Integrating Systems and Software Engineering: Observations in Practice October 29, 2007 Tom Schroeder FCS MGV Software Engineering Manager BAE Systems, Ground Systems October 29, 2007 © 2007 BAE Systems Land & Armaments L.P. 1 Purpose Provide a perspective on: Integrating Systems and Software Engineering for Complex Systems. • From the perspective of a major supplier and integrator • Starting Build 2 What are the significant issues and concerns expressed by project suppliers? What works in practice? What doesn’t? How can the Incremental Commitment Model be improved? October 29, 2007 © 2007 BAE Systems Land & Armaments L.P. 2 Ground Systems – A Summary Protected Fighting Platforms for Today’s Warfighter as well as the Battlefield of Tomorrow • Predominant Supplier to the U.S. Army Heavy Brigades with • • Bradley, HERCULES, Paladin, M113 Mine-Protected Wheeled Vehicles FCS Manned Ground Vehicles and Armed Robotic Vehicle Key Technologies • Advanced Protection and Mobility Solutions for Soldiers, Manned • • Vehicles and Robots Outstanding Program Management and Experienced Workforce 3,250 employees, including more than 600 technologists World-Class Development Processes • CMMI Level 5 Software and Systems Engineering Process • Physics-Based Models & Real-Time Simulation Capabilities • Rapid Prototyping of Complex Systems Lean, Cost-effective Production Facilities GS is a modern, efficient, full-spectrum developer, integrator and supplier of survivable, lethal ground combat platforms and advanced technologies October 29, 2007 © 2007 BAE Systems Land & Armaments L.P. 3 FCS Brigade Combat Team (BCT) 18 Integrated Systems + 1 Network + 1 Soldier Infantry Carrier Vehicle (ICV) Manned Systems Unmanned Aerial Vehicles Class IV Command and Control Vehicle (C2V) Class II Class I Class III Mounted Combat System (MCS) Unattended Munitions Unattended Ground Sensors NLOS LS Intelligent Munitions Systems Reconnaissance and Surveillance Vehicle (RSV) Unmanned Ground Vehicles Non-Line of Sight Cannon (NLOS-C) Non-Line of Sight Mortar (NLOS-M) Small (Manpackable) UGV ARV Aslt ARV RSTA FCS Recovery and Maintenance Vehicle (FRMV) Armed Robotic Vehicle (ARV) Medical Vehicle Treatment (MV-T) Medical Vehicle Evacuation (MV-E) ARV-A (L) MULE (Countermine) MULE (Transport) FCS is about the 21st Century Soldier October 29, 2007 © 2007 BAE Systems Land & Armaments L.P. Approved for Public Release, Distribution Unlimited, TACOM 20 SEP 2006, case 06-208. 4 MGV Base Platform Software Context Real–Time, Deterministic Software - Isolated from the C2 Network An MGV Vehicle Platform Base Vehicle ISR Inter-Platform Behaviors: “Brigade Combat Team SOS” C2 Logistics / Sustainment Training Integrated Platform: “MGV Platform” Vehicle Platforms Must be Designed for Integration and Evolution October 29, 2007 © 2007 BAE Systems Land & Armaments L.P. 5 Characteristics of Project from a Software Perspective Extremely large System of Systems project Target computing hardware developed concurrently with software • Includes vehicle computers, remote interface units, servo control units • Includes many additional subsystems internally produced/procured and externally provided • Most final hardware not available during early build iterations Must substitute “host” and “surrogate” development environments Evolve Simulators to Emulators and Stimulators Many decisions not under platform supplier’s control Interface contracts extremely important due to size of SOS • Successive refinement and elaboration through multiple levels of Systems Engineering to • • • October 29, 2007 Software Engineering At Software level, utilize IDL and interface code generators to minimize architecture dependencies Utilize common design patterns for communications across deployable software entities Pub-sub, proxy, etc. Ideally generate IDL directly from tagged attributes in shared design model Requires all groups to use same interface modeling approach Allows for one data dictionary © 2007 BAE Systems Land & Armaments L.P. 6 Software Build 1 Objectives Build initial prototype vehicles for one vehicle type (“variant”), the NonLine of Sight Cannon (NLOS-C), to be assembled in 2008 Develop “threshold path” common components and software for a common chassis. • Hybrid Electric Drive Powertrain, Driving functions, Vehicle Management • Power Distribution, Remote Interface Units, Servo Control Units • Embedded Training Develop “threshold path” mission equipment and software for the weapon and mission control functions Develop low-cost software and hardware “surrogates” to stand-in for functionality that is not yet available, such as the sustainment system, the displays and user interface system, etc. Develop and improve processes for software development and integration Reduce risk for objective vehicles and software development October 29, 2007 © 2007 BAE Systems Land & Armaments L.P. 7 Software Build 2 Objectives Build out software infrastructure for all MGV variants Ensure that MGV common components and common software can be configured for every MGV variant Develop vehicle and mission module control functions for all MGV variants Utilize and integrate externally provided software and subsystems • Prove viability of layered software infrastructure • Define peer interfaces at application level across SOS Continue to develop and improve processes for systems and software development and integration • Better integrate MGV Systems and Software Engineering Workflows • Improve Coordination with SOS Development Continue to reduce risk for objective vehicles and software development October 29, 2007 © 2007 BAE Systems Land & Armaments L.P. 8 Integration of Systems and Software Engineering: MGV Sys/Sw WorkflowWorkflow Design Stages Engineering Cycle 2.n+1 @ Points in Time Engineering Cycle 2.n Sys PIDS/AL1 Rq Engineering Cycle 2.n+2 Sys PIDS/AL1 Rq Sys Design Sys PIDS/AL1 Rq Sys Design Sys Spec/Alloc Sys Spec/Alloc Systems EC 2.n+1 SSys AL2 Rq Sys Design Sys Spec/Alloc SSys AL2 Rq SSys Design pec/Alloc SSys Design SSys Spec/Alloc SW Mgmt & Env SSys AL2 Rq SSys Spec/Alloc SW Mgmt & Env SW Rqmts SSys Des SW Mgmt & Env SW Rqmts Software EC 2.n SW Design SW C&UT SW Rqmts SW Design SW C&UT SW I&T CMN SW I&T Env MN SW Int SW Design SW C&UT SW I&T T Env VNT SW Int VNT SW Test CMN SW CMN SW Int CMN SW Test VNT SW I&T Env SW I&T CMN SW I&T Env CMN SW Int CMN SW Test SSy Software IntegrationVNT SW I&T Env EC 2.n-1 VNT SW Int VNT SW Test CMN SW Test VNT SW I&T Env VNT SW Int VNT SW Test 4 months October 29, 2007 © 2007 BAE Systems Land & Armaments L.P. 9 Idealized Systems and Software Integrated Build Life Cycle Start DI 2.1 S/SS Year 1 SW By adding Systems & Subsystems requirements and design engineering stages synchronized with Software development increments, upstream work products can be worked iteratively to support Software. SIT SyLCO DI 2.2 S/SS Consider introduction of Systems LCO and DI 2.3 LCA events. SwLCO (ACR) SW SIT S/SS SW DI b.c IOC LCA LCO S/SS SI SIT SW SwLCA SwLCO SyLCA SyLCO TRR DI 2.4 Acronyms and Abbreviations Development Increment, Build b, Increment c Initial Operational Capability Life Cycle Architecture Milestone Life Cycle Objectives Milestone Systems/Subsystems (IPT) Engineering Software Item Software Subsys/Sys Integration & Test Software Engineering (incl. SI-Level Int & Test) Software LCA Software LCO Systems/Subsystems LCA Systems/Subsystems LCO Test Readiness Review (SI Level) October 29, 2007 S/SS DI 2.5 Inception Phase SIT SyLCA Need to keep as separate Sys/SW events to coordinate workflows? Year 2 SwLCA (DCR) SW SIT Software can provide valuable implementation feedback to Systems & Subsystems teams, with preplanned learning adjustments. S/SS SW SIT S/SS SW DI 2.6 Construction Phase SIT TRRs DI 2.7 S/SS DI 2.8 © 2007 BAE Systems Land & Armaments L.P. Elaboration Phase SW SIT S/SS SW IOC SIT Transition Phase 10 Systems Engineering Support to LCO (ACR) LCO viewed by MGV Systems Engineering as a “Software Event,” but SE provided required support. Created internal RBR (Requirements Baseline Review) milestone to have the LCO equivalent for Systems within MGV. Created mappings from MGV System Use Cases to MGV Software Capabilities, and scheduled Systems, Software, and Integration Thread capabilities across all Development Iterations. What level of completion of Systems Engineering work products is needed for Software to have a successful LCO? Recent RBR Experience: All requirements allocated to software were specified at a broad level, with traceability to upper-level models. • Challenge was leveling MGV IPTs to consistent levels of detail, and gaining consistent support across every team • Factored out common use cases to ensure consistent requirements Some systems requirements were developed in greater depth • Vehicle start-up, shut-down, mode changes Other documents were made available as reference (works in progress) which will be matured leading up to System PDR: • System Architecture which may include applicable DDM's, such as Vehicle Electronics Architecture, • • • • • October 29, 2007 System/Subsystem Schematics, Specialty Engineering Reports (Security, HFE/MANPRINT, Safety, Maintainability), Sustainment and Training Design Concepts Thread based performance analyses Current Version of Common Subsystem and Variant Level S/SDD's Vehicle External ICD’s - Interfaces to all C4ISR, Sustainment, and Training supplied subsystems. Vehicle Internal ICD’s – Interfaces to MGV Subsystems HW Configuration Item Specifications (HW CIDS - as available) © 2007 BAE Systems Land & Armaments L.P. 11 Feasibility Rationale - Key Point: Need to Show Evidence B. Boehm, “A Case for Anchor Point Milestones and Feasibility Rationales,” April 2005 Not just traceability matrices and PowerPoint charts Evidence can include results of • • • • • • • Prototypes: networks, robots, user interfaces, COTS interoperability Benchmarks: performance, scalability, accuracy Exercises: mission performance, interoperability, security Models: cost, schedule, performance, reliability; tradeoffs Simulations: mission scalability, performance, reliability Early working versions: infrastructure, data fusion, legacy compatibility Combinations of the above Validated by independent experts • • • • October 29, 2007 Realism of assumptions Representativeness of scenarios Thoroughness of analysis Coverage of key off-nominal conditions © 2007 BAE Systems Land & Armaments L.P. 12 Feasibility Analysis Planning Workshop Review Build Objectives, incl. Risks, Concerns, Issues Develop candidate software prototypes and integration threads by IPT Software Team using provided template Integrate results Template with example provided to teams: Concern (significant LCO or LCA Risk, Rqmt, Arch) Handling Approach, Value (Prototype, Thread, Analysis) Plan (EC, Description, ROM Cost, Product) Sensor/Weapon I/F Integration Prototype Interface with SDM, Exercise Preliminary Thread, High Value prior to LCA EC 2.3, Prototype sensor/weapon slew I/F, 80 hrs, SADD & FR input. Integrate with SDM LoFi in lab. … Result Summary: • 10 Subsystem and Vehicle Software Teams developed a total of 36 candidate feasibility analyses Next steps are to expand participants, prioritize, isolate common concerns, plan and schedule analyses Need to find ways to increase involvement of Systems Engineering and External SOS groups in Feasibility Analysis October 29, 2007 © 2007 BAE Systems Land & Armaments L.P. 13 Span of Feasibility Concerns Notional SOS Project Summary of Initial Results Org, 3, 8% MGV MGV, 12, 33% SOS, 21, 59% SOS Organization Contract/Agreement Relationship Example: 2 Organizations SOS Span Distance = 5 SOS Organizations = 6 • Most concerns are about the feasibility of interfacing to and using products produced across SOS organizations. • Communications burden and contractual boundaries rapidly expand as the number of SOS organizations increases. October 29, 2007 © 2007 BAE Systems Land & Armaments L.P. Span Distance Within Org = 0 MGV = 1 or 2 SOS = 2 to 5+ 14 SOS Challenges Synchronizing development to similar iterations across many organizations, each with its own development processes, is extremely difficult. • Unsynchronized workflows increase the number “surprise” requirements changes Greater numbers of participants tend to stretch out the iterations to make the same amount of progress. • Takes time to discover who to coordinate with, and if they have tasking to do so • Organizations change, people move around, responsibilities shift, POC’s change • Potential O(2) effect with more interactions and interfaces Interface and scope negotiation across many associate teams adds activities and stretches iterations. • To make progress, teams are forced to make unilateral decisions • From SOS perspective, may be sub-optimal or not even viable system-wide Coordinating compatible requirements to achieve viable functioning threads across associate contractors for small increments is difficult. Incremental SOS use of tactical software can be prohibitively expensive in terms of numbers of computers, simulators, emulators, and plant models needed prior to long-lead time dedicated hardware being available. Intellectual property protection, licensing, and legal involvement increases. October 29, 2007 © 2007 BAE Systems Land & Armaments L.P. 15