling and design infrastructure_816e2aa8-97e6-11e5-8f8e-68bf856e8e00TA2.2.2.1IntegrationDefine requirements for TA1 integration_816e2bb6-97e6-11e5-8f8e-68bf856e8e00TA2.2.2.2StrategyDefine an insertion and testing strategy for the new capability._816e2cba-97e6-11e5-8f8e-68bf856e8e00TA2.2.2.3CapabilityProposers must also identify a strategy for extending and exercising the capability to meet the milestones in Section I.D._816e3278-97e6-11e5-8f8e-68bf856e8e00TA2.2.2.4System Assessment Metrics & DataDefine data or other validated models that are available for benchmarking and challenge problem formulation._816e3430-97e6-11e5-8f8e-68bf856e8e00TA2.2.3Proposers must define what data or other validated models are available for benchmarking and challenge problem formulation. Proposers must identify where data are not available and articulate a modeling strategy for defining such data. TA2 performers are expected to share benchmarking data with the broader application community in either the military SoS or urban resilience areas (see Section I.F for testbed information). Proposers must also define figures of merit for system functional outcomes and relevant characteristics of constituent structures or behaviors related to these figures of merit. Proposers should anticipate refinement or modification of these figures of merit after discussion with the Challenge/IV&V partners.2015-11-232015-11-30