Composable Architecture & Design Applying Product Line and Systems of Systems Concepts to the Design of Unique, Complex Cyber-Physical Systems 12/03/2014 Christopher Oster
About Me B.S. Computer Science, M.S. Computer Engineering Penn State University 11 years at Lockheed Martin Systems & Software Architect at Lockheed Martin Advanced Technologies Lab, Chief Architect for DARPA Insight Program Research Interests Analytics Systems Thinking Decision Analysis Engineering Methodologies Systems Architecture 3
Chair: Dr. Jon Wade Committee Chair and Members Members Dr. Rob Cloutier (Stevens, SSE) Dr. Souran Manoochehri (Stevens, Mechanical Engineering) Dr. Ricardo Valerdi (Arizona State) Mr. Sanford Friedenthal
Problem Statement Product line and platform-based design approaches can drive down the recurring cost of new unique mission systems, however traditional product line approaches do not work well for system families where each implementation has significantly unique operational requirements (i.e. aerospace & defense) Organizations developing complex cyber-physical systems have often struggled with realizing the full value of their organizational intellectual property as structured, managed reuse has proven elusive, especially within the Aerospace & Defense market Composable design is a systems architecture and development concept focusing on composing new systems from known components, designs, product lines and reference architectures as opposed to focusing on "blank sheet" designs based on requirements decomposition alone 5
Traditional Product Lines A set of [systems] that share a common, managed set of features satisfying the specific needs of a particular market segment or mission and that are developed from a common set of core assets in a prescribed way Carnegie Mellon SEI Bosch s Product Line Taxonomy [1] 6
Composable Architectures Capture variability through an overarching reference architecture, mathematical constraints (design rules) and a catalog of alternatives at the feature, component, subsystem (or system) level The system reference architecture establishes the design context for a composable, optimization-based decision support tool suite Mathematical rules indicate what selections are compatible with other selections, what requirements constrain specific design choices and what attribute values are allowed within the context of a valid system architecture The catalog model supports strategic reuse of existing components and designs by ensuring they are considered in context by architects during new system designs 7
Hypotheses Traditional product line design methods, centered on feature tree decomposition, do not scale well into markets and problem types where constraining system configurations to a limited set of variants is not desired, but where both commonality and reuse of system and infrastructure elements are. A composable design approach for managing this class of product lines provides a mechanism for strategically managing commonality while supporting high variability in final products. The specific value of implementing a composable design framework in a given organization will be tied to the impact of reuse of given elements (including design elements, supporting infrastructure, intellectual property, etc) by category. Specifically for the design, development, production and test of a cyber-physical system, this value will exceed 25% of the cost of a system, as compared to a design created from scratch. The implementation of a composable design framework for an organization will drive up the likelihood of reuse of a given component over more traditional reuse methodologies, increasing the value of a given reusable asset by greater than 25%.
Innovation and Contributions to Field Composable design provides a mechanism for applying product line concepts to families of solutions with widely varying mission use cases This research is providing the following contributions to the systems engineering field Formalizing a composable data modeling methodology for complex cyber-physical product families Developing and verifying a mathematical constraint language and toolkit extending SysML to capture variability rules and capability interfaces to ensure composed solutions are valid and mission compliant Designing and Implementing an optimization guided design synthesis capability applying a novel multi-tier design optimization approach 9
Composable Design Reference Architecture Composable System Reference Architecture Structural Architecture Model Architectural Variability Model Component Capabilities Model Variability Constraints Model System Verification Model System Optimization Model Descriptive Model Parametric Model Analysis Model SoS Mission/ SoS Threat Mission/ SoS Models Threat Mission/ Capability Models Performance Threat Models Models SoS Mission/ SoS Threat Mission/ SoS Models Threat Mission/ Mission Models Threat Analysis Models Models Cost Model 10
Constraint-based Design Input 1 Input 2 Input 3 Input 4 Input 5 Input 6 Constraints IC1: IC2: IC3: SC1: SC2: SC3: 1. Establish constraint order based on data dependency graph 2. Navigate model for constraint input values 3. Populate execution framework with known data values 4. Execute independent constraints 5. For each available configuration 1. Execute each selection-dependent constraint in the constraint set 2. If all constraints pass, configuration is valid 3. If any constraint fail, record which constraint resulted in failure and end 6. Populate results into output variables and establish valid configuration list for this constraint 7. Move to next constraint set and repeat Available Configs Valid Configs Required_Component_Bandwidth must be less than or equal to Total_Comms_Bandwidth SUM(Selected_Components_Bandwidth) Multiple sets of individual constraints allow architects to establish rules of design without pre-establishing a full feature tree common in product line approaches and variability modeling 11
Approach Guided Synthesis Component Technologies Representative Missions Reference Architecture Tier One: SoS Characterization Batch Mode (offline) Interactive Tier Two: Tradespace Trimming Mission Definition Selection Made Tier Three: Guided Optimization If we have Governing reference architecture Model-based relationships and constraints Links to digital design artifacts Analyses for system level KPPs Candidate Configuration(s) System Architect We can optimize a system design for a given mission need, but Finding an optimal design is computationally challenging for a sufficiently complex system given the combinatorics of the problem. This tiered approach is structured to allow for real-time assessment such that an architect can interactively design a near-optimal system. 12
Tool Architecture 13
Predicted benefits of Composable Design Faster turnaround time for bids and bid architectures Better handoff of information from bid to program start Better control of organizational reuse Drive cost out of supply chain through common parts Increase the value of infrastructure investments Limit the cost of engineering release Initial pilot effort underway with Lockheed Martin Results
Conclusions/Future Research Composable Design methods provide a mechanism for managing organizational intellectual property in a value-driven way Methods and framework will continue to be developed in accordance with the research design previously discussed True impact of the methodology will be measured through industry piloting. Lockheed Martin Space Systems pilot already underway Additional engagement from non-aerospace desired
Contact Christopher Oster coster@stevens.edu