11
Global Journal of Flexible Systems Management 2009, Vol. 10, No. 4, pp 45-54 Impact of Identified Causal Factors to "System of Systems" Integration Complexity from a Defense Industry Perspective Jessica Jovel Graduate Student Stevens Institute of Technology, Email: [email protected] Rashmi Jain Associate Professor Stevens Institute of Technology. Email: [email protected] Abstract Multi-mission projects such as Mis.sile Defense Agency's (MDA) Ballistic Missile Defense (BMD) program involve .systems with unique capabilities developed by different companies to communicate together. While .çy.ç/cm integration for an independent .•system is a challenge in itself, integration for system of systems is a huge complex effort in part due to interoperability challenges. The focus of this paper is research, analyze, and prioritize some of the causal factors of system integration complexity, such as interoperability, for complex .systems such as BMD. Complex system integration is used interchangeably with system of systems integration in this paper because currently in the defense industry, system of systems integration is the new complex system integration. From the defense industry perspective, an attempt is made to contribute in developing a cause and effect relationship model between system requirements, system architecture (specifically open system architecture), and system of .systems integration process complexity. Keywords: complexity, systems engineering, system of systems, systems architecture, systems integration. Introduction In the defense industry, there are numerous systems that are composed of subsystems that are developed by different cotnpanies. For example, in the Ballistic Missile Defense (BMD) program, a ground-based radar developed by Raytheon needs to communicate with a ship-based radar developed by Lockheed Martin about the status of target objects such as missiles or enemy aircrafts. Although this is a large-scale type system of systems integration, smaller- scale integration efforts involve companies subcontracting different software or hardware elements and become the |irime integrators of these elements. Carney and Obemdorf acknowledge in their presentation "Integration and Interoperability Models for Systems of Systems" that the current trend in the technological industry is new systems are mostly expected to cooperate with other systems. This cooperation, which they term interoperability, is one of the challenges systems integration face. The continuous demand for systems to work with other systems makes "system of systems" "uncertainly unbounded", a concept defined by Dr. Kaplan. Uncertainly unbounded is when the customer or system contractor does not know when the system will stop growing making it difficult to have control during pre-system integration. Not knowing when the network of systems will stop growing or changing makes it difficult to design an interoperable network of systems. Also, not knowing when the system will be fixed from expanding makes it difficult to design the system architecture to reduce system integration complexity. This is an issue that defen.se systems have faced for the past 20 years. Defense system acquisitions made by the government are now based on the ability of a system to be affordable, dynamic, and support multi-mission modes. Currently in defense products there are many legacy systems that are in need of upgrades but use unique infrastructures that make it a challenge to upgrade (Hanratty, Lightsey & Larson 1999). A change that many defense contractors have embraced is the approach to system design. Instead of creating unique and closed system architecture, an open architecture approach and using commercial-off-the-shelf (COTS) parts have become part of system architecture design (Hanratty, Lightsey & Larson, 1999). Using the open systems approach when developing a new system or an upgrade of legacy products reduces system integration complexity in the future for the system. In 1999 an article about open system approach to weapon systems was written by Hanratty et. al. According to the authors, using an open systems approach to system architecture/design lessens the risk of the product becoming obsolete by making room for upgrade feasibility. The development of defense related complex systems may take several years because of the politics, disagreements, changing requirements, etc, (Hanratty, Lightsey & Larson 1999). With technology ever evolving, the risk of a technical © 2009, Global Institute of Flexible Systems Management

Impact of Identified Causal Factors to System of Systems ...jainra/Research_Papers/R4.pdf.•system is a challenge in itself, integration for system of systems is a huge complex effort

  • Upload
    others

  • View
    10

  • Download
    0

Embed Size (px)

Citation preview

  • Global Journal ofFlexible Systems Management2009, Vol. 10, No. 4, pp 45-54

    Impact of Identified Causal Factors to "System of Systems"Integration Complexity from a Defense Industry Perspective

    Jessica JovelGraduate Student

    Stevens Institute of Technology, Email: [email protected]

    Rashmi JainAssociate Professor

    Stevens Institute of Technology. Email: [email protected]

    Abstract

    Multi-mission projects such as Mis.sile Defense Agency's (MDA) Ballistic Missile Defense (BMD) program involve .systems withunique capabilities developed by different companies to communicate together. While .çy.ç/cm integration for an independent.•system is a challenge in itself, integration for system of systems is a huge complex effort in part due to interoperability challenges.The focus of this paper is research, analyze, and prioritize some of the causal factors of system integration complexity, such asinteroperability, for complex .systems such as BMD.

    Complex system integration is used interchangeably with system of systems integration in this paper because currently in thedefense industry, system of systems integration is the new complex system integration. From the defense industry perspective, anattempt is made to contribute in developing a cause and effect relationship model between system requirements, system architecture(specifically open system architecture), and system of .systems integration process complexity.

    Keywords: complexity, systems engineering, system of systems, systems architecture, systems integration.

    Introduction

    In the defense industry, there are numerous systems that arecomposed of subsystems that are developed by differentcotnpanies. For example, in the Ballistic Missile Defense(BMD) program, a ground-based radar developed byRaytheon needs to communicate with a ship-based radardeveloped by Lockheed Martin about the status of targetobjects such as missiles or enemy aircrafts. Although this isa large-scale type system of systems integration, smaller-scale integration efforts involve companies subcontractingdifferent software or hardware elements and become the|irime integrators of these elements.

    Carney and Obemdorf acknowledge in their presentation"Integration and Interoperability Models for Systems ofSystems" that the current trend in the technological industryis new systems are mostly expected to cooperate with othersystems. This cooperation, which they term interoperability,is one of the challenges systems integration face. Thecontinuous demand for systems to work with other systemsmakes "system of systems" "uncertainly unbounded", aconcept defined by Dr. Kaplan. Uncertainly unbounded iswhen the customer or system contractor does not know whenthe system will stop growing making it difficult to havecontrol during pre-system integration. Not knowing when thenetwork of systems will stop growing or changing makes itdifficult to design an interoperable network of systems. Also,not knowing when the system will be fixed from expanding

    makes it difficult to design the system architecture to reducesystem integration complexity. This is an issue that defen.sesystems have faced for the past 20 years.

    Defense system acquisitions made by the government arenow based on the ability of a system to be affordable,dynamic, and support multi-mission modes. Currently indefense products there are many legacy systems that are inneed of upgrades but use unique infrastructures that makeit a challenge to upgrade (Hanratty, Lightsey & Larson1999). A change that many defense contractors haveembraced is the approach to system design. Instead ofcreating unique and closed system architecture, an openarchitecture approach and using commercial-off-the-shelf(COTS) parts have become part of system architecture design(Hanratty, Lightsey & Larson, 1999). Using the open systemsapproach when developing a new system or an upgrade oflegacy products reduces system integration complexity inthe future for the system.

    In 1999 an article about open system approach toweapon systems was written by Hanratty et. al. Accordingto the authors, using an open systems approach to systemarchitecture/design lessens the risk of the product becomingobsolete by making room for upgrade feasibility. Thedevelopment of defense related complex systems may takeseveral years because of the politics, disagreements,changing requirements, etc, (Hanratty, Lightsey & Larson1999). With technology ever evolving, the risk of a technical

    © 2009, Global Institute ofFlexible Systems Management

  • 46 Jessica jovel and Rashmi jain

    product becoming obsolete after the concept has beendesigned is high in defense because by the time a system isin its production phase new technology has been out foryears. When a system has been designed to be an opensystem, the possibility of easily adding functionality orperformance upgrades is high and at a more affordable costwithout having to start from scratch; in addition the systemlife-cycle is extended. i

    One of the challenges that face open architecture isgrasping where in the system it makes most sense to useCOTS, use subcontractors for parts, and where some uniqueparts should be used in order to decrease system integrationcomplexity (Hanratty. Lightsey & Larson 1999). Systemattributes already defined should then be used to gainknowledge and understanding of how to design the systemarchitecture to become an open architecture design and resultin decreased system integration complexity. Figure 1 showsthe relationship between system requirements, andarchitecture design for legacy systemupgrades. It was developed to try tomake system integration easier.

    Identify rapidly changingtechnologies applicable tothe system

    Identify subsystems that arelikely to grow and evolveover the course of thesystem's life

    Identify high life cycle costdrivers

    Establish a listof criticalinterfaces

    iIdentify openstandards forall criticalinterfaces

    JSystem Requirements Design synthesis and architecture

    development

    Almost 10 years later, systemintegration for legacy systems still face challenges. Furtherresearch has been made into the cause and effect relationshipof system integration complexity. The system integrationprocess complexity is the "interaction between degree offeasibility and level of effort required to understand,describe, implement, manage and document the systemintegration process for a given system development andoperation environment" (Jain, et al.. 2008). Systemrequirements and system architecture have been investigatedto determine which have most impact on system integrationcomplexity. Concepts from the earlier research by the secondauthor (Jain et al. 2(X)8) are used for this research. This papersummarises our findings based on the application andextension of the earlier research. A difference however issystem architecture is considered for both hardware andsoftware systes and is studied from mostly system engineer'sperspective instead of a principally software standpoint.

    The application of the framework (Jain et al. 2(X)8) inthis research is toward defense systems, which as previouslymentioned involve system of systems. The survey used inthe earlier research was used as part of this paper's researchmethodology to capture the perspective of defense industryengineers. The survey was given to lead/chief engineers inthe defense industry who work in

    areas

    Current trend in the technologicalindustry is new systems are mostlyexpected to cooperate with other systems.

    Figure 1 : Relationship between System Requirements andSystem Architecture for Legacy System Upgrades (Hanratty,I Ligbtsey & Larson 1999)

    Causal Factors Investigated

    Exisitng literature indicates that one of the causal factorsthat add to system integration complexity is the continuous

    change of software or hardwareelements within the system such asupgrades and additionallunctionality. As mentionedpreviously, currently many projects

    are becoming "system of systems" that are "uncertainlyunbounded". There is no determination if the system willbe expected to support other missions in the future orcompletely change its purpose altogether. As a result,systems need to be backward compatible to existing productsand forward compatible to emerging technology.

    Camey and Obemdorf propose in their presentation thatsynchronicity of the system may help interoperability.Synchronicity allows all parties (customers, developers) tosee the rates at which subsystems undergo change such asupdates, repair, and modifications in order to keep the wholesystem synchronized. With synchronicity of all elements,interoperability between individual systems is more feasible.However synchronicity requires extensive documentation toshow the growth of the modified system. The level ofdocumentation of synchronicity may add to systemintegration complexity becau.se of the effort needed.

    During software/hardware subsystem contract awards,vendors are selected based on price, product turnaround,availability, and product technology features. Many vendorsclaim open architecture capability however not all productsare completely based on open architecture which many

    times is manifested during systemdifferent engineering areas such as ^^ open systems approach tO system nnegration and verification &software, hardware, and systems io architecture/design lessens the risk of validation. One problem is thatacquire their perspectives on factors the product becoming obsolete by vendors are not informed of otherthey think most impact system making room for upgrade feasibility. vendor's products or applicarionintegration complexity. The responseshelped produce a cause and effect relationship betweensystem requirements, system architecture and systemintegration complexity. The research methodology alsoinvolved reviewing exisiting literature on system of systemsintegration practices & from personal experience of the authors

    information. When the primecontractor uses producís from two different vendors, theproblem lies in either interoperability issues between thedifferent vendors or with the prime contractor's subsystems.This interoperability between three systems in this case addsanother layer to system integration complexity.

    in system integration & test. One possibility to overcome this issue is information

  • impact of identified Causal Factors to "System of Systems" Integration Complexity from a Defense Industry... 47

    sharing between vendors. During the design phase thesevendors could communicate the types of standards usedwithin each element in their product (since these vendorsin turn may sub-contract some parts or materials), designintent for their applications, schematics or high levelinformation on coding scheme. A team of system integratorswithin the prime contractor may be needed to keep eachvendor aware of some issues that may arise in their productwhen integrating it to other vendor products. Knowledgeof this information during the design phase infiuences thesystem integration to run more smoothly. The desire totnaintain a competitive advantage decreases vendors' interestin information sharing,making it difficult for .1system to achieve interfaceopenness. This causal factormay be difficult to overcome.

    During the requiremetiisphase in defining stakeholderneeds, focus is given to the system's current capabilities thatwill define the performance and constraints (Buede 2000).The National Society of Professional Engineers modelconsiders the requirements phase to be a part of the conceptdevelopment phase (Kossiakoff & Sweet 2003). The conceptdevelopment phase contains needs analysis, conceptexploration, and concept definition. None of these phasescontain future system capabilities as a task. As a result notmuch discussion is centered on the future capabilities of thesystem. One of the reasons may be because it may causemore constraints or performance requirements that the primesystem integrator may not desire.

    When a system is uncertainly unbounded it isunderstandable that not much focus is placed on futurecapability of the system. The capability is unknown.However, the minimal communication on future capabilitiesof a system may cause the interoperability challenges in thefuture during system integration. Currently requirements arebeing written in such a way that openness is achievedwhether by commonality or

    and littoral operations.

    The system was designed to be open-architecture and useCOTS modules including receiver and digital signalprocessor. If it's a mix (customer, business developmentleaders, and engineers) then perhaps all parties can think ofpotential uses for a system; such as the case in recent newsabout a falling satellite being destroyed by a missile defenseinterceptor designed for missile defense. It was converted tohave anti-satellite capability. Instead of starting from scratchthe missile interceptor was reprogramtned to complete a taskit was not tested for.

    When a system has been designed to be an opensystem, the possibility of easily adding functionalityor performance upgrades is high and at a moreaffordable cost without having to start from scratch;in addition the system life-cycle is extended. Ä

    It is difficult to fully knowthe environment in whichuncertain unbounded systemswill function in order to drive the system architecture. Dothe customers give this information or the lead engineer ofthe system with business developtnent leaders or a mix ofboth? In the case of the customer, he/she may have a limitedview on the system since they are focused on satisfying aneed whether it's large-scale or a simple task. On the otherhand, business development leaders along with the lead/chief engineers tnay have a bigger picture of future uses forsystetn therefore pitching to customers a flexible system thatcould be used for different missions. For example the SolidState S-band radar (S4R) system developed at LockheedMartin is a system designed to be scalable to supportmultiple missions, including air surveillance, cruise missiledefense, ballistic rnissile defense, counter target acquisition

    factors that add to systemintegration complexity is the continuous change ofsoftware or hardware elements within the systemsuch as upgrades and additional functionality.

    There are some instanceswhen systems are beingenhanced to support morecomplex missions without riskanalysis. Of course anytimenew software is added thereare potential conflicts withexisting code. One way to

    work around this is install software patches to the systemfor issues that are found. These patches are somewhat abridge between both codes. However, we run into theproblem where other parts of the system may be using someof the patch code only to find that it does not cooperatewith a subsystem code. These types of issues are seen overand over again in current systems that are becoming systemof systems. Either developing an abstract system architectureor following standards when designing the systemarchitecture helps this issue. For example, Prasad and Rossakdefine four types of integration architectures (Rossak &Prasad 1991). These types of architectures can be given asguidelines for type of system architecture and flowed downto vendors as which type to use. One drawback is theinfiexibility of having to use a defined type; however witha system of systems this would provide huge benefits fromreducing system integration cotrtplexity. Integrating theAbstract Design Paradigm introduced by Dogru et. al. to the

    systems architecture/designphase is one way of reducinginteroperability issues. In thisparadigm, the design isviewed from an integrationperspective rather than inphases (Dogru, Delcambr,

    Bayrak, Christiansen & Tanik 1992),

    Documentation for legacy systems is not always detailed,coherent, or even complete. Proper documentation isnecessary to flow down the technical infonnation to newengineers, technicians, etc...trying to understand the existingtechnology to cooperate with the new one. Many engineerswho worked on the legacy system go onto new projects,different companies, without leaving their knowledge indocumentation form. The cycle can continue to be on-goinguntil proper documentation guidelines and rules arefollowed. Other losses resulting from engineers who left anddid not document their design understand the intent of theirwork. Why did they do it that way and not this way? What

    © 2009, Global Institute ofFlexible Systems Management

  • 48 Jessica Jovel and Rashmi Jain

    was the intent behind that? Knowing these details helpbetter design the updated system architecture to integratewith the existing. Interoperability, understanding of subsystemsand documentation are all factors during integration.

    A summary of the causal

    focuses on causal factors that may have an impact tointegration and test. A brief description of such factorsis given below.

    1. Modularity according to Kossiakoff is a measure of thedegree of mutual independence of

    factors described so far as well as Jfie level of documentation of synchronicity the individual system components,possible solutions are listed in ffmy ddd ¡Q system integration complexity According to Kossiakoff a systemTable 1. A survey was given to five /.„^^„„p nf thp effort npeded should achieve a high degree ofengineers to assess the system •^ "'•' ' modularity such as makingintegration level of effort they perceive each solution to each interfaces and interactions as simple as possible for ea.secausal factor described in this paper will create. The averagesof engineers' responses on the survey were calculated foreach causal factor solution to get the resulting level of effortduring system integration. A high level of effort for systemintegration means an increase in system integrationcomplexity.

    Table 1 lists the phases of the system life-cycle eacharchitecture issue and solution impacts as well as engineers'responses to the level of effort needed for each solution.

    The purpose of the survey was to determine whetherpossible solutions that have been described to mitigate theeffect of causal factors to "system of system" integrationcomplexity add yet another layer of integration complexity.Our results show that vendor compatibility andinteroperability and defining a specific type of systemarchitecture (whether DoDAF, object-oriented, etc..) for allvendors to use does not involve a high level of effort. Bothof these elements are part of system architecture design,which gives a preliminary

    of system integration and test. Physical modularity intheory should make it easier (less complex duringsystem integration) to add system upgrades by beingable to replace a unit or add another unit onto theexisting system. However when different vendors areresponsible for different modules system integration hasthe potential of getting quite difficult in the vendor isunreliable or unknown (stemming from ftrst time beingused, unfamiliar technology)

    2. Commonality according to this earlier research is theease of ability to use a component or subsystem in morethan one place within the system. This categoryincludes physical familiarity such as percent of vendors,subcontractors known and hardware and softwaretechnology known, and operational commonality suchas number of unique skill codes required or estimatedoperational training time.

    3. Standards based system architecture is designed intclation to open standards. In

    assessment that system Proper documentation is necessary to flow down ^^^ standards based systemarchitecture does result in the technical information to new engineers, ¡tems such as interface,increasing or decreasing system technicians, etc...trying to understand the existing hardware, and softwareintegration complexity, technology to cooperate with the new one. standards are considered. Thewhichever case may apply. number of multiple vendors

    that manufacture the same product are considered aswell as multiple business domains the system will be

    System requirements, in the earlier research of the used for whether medical, commercial or defense. Alsosecond author, are analyzed as inputs to system consistency orientation determines the architecture suchintegration complexity. This research (Jain et al. 2008) as using common guidelines to implement fault

    Table 1 : System Ititegration Level of Effort

    System Integration Complexity Survey

    Isstie

    Not enough focus given to sy.stem futurecapabilities during concept development

    Interoperability of existing system to newtechnology caused by continuous change

    Vendor compatibility / interoperability

    Continuous system changes

    System Enhancements cause unforseenissues in other subsystems

    New engineers having unfamiliarity withlegacy systems

    Pos.sibleSolution

    Need business development leaders, cu.stomer,and lead engineers to focus on potentialfuture uses for system

    System design needs to be forward, present,and backward compatible

    Vendor information sharing

    Documentation for synchronization

    Defmed system architecture type to makeenhancements easier to add

    Thorough documentation of legacysystems(intent of design, lessonsleamed)

    Phase

    Requirements

    ArchitectureDesign

    ArchitectureDesign

    Integration

    ArchitectureDesign

    Integration

    Resulting Level ofEffort during System

    I ntegration (1-10)

    S

    10

    1

    7.5

    3

    10

  • Impact of Identified Causal Eactors to "System of Systems" Integration Complexity from a Defense Industry... 49

    Table 2: Survey Categoriesdetection and isolation.

    4. RMT embodies reliability, maintainability, andtestability. Lessons from Steven Institute systemsengineering courses (SYS605: System Integration) stressRMT helps extend the life-cycle of the system. RMThelps verification and validation by way of havingreliable equipment testing, help write test plans tothoroughly test reliability and maintainability againstrequirements.

    This earlier research (Jain et al. 2008) included a surveyquestionnaire that addressed the factors identifted as havina big impact on system integration complexity and qualityof veriflcation and validation (V&V) to form a cause andeffect relationship. Verification and validation are animportant part of the system integration process whichBuede terms "Integration and Qualification". Qualificationis the process of verifying and validating the system designand then obtaining the stakeholders' acceptance of thedesign (Buede 2000).

    The survey was modified for this research paper in twoways. First, a new grouping of the questions was created fromthe original survey into different categories such ascommonality, modularity, and requirements. The purpose ofthe grouped questions was to assess thoughts from engineersabout elements within each category which they thoughtwere the biggest contributors to system integrationcomplexity and V&V.

    There were 3 percentage choices for the survey in orderto assess engineers' perspective on each element's impact tosystem integration complexity. The 3 percentage choiceswere 75%. Thequestions that had more than >75% as ratings under eachcategory were then selected to be part of the second survey.Categories were broken down as shown in Table 2. Surveyresults showed that factors considered being most influentialcommonality, modularity, standards (hardware, software) andsystem requirement as shown in Table 2. Results showed that the second survey. They ranged from lead engineers toreliability, maintainability, and test had much some principal engineers in software, hardware and systeminfluence on integration and engineering. Results from thetest but not enough to increase Results show that vendor compatibility and survey are shown inthe complexity by too much. . . L-I-. J • r. • •/- ^ j- .

    ^ ^ ^ interoperability and defining a specific type of systemFor the first part of the architecture for all vendors to use does not involve

    ea system ^ ^ • ^ ^̂ ^̂ ^ /• gff^yf^ ß^^j^ ^/- these elements gives aengineers ,r . , . . *

    arniwprpH Fnr thp srnnH mrt preliminary assessment that system architecture doesanswered, hor the second part i- / _ J' _ _ number of components is aof the survey nine engineers result in increasing or decreasing system integration significant factor in systemtook the survey. Two third of complexity, whichever case may apply.the engineers are principal or

    Category

    Commonality

    Commonality in hardwareand software subsystems

    Operational commonality

    Physical familiarity

    Percentage of known vendorsand subcontractors

    % of familiar technology

    Modularity

    Physical modularity

    Functional modularity

    Orthogonality

    Abstraction of systemarchitecture

    Standards Based

    Interface Openness

    Open system orientation

    Consistency Orientation

    RMT

    Required level of reliability

    Required level ofmaintainability

    Testability

    Requirements

    Clear identification of thesystem requirements

    Prioritization of systemrequirements

    System requirementsfeasibility analysis

    System requirementsrisk analysis

    Categorization ofsystem requirements

    Importance to SystemIntegration

    75% importance to systemintegration complexity. Nine engineers were selected to take

    integration complexity.System requirements can be a

    factor in decreasing the number of components such as LRUsto control system integration level of effort. Even in openarchitecture systems, there are many unique elements for thesimple fact that vendors are subcontracted to develop somehardware or software components. Although a company mayuse vendors due to lowered costs or fast delivery turn-around instead of using in house resources, considerationshould be made as to the number of allowed unique

    © 2009, Global Institute ofFlexible Systems Management

  • 50 t 1

    Commonality

    • Physical Commonality(Within the system)• Hardware Commonality

    • Number of uniqueLRUs• Unique cables• Unique standards• Unique connectors

    • Software Commonality• Number of

    Languages• Number of

    Compilers• Number of unique

    Code implementation• Physical familiarity

    (from other systems)• % of vendors

    subcontractors known• % hardware and

    software technologyfamiliarity

    • Operational Commonality• Number of unique skill

    required• Estimated training time

    for maintenance• Estimated training time

    for operation

    ••• :i 1

    Jessica Jove

    Modularity

    • Physical modularity• Ease of system element

    upgrade• System rework• Modifying lines of code

    • Ease of operating systemupgrade

    • Functional modularize• Ease of adding new

    functionality• Orthogonality

    • Amount of functionalrequirement fragmentedacross multiple processingelements and interface

    • Amount of throughputrequirements acrossinterfaces

    • Abstraction• Does the system

    architecture provide andoption for informationhiding

    • Interfaces• Number of unique

    interfaces per systemelement

    • Number of differentnetworking protocols

    • Number of explicitimplicit interfaces

    / and Rashmi Jain

    Standards Based

    • Open Systems Orientation• Interface Standards

    •# of interface standards/*of interfaces

    • Multiple VendorsGreater than 5) exist forproducts based onstandards

    • Multiple businessdomains apply/usestandards (aerospace.medical.(telecommunications)

    • Hardware Standards•# of Form Factors•# of LRU• Multiple Vendors• Multiple Business

    Domain Apply/Use• Standard Maturity

    • Software Standards•# of proprietary & unique

    operating systems•# of non standard databases

    ware• Consistency Orientation

    • Common Guidelines forimplementing diagnosticsand performance

    RMT

    • Reliability• Fault Tolerance

    • % of missionfunctions with singlepoints of failure

    • % of safety criticalfunctions with singlepoints of failure

    • Critical Points ofDelicateness (SystemLoading• % Processor Loading• % Memory Loading• '7o Network Loading

    • Maintainability• Expected MTTR• Maximum Fault Group

    Size• Whether system

    operational undermaintenance

    • Accessibility• Testability

    •# of LRUs covered by BIT• Reproducibility of Errors• Online Testing• Automated

    Input/Simulation Insertion

    Figure 2: Architecture Assessment and Evaluation Parameters and Priorities (Verma 2000)

    components. This goes into play with the vendorinformation-sharing problem. Willimore vendors sharing their

    number of components used. Since there are more spares that„ . , ^ , , .^ . . . . . . . can be shared the turnaround timeRMT helps verification and validation by way ^^^ ^^^^^^ integration increases

    architecture, the system integrators of having reliable equipment testing, help and requires less troubleshootinghave more control and feedback u> write test plans to thoroughly test reliability during integration. This isall vendors to try and limit tiic ^nd maintainability against requirements.number of unique elements.

    Most survey takers made similar comments aboutcommonality in hardware and software making integrationeasier by 50-80% because when engineers are working withfamiliar items, efficiency is increased. The biggest concernwas compatibility issues that stem from integrating a highnumber of unique components within the system. Theprobability of discovering unfamiliar issues increases with

    ISystem Integration Complexity Impact

    assuming that the product is notdefective and integration issues

    have already been overcome. Some said that integrationeffort is reduced in relationship by the number of uniquepieces of equipment. However, there needs to be a definedreference point before determining how much better havingcommonality is to system integration. One additional detailto include in commonality of hardware is the productionequipment involved in assembly and test of subsystems preand post system integration. Common-off-the-shelf (COTS)

    System Verification & Validation Impact

    Figure 3: System Integration Complexity Survey Results Figure 4 : System Verification and Validation Results

  • Impact of Identified Causal Eactors to "System of Systems" Integration Complexity from a Defense Industry...

    change due to changing environmental regulation such asRestriction of the Use of Certain Hazardous Substances(RoHS): therefore testing tools and products vary. However,the methodology, analysis and processes to implement themodified COTS or assembly will not. One survey takercommented that the commonality between these two(equipment and testine

    The biggest concern was compatibility issuesthat stem from integrating a high numberof unique components within the system.

    51

    methodologies) is what makessystem integration less complex.

    The response to whetherlower percentage of known vendorsand subcontractors lessen system integration complexity was80% no meaning it is best to use a higher percentage ofknown vendors. One of the reasons was integration becomesdifficult when percentage of known vendors decrease.Unknown issues can arise from electronic parts or softwarepackages. The less percentage of known vendors the greater

    Table 3: System Integration ComplexitySurvey Scaled Responses

    Sy.stcni IntegrationComplexity

    ConmioiKilitN 111 I IW

    and S W subsy.slcm.s

    Lower 9f> of knownvendors/subcontractors

    Increase in % oflamlliar technology

    lncrea.se in physicalmodularity

    Increase in functionalmodularity

    Increase in opensystem orientation

    t)ecrease in interfaceopenness

    Prioritizatlon ofsystem requirements

    .System requirementrisk analysis

    Scale

    C o m m o n a l i t y in I IWand SW subsystems

    Lower % of knownvendors/subcontractors

    Increase in % offamiliar technology

    Increase in physicalmodularity

    Increase in functionalmodularity

    Increase in opensystem orientation

    t)ecrease in interfaceopenness

    Prioritization ofsystem requirements

    System requirementrisk analysis

    No

    0

    7

    1

    1

    1

    0

    3

    ">

    1

    1

    0

    7

    1

    1

    1

    0

    3

    2

    1

    80%

    -)

    0

    4

    3

    1

    1

    0

    5

    10

    0

    10

    20

    \5

    2^

    5

    5

    0

    Total

    9

    9

    9

    9

    9

    9

    9

    9

    9

    Total

    4

    I

    4

    4

    4

    4

    3

    3

    3

    the likelihood of unknown trends, characteristics, reliabilitymethods and products used. Issues raised from usingunknown vendors again stems from the minimalcommunication and information given between primecontractor and vendor during product development. Vendorsgive the technical information after the product has been

    produced as even thesedocumentations can be insufficientor unintelligible. Choice of theright vendor is critical. Fewervendors lead to less competition

    causing less product innovation and higher prices forproducts. Both of these problems offset the benefits of fewervendors. In the current marketplace, benchmarking ofproducts (periodic verification of current markets for criticalproducts to verify the vendors/costs) by programs hasbecome the standard and funded by contracts for the bestprice/product. Product vision (with vendors) in the future iscritical for system integration to be effective.

    Physical and functional modularity botli in general wasconsidered by more than 75% of responses to have 50% ormore impact in lessening system integration complexity.Answers ranged from modularity making redesign simplerto making substitution of a failing module easier to do. Inhardware, when proper initial integration analysis is done,modularity allows this to be leveraged for future upgrades,future packages or cabinet improvement within heating,cooling, power, weight, and center of gravity limitations.With modular design controls comes simplified I/O,mounting, and installation. In functional modularity, thecapability to integrate one function at a time allows foreasier troubleshooting and debugging.

    Engineers that claimed a lower percentage of knownvendors help integration by 20% up to 80% believed thatwith a smaller number of known vendors, the leaming curvedecreases of what to expect. It also helps make integrationeasier to manage. The burden on the prime contractor forinter-subcontractor communication is reduced. The claimsare with the caveat that the vendors that are being used havereliable products. However, if the single vendor/subcontractoris not reliable, then having a fewer number of vendors andsub-contractors is a risk.

    The responses for increased percentage of familiartechnology used generally (more than 85% agreed) were thatit led to less system integration complexity. Many of theunknowns for the technological product decrease when usingfamiliar technology making it easier to integrate with othersubsystems. Engineers that said it made sj-stem integrationeasier by more than 80% commented that efficiency isincreased whenever working with familiar technology. Whenengineers become familiar with capabilities and limitationsof a technology, less time is spent on integrating.

    One remark about familiar technology not leading tobetter integration was due to the fact that good engineeringefforts and analysis is still needed to ensure the applicationof familiar technology such as COTS are integrated correctlyand within the performance capabilities of the system.

    © 2009, Global Institute ofFlexible Systems Management

  • 52 Jessica Jovel and Rashmi Jain

    Without diligent system integration team the interfaces,interconnections, mounting, and technical performance ofthe system will not work or communicate even with familiartechnology used.

    About 65% of engineers who took the survey agreed thata decrease in interface openness helped ease systemintegration by at least 20%. It is difficult to comply withdifferent components especially from different vendors; itrequires additional effort which increases time and cost.Ideally, the use of common standards.and interfaces across components and Prioritization of system requirementssubsystems developed by separate giygg ^ ^/^^^ understanding of the

    should reduce the incidence of

    lessons learned from solving integration hurdles with a setof interfaces can be applied when integrating the same typeof interface somewhere else in the system. The one engineerthat did not agree that a decrease in interface openness helpsSI mentioned that interfaces need to be well understoodbefore really assessing that this improves the systemintegration.

    In system requirements related questions about 40%agreed that system integration complexity decreases by more

    than 50%. Within systemrequirements, that prioritization

    system integration

    problems, when teams are working totechnical performance requirements and

    the same set of rules. However, this program requirements that are necessaryis not always the case. With a for successful system integration.decrease in interface openness, the

    sitntlar to Dr. Jatn s results m whtchagreed that system requirement

    prioritization made systemintegration easier by 50-80%.

    Table 4: System Verification andValidation Survey Scaled Responses

    System Verificationand Validation

    Commonality in HWand SW subsystems

    Lower % of knownvendors/subcontractors

    Increase in % offamiliar technology

    Increase in physicalmodularity

    Increase in functionalmodularity

    Increase in opensystem orientation

    Decrease in interfaceopenness

    Prioritization ofsystem requirements

    Systetn requirementrisk analysis

    Scale

    Commonality in HWand SW subsystems

    Lower % of knownvendors/subcontractors

    Increase in % offamiliar technology

    Increase in physicalmodularity

    Increase in functionalmodularity

    Increase in opensystem orientation

    Decrease in interfaceopenness

    Prioritization ofsystem requirements

    System requirementrisk analysis

    No

    2

    4

    3

    1

    1

    0

    2

    1

    2

    1

    2

    4

    3

    1

    1

    0

    2

    1

    2

    2

    0

    4

    0

    2

    0

    0

    2

    ->

    4

    20-50%

    0

    1

    4

    1

    1

    1

    2

    3

    •}

    3

    0

    3

    12

    3

    3

    3

    6

    9

    6

    50-80%

    7

    2

    0

    2

    4

    3

    3

    3

    2

    4

    28

    8

    0

    8

    16

    12

    12

    12

    8

    >80%

    0

    0

    2

    4

    3

    5

    1

    1

    1

    5

    0

    0

    III

    20

    15

    25

    5

    5

    5

    Total

    9

    9

    9

    y

    9

    9

    9

    9

    9

    Total

    3

    2

    3

    4

    4

    4

    3

    3

    3

    Prioritization of system requirements gives a clearunderstanding of the technical performance requirements andprogram requirements that are necessary for successfulsystem integration. All inputs from the technical experts,stakeholders and functional managers are key to ensure norequirements are overlooked or left out (this is tnostimportant during verification and validation). Prioritizationof system requirements ensures proper focus is applied tothe requirements (frotn highest priority to lesser priority) andwill result is effective verification and validation when theproper stakeholders are consulted. Engineers who did notagree that prioritization of requirements helped systemintegration said that stakeholder requirements do notnecessarily map well into what needs to take place in orderto get the system functional.

    Risk analysis of system requirement feasibility had highimpact to system integration as well; 90% of engineersagreed that it improved integration efforts by at least 20%especially if risk is based on previous products with similarcapability. Performing requirement risk analysis helps futuremodification of the product.

    One of the disagreements to the common census is dueto the fact that requirement risk analysis may have noimpact to system integration. It was thought that effectiveProgram Risk Management is more important than just riskanalysis of requirements always reduced complexity andmitigated problems before system integration began. It wasthought that identifying key risk areas helps plan theintegration effort to schedule appropriate amount of effortto higher risk functionality. This sense may improve thesystem integration but it will not facilitate the actualprocess. Identifying program weakness and deficiencies willalways result in an effective system integration effort. This,however, does not come with a cost as the resolution ofproblems will require additional efforts, schedule andproduct performance specifics identified at a level notcommonly addressed at the program level.

    The same questions were given for verification andvalidation since this is another part of system integration.

  • Impact of Identified Causal Factors to "System of Systems" Integration Complexity from a Defense Industry... 53

    Similarities between verification & validation and systemintegration complexity survey results were physicalmodularity and open system orientation had the highestpercent of making more than 80% impact. The use ofcommon hardware and software sub-system allow test plans

    validate the requirement.

    Cause and Effect Relationship

    Survey responses for system integration complexity andsystem verification and validation were weighted to find the

    and equipment from previous modules . . _ „ „i. w • causal factors with most impact. Theto be leveraged for verification and H is important to note that leveraging ^^^^^.^^^ ^^^ then averaged by the

    ff f b b l d l b i l dvalidation. Commonality helps make off of Subassembly modular build number of responses and rounded toverification and validation less successes makes the probability for nearest integer to help form a causecomplex for next generations of the success greater for VáíV. and effect relationship for systemsystem (or subsystems). For example,after the initial installation, testing of subsystems and thefully integrated system have been complete for the ftrst timeon a production build, lessons learned can be leveraged forihe next generations.

    Some argued that if system integration was complete andissues were resolved, V&V would not rely on commonalityof hardware and software subsystems. It was also arguedthat the number of unique parts does not lead to easiersystem validation, but it does lead to easier verificationbecause there are fewer unique items to verify. With morecommon hardware/software in a system, common methodsand techniques can be applied therefore the number oftmique methods that need to be proven or demonstrated areless.

    Although modularity in general was thought to be apositive effect on V&V (many of the comments as to why itwas helpful were the same as for system integration), therewas argument on this idea. The argument was that samefailure rates are shown in modular assembly as non-modularas.scmblies. It is important to note that leveraging off of sub-assembly modular build successes makes the probability forsuccess greater for V&V.

    Unlike in system integration a lower percentage ofknown vendors and subcontractors actually simplifiedverification and validation in this case. Control of productdevelopment with the correct vendor (having technicalexpertise and the knowledge base with key personnel) willtninimize schedule overruns, test/qualification risks and theintegration of the qualified equipment into the designedsystem. The correct vendor is crucial however to theassessment of a decrease in vendor having a positive impactto V&V. Only when the vendor offers reliable products arefewer errors are encountered during V&V. Fewer knownvendors may lead to better conditions if control of processesis in place.

    Increasing the percentage of "familiar technology" forV&V helps set up the process for success. Resources are notwasted on determining whether any bad results are a resultof the system being tested or due to unreliable vendortesting equipment.

    Prioritization of system requirements in V&V helps aidthe process by understanding the most importantrcquiretiients that need to be validated and verifted. Riskanalysis does not benefit verification and validation muchttnless the part of the risk is in the capability to verify/

    requirements, system architecture, andsystem integration complexity. The scale for answercategories in the survey are:

    No

    80%

    1

    2

    3

    4

    5

    The scaled response of system integration show that thefollowing had most impact (50-80%)

    • Commonality in HW and SW subsystems

    • Increase in % of familiar technology

    • Increase in physical modularity

    • Increase in functional modularity

    • Increase in open system orientation

    The scaled response of system integration show that thefollowing had most impact (50-80%)

    • Increase in physical modularity

    • Increase in functional modularity

    • Increase in open system orientation

    Elements shown in the cause and effect relationshipdiagram. Figure 5, are all practiced in system architecture.System requirements in this case did not affect systemintegration complexity much. If system requirementscontained constraints or a required level of commonality,standards, or modularity in system architecture, systemrequirements would indirectly have an affect on SI. For

    Commonality inHW and SWsubsystems

    \

    tncTMse tn opensystem

    orientation

    \

    tncTÉAse Inhincttonalmodularity

    \

    Incrtate in % offamiliar

    technology

    \

    /

    tncrus« tnphysical

    modularity

    /

    s,\ 8y«l«fnIntegritlon

    Comptutty

    ...

    Figure 5: Cause and Effect Relationship forSystem Integration Complexity

    © 2009, Global Institute ofFlexible Systems Management

  • 54 Jessica jovel and Rashmi Jain

    thatuse the mentioned elements to

    technology; also modularitysuch as physical and functionalmodularity, as well as open systems

    example, low levels of hardware andsoftware commonality in systemsarchitecture have negative impact tointegration complexity. System their system architecture approach orientation. Addressing these elementsrequirements could include constraints will help ease system integration lor the architecture greatly reduceson the number of unique hardware and ^ „ j verification and validation. -system of system integration efforts,software components within the system Awarding contracts to vendors that useas a way to decrease system complexity. the mentioned elements to their system architecture

    . , . .r. • , ,• 1 • approach will help ease system integration and verificationSimilarly in system venfication and validation, system ^^ ,•

    •',,•' . , . „. , and validation,architecture has the most impact as shown in Figure 6.

    References

    Buede D. The Engineering Design of Systems: Models and Methods. NewYork: Wiley Series in Systems Engineering, 2000.

    Camey D, and P Oberndorf. Integration and Interoperability Models forSystems of Systems. Systems and Software Technology Conference. 2004.

    Dogru A, S Delcambr. C Bayrak. M Christiansen and M Tanik. Thedevelopment of an integrated system design environment. ICSl, 1992:691-698.

    Elias G and Jain R. Exploring Attributes for Systems ArchitectureEvaluation. CSER, 2007.

    Hanratty M, R H Lightsey and A. G. Larson. Open Systems and TheSystem Engineering Process. Acquisition Review Quarterly, 1999: 47-60.

    Jain R., A. Chandrasekaran. G Elias and R Cloutier. Exploring the Impactof Systems Architecture and Systems Requirements on Systems IntegrationComplexity, IEEE Systems Journal, 2(2) June 2008.

    Kaplan J. Challenges and Approaches to System of Systems Engineering.ICAF. 2005.

    Kossiakoff A and W N Sweet. System Engineering: Principles andPractices. New York: John Wiley & Sons, Inc, 2003.

    Rossak W and S. Prasad. Integration Architectures - A Framework forSystems Integration Decisions. Proceedings of the IEEE InternationalConference on Systems, Man, Cybernetics. 1991. 545-550.

    Verma D. Application of Innovative Methods to Early SystemsEngineering Phases: Need Analysis and Requirements Defmition. ConceptEvaluation, and System Architecture Development. NSEBS. 2000.

    Increase infoundation almodutarlty

    ncreasosyst<

    orient«

    n openïmtfon

    \ s./

    Increase inphysicai

    modularity

    SystemIntegrationComptexity

    Figure 6; Cause and Effect Relationship for SystemVerification and Validation

    Summary

    The cause and effect relationship between systemarchitecture and system integration complexity indicatesthat the former is most significant in impacting systemintegration complexity and system verification andvalidation. The most important aspects of system architectureimpacting system integration complexity are: commonalitysuch as hardware & software commonality, a high percent

    Ms. Jessica Jovel recently completed her Masters of Systems Engineering at Stevens Institute of Technology. She has 4 years ofexperience in the defense industry working as a systems engineer. Her line of work has mostly been concentrated in systemintegration, verification & validation.

    Dr. Rashmi Jain is an Associate Professor of Systems Engineering at Stevens Institute of Technology. Dr. Jain has over 15 yearsof experience of working on Information Technology (IT) systems. Prior to joining Stevens she was with Accenture (formerlyknown as Andersen Consulting). Over the course of her career she has been involved in leading the implementation of large andcomplex systems engineering and integration projects. She has done invited lectures internationally at Keio University, and ShibauraInstitute of Technology. Japan, Overseas Chinese Institute of Technology (OCIT), Taiwan. Indian Institute of Technology - Delhietc. She is a visiting professor for System Architecture and Integration at Keio University. Her teaching and research interestsinclude systems integration, systems architecture and design, business process reengineering, and rapid systems engineering. Dr.Jain has authored several papers on these topics. She holds Ph.D. and M.S. degrees in Technology Management from StevensIn.stitute of Technology.

  • Copyright of Global Journal of Flexible Systems Management is the property of Global Institute of Flexible

    Systems Management and its content may not be copied or emailed to multiple sites or posted to a listserv

    without the copyright holder's express written permission. However, users may print, download, or email

    articles for individual use.