Intelligent Industrial Processes: Enabling Research

Intelligent Industrial Processes Enabling research challenges by Dependable Communication and Computation Evgeny Osipov Valeriy Vyatkin Preface Effici...

0 downloads 37 Views 4MB Size
Intelligent Industrial Processes: Enabling Research Challenges by Dependable Communication and Computation

Evgeny Osipov Valeriy Vyatkin

Luleå University of Technology Department of Computer Science, Electrical and Space Engineering Division of Computer Science

ISSN 1402-1536 ISBN 978-91-7583-205-0 (pdf) Luleå 2014 www.ltu.se

Intelligent Industrial Processes Enabling research challenges by Dependable Communication and Computation Evgeny  Osipov  Valeriy  Vyatkin

Preface Efficient   industrial   processes   are   a   key   enabling   factor   of   economic   and   social   growth  in  modern   societies. Several   initiatives   on   projecting   the   development   of   industrial   process   automation   in   the   future   were undertaken  by  different  research  groups,  industrial  forums  and  governmental  organizations. This   document   reflects   upon   the   major   industrial   trends,   socio-­economical   demands   as   well   as   long   term visions   of   the   future   technology.   The   reflection   focuses   on   scientific   challenges   in   the  areas  of   computation and   communication   systems,   which   need   to   be   addressed   in   order   to   achieve   the  desired   functionality   and foresee  intelligence  in  future  industrial  processes. The   document   consists   of   two   parts.   Part   I   introduces   concepts   and   categories   to   be   used   throughout   the text   and   presents   the   results   of   the   analysis   of   the   international   research   agendas.   Part   II   presents   the research  challenges  based  on  the  analysis  of  the  state-­of-­the-­art.

Part I. Industrial development 2030 and beyond Part  I  of  the  document  is  structured  as  follows.  Section  1.1  summarizes  the  concepts  and  categories  to  be used  throughout  the  document.  Section  2  presents  the  analysis  of  the  considered  research  agendas.

1. Concepts and Categories The   main   focus   of   the   present   study   is   future   industries.   The   definition   of   the   modern   industry   spans   way beyond   a   single   factory   focusing   on   manufacturing   of   a   particular   product.   In   the   context   of   this   report   by industry   it   is   understood   a   set   of   actors   along   with   the   business,   social   and   technological   relationships between  them  resulting  in  the  production  of  one  or  another  final  value. The   business,   social   and   technological   relationships   in   this   context   refer   to   the   relationships   across   the entire   value   chain   leading   to   the   production   of   a   certain   value   including   business-­level   contacts,   logistics, manufacturing,  services,    etc. The  creation  of  a  certain  value  happens  thus  through  a  chain  of  industrial  processes. The   traditional   definition   of   an   industrial   process   tells   that   the   industrial   process   is   a  systematic  series  of mechanical   or   chemical   operations   that   produce   or   manufacture   something.   Another   definition   of   a   similar kind   tells   that   the   industrial   processes  are   procedures   involving  chemical   or  mechanical  steps  to   aid  in  the manufacture  of  an  item  or  items,  usually  carried  out  on  a  very  large  scale. In   the   context   of   this   document   this   definition   is   extended   as   a   systematic   series   of   operations  on   different layers  of  business,  social  and  technological  relationships  leading  to    production    or  manufacturing  a  value.

1

Modern   industrial   processes   feature   a   great   deal   of   informatization,   automation   and   intelligence.   The meaning   of   the   process   automation   is   rather   straightforward   and   self-­explanatory.   In   different   industries   the automation   is   either   achieved   by   the   usage   of   automatically   operating   machines   and   robots,   equipped  with computer-­based   control   system.   The   new   concept   of   informatization   is   seen   in   the   wide-­spread   usage   of modern   information   and   communication   technologies   for   the   purposes   of   planning   and   monitoring   industrial processes,  without  necessarily  increased  degree  of  their  automation. By   intelligence   in   modern   industrial   processes  often  understood  an   ability   of   the   automation  system  to  adapt to   the  changing  environment   conditions  through  self-­configuration  and  re-­configuration  as  well  as  its  ability  to deduct   simple   logical   relationships   given   the   input   and   output   data.   In   Section   3   an   elaborated   definition   of intelligent  automation  is  presented.

2. IIP challenges in light of international development agendas for future of industrial processes During   the   last   five   years   the   different   international   consortia   including   representatives   from   the   world-­wide leading   academic   institutions,   industries   and   governments   have   discussed   the   plans   and   strategies   for   the long-­term   development   of   the   industry.   These   forums   resulted   in   several   developed   agendas   projecting   the industrial   development   beyond   2030.   This   document   is   based   on   the   analysis   of   the   following   strategic agendas: 1. 2.

3. 4. 5.

Securing   the   future   of   German   Manufacturing   Industry:   Recommendations   for   implementing   the strategic  initiative  INDUSTRIE  4.0.  Final  report  of  the  Industrie  4.0  Working  group.  April  2013. Swedish   production   2025   Strategic   research   and   innovation   agenda   to   meet   the   global   challenges. September   2011.   (Svensk   produktion   2025   Strategisk   forsknings-­   och   innovationsagenda   för   att möta  de  globala  utmaningarna.  Sept.  2011). Factories   of   the   future   PPP   strategic   multi-­annual   roadmap.   Prepared   by   the   Ad-­hoc   Industrial Advisory  Group.  2010. European  Roadmap  for  Industrial  Process  Automation.  ProcessIT.EU  2012. A  Roadmap  for  U.S.  Robotics  From  Internet  to  Robotics  2013  Edition.

Without   a   loss   of   generality  the  authors   conjecture   that  the  main  development   milestones,   requirements   and envisioned   challenges   are   similarly   formulated   in   all   different   strategic   agendas.   For   the   purposes   of   the discussion,   we   recapitulate   several   issues   from   selected   agendas   which   allows   to   put   the   requirements under  a  common  denominator. According   to   German   development   agenda   Industrie   4.0,   the   overall   goal   of   the   set   for   the   industry   is   to fundamentally   improve   the   industrial   processes   involved   in   manufacturing,   engineering,  supply   chain  and   the life   cycle   management.   The   main   driving   force   of   the   new   industrial   revolution   according  to  the  Industrie 4.0   vision   is   the   Internet   of   Things   (IoT)   and   the   Cyber   Physical   Systems   (CPS).   In   the   future, machinery,   warehousing   systems   and   production   facilities   will   be   incorporated   using   global   networks   and form  a   Cyber  Physical   System.   In   the   manufacturing  environment,  these  Cyber-­Physical  Systems  comprise smart   machines,   storage   systems   and   production   facilities   capable   of   autonomously   exchanging information,   triggering   actions   and   controlling   each   other   independently.   The   final   report   of   the   Industrie  4.0

2

working   group   defines   the   major   development   and   research   challenges   which   needed   to   be   addressed   to enable  the  projected  development. Another   notable   moment   is   stated   in   Swedish   production  2025  Strategic   research  and   innovation  agenda   by Vinnova.   This   document   recognizes   the   current   shift   of   production   paradigm   from   simple   production   of goods   to   all-­growing   role   of   services   connected   to   products   and   the   associated   quality   of   experience.   This paradigm   shift   requires   new   methods  to  shorten  the  production  cycle   and   capabilities   of   industries  for  highly customized   production.   The   key   property   characterizing   the   industry   of   the   future   is   the   flexibility   across the   entire   production   chain.   Vinnova   conjectures   that   new   more   resource   efficient   behavior   of   the consumers   will   also   give   a   start   to   completely   new   business   models,   which   in   its   turn   will   affect   the production   processes.   Vinnova’s   development   strategy   identifies   the   following   driving   forces   behind   the transformation   of   the   production   industry:   individualized   production,   resource-­smart   design   and   production as   well   as   accentuated  customer   usefulness.  The   report   identifies  the  research   and  development   challenges which   needs   to   be   addressed   to  meet   the   demands  of  the  driving   forces.   Another  aspect   of  the  development agenda   which   makes   it   outstanding   from   other   similar   documents   is   accentuating   the   importance   of development   of   new   work   forms   and   relationships   between   workers   and   the   industry.   In   particular   new advanced   Human-­Machine   interactions   techniques   should   be  proposed   as  well   as  novel   methods  for   efficient development  of  new  skills  should  be  in  place  in  order  to  achieve  the  goal.

Figure  1.    Intelligent  Industries  by  2030:  Demands  and  Driving  forces.

3

The   result   of   the   analysis   of  the  major   research   agendas  projecting   the   development  of  the  industries   beyond year   2030   is   summarized   in   Figure   1,   Figure   2   and   Figure   3.  It  is   a  common   understanding  that   the  current state   of   the   industrial   processes   could   be   characterized   as   globally   interdependent,   featuring   advanced automation   and   a   certain   level   of   intelligence.   To   this   moment   the   intelligence   could   be   defined   as “Engineered   intelligence”   involving   a   pre-­engineered   set   of   operations   of   the   automation   components   on different   levels,   which   allows   execution  of  alternative  operation  paths  leading  to   more   efficient  performance  of the   target   industrial   process.   Despite   of   the   truly   sophisticated   and   to   a   large   extend   fully   automated industrial   processes   all   leading   economies   univocally   conclude   that   in   the   current   state   the   industries cannot   meet   the   upcoming   societal   demands.   The   major   key   word   commonly   identifying  these   demands  is flexibility.   Flexibility   in   this   context   is   an   ability   of   an   industrial   process   to   adapt   accordingly   in   order   to meet   the   time   varying   demand   on  the  one  hand  and   a   capability  to  quick  reconfiguration  in  order   to   increase the  quality  of  the  final  product  and  to  improve  its  environmental  footprint. All  major  strategic  development  agendas  agree  on  the  following  driving  forces  for  the  future  industries: ● Flexible  production  and  manufacturing, ● Integrated  operation  of  the  entire  value  chain  for  a  particular  product;; ● Resource-­smart  design  and  production;;  and  finally ● New  work  forms  and  social  relationships. The   flexibility   can   be   achieved   through   yet   more   advanced   automation  and   intelligence   behind  the  operation and   management   of   the   industrial   processes.   It   is   a   common   understanding   that   the   major   technology which   will   advance   the   industrial   processes   is   virtualization   of   the   machines   and   their   components,   the environment   and   the   human   into   integrated   Cyber-­Physical   Systems.   While   recognizing   the   great importance   and   potential   impact   of   theiIndustrial   and   governmental   roadmaps   and  agendas,   the  authors  of this   document   would   like   to  make   a  remark   concerning  the  basic   definitions.   Concerned  with  future   trends, such   as   the   IoT   and  CPS   concepts,   most   of   the   documents  offer  very   imprecise  definitions  that   lead  to  mis-­ interpretation   and  can  bring   more   harm  than  value.  One   shall  distinguish   the   Cyber-­Physical  approach   to  the design   of   computer-­controlled   Systems,   from   the   vague   description   of   future   manufacturing   systems   as being   CPS.   Any   modern   manufacturing   system   is   already   a   cyber-­physical   system,   because   it   is “integrations   of   computation   with   physical   processes”.   It   already   “comprises   smart   machines,   storage systems   and   production   facilities   capable   of   autonomously   exchanging   information,   triggering   actions   and controlling   each   other   independently”.   Besides,   such   features   as   “autonomicity”   and   “independence   of operations”   cannot   be   considered   as   characteristic   of   CPS,   they   would   rather   belong   to   “Intelligent Manufacturing   Systems”.   On   the   other   hand,   adopting   the  “true   CPS”   view   and   design   approach  is   capable of  bringing  numerous  benefits  to  the  future  industries.

4

Figure  2.  Evolution  of  industrial  intelligence. Connected   to   the   last   remark   it   is   important   to   characterize   the   state   of   the   technology   in   the   period   in question   in   order   to   proceed   with   the   identification   of   the   academic   challenges.   We   also   intend   to crystallize   the   definition   of   “intelligence”   depending   on   the   stage   of   the   technology   development.   These definitions   are   also   summarizedss  in   Figure  3.  As  is   stated  above   by  the  present   time  the  industry   features a   great   deal   of   automation.   To   some   extend   the   current   level   of   automation   of   the   industrial   processes allows   to   talk   about   certain   level   of   intelligence.  The   intelligence   of   most   of   the   modern   industries   manifests itself   through   the   usage   of   expert   systems   for   performing   root-­cause   analysis  and   inferring   of   observation based   process  models;;   neural  networks  are   widely  used  for  classification  purposes;;  distributed  planning  and execution   of   alternative   operation   plans.   Since   all   of   the   methods   require   a   great   deal   of   engineering   by human  developers  we  will  call  such  intelligence  “Manual    intelligence”. Already   at   the   present   time   there   is   a   great   trend   of   adopting   service   oriented   architectures   for semi-­automatic   orchestration   of   distributed   functionality,   adoption   of   agent-­based   technology   for   increasing the   level   of   flexibility.   We   will   call   this   era   -­   the  era  of   “Automated  intelligence”.  In  the  opinion  of   the   authors and   according   to   the   reviewed   research   and   development   agendas   realistically   this   era   will   span   all   the way  until  2030  when  most  of  the  agenda  envision  full  adaptation  of  the  SOA  . Looking   beyond   2030,   the   level   of   technological   development   should   allow   yet   higher   degree   of   processes’ flexibility   including   fully  automatic  orchestration   and  consistency  verification,   runtime  automatic   re-­tasking   of machinery   and   control   system   as   well   as   industrial   machinery   learning   new   tasks   by   observation.   We   call this  level  of  intelligence  as  “Evolving  Intelligence”.

5

Figure  3.  Definition  of  Industrial  Intelligence  .

Part II. Detailed research roadmap in selected areas of the identified research challenges. The   minimalistic   definition   of   cyber-­physical   system   given   in   [6]   “(CPS)   are  integrations   of  computation  with physical   processes”   has   given   raise   to   the   synergetic   system   design   and   analysis   methodology   that considers   both   computing   and   physical   processes   as   two   integral   parts   of   one   system   with   complex interrelations   between  each  other   that   can   have   substantial   impact  on   their   operation  and   performance.  Most if   not   all   of   the   industrial   processes   are   certainly   cyber-­physical   already   (and   have   been   for   long   time),   but approach   to   their   design  has  not  been  much   following  the   CPS  agenda.  In  this  section   we  propose  a  number of   steps   to   bridge   this   gap   to   the   benefit   of   the   industry   based   on   the   analysis   of   the   current   trends.   The reasoning   line   of   this   chapter   adopts   the   top-­down   approach   starting   with   consideration   of   the   trends   and gaps   on   higher   (business   relationship)   abstraction   level   descending   to   the   developments   on   the   level   of particular  software  and  hardware  components  (sensors  and  actuators) At   the   highest   level   we   conjecture   that   industrial   automation   will   remain   to   be   the   user   (rather   than main   development   driver)   of   advanced   computing   platforms   initially   developed   for   other application   areas,   where   higher   investments   is   possible   due   to   high-­volume   production   opportunities including   military,   space,   consumer   electronics,   robotics,   automotive   electronics,   business   computing, gaming   (See   illustration   in   Figure   4).   The   authors   of   this   document   see   the   need   to   monitor   the progress   in   those   areas   and   timely   pick   and   apply   promising   technologies   in   the   industrial automation  context.

6

Figure  4.  Sources  of  inspiration  for    IIP  solutions. It  is  likely  to  see  the  use  in  industrial  automation  of  such  recent  inventions  as: ● mobile  devices  with  augmented  reality  (like  Google  glass);; ● modular  mobile  devices;; ● mobile  robots  (like  quadrocopters);; ● Gesture  steered  robots;; ● adaptable  wireless  networks  providing  acceptable  quality  of  service  in  noisy  environments;; ● precise  time  synchronization;; ● wireless   sensor   networks,   in   particular,   body   area   networks   that   can   be   used   as   a   part   of  workers uniform  in  human-­machine  smart  industrial  environments. ● dependable   networking   and   embedded   computation   technologies   initially   developed   for   automotive and  military  applications. ● modeling  of  user  behavior  in  different  social  groups.

2.1 Globally integrated industries Effectively   the   availability   of   globally   integrated   industries   by   2030   projected   by   the   main-­stream   research agendas   implies   enabling   the   access   to   information   on   all   levels   of   granularity   down   to   every   single intelligent  component  and  providing  the  infrastructure  for  the  efficient  management  of  huge  data  flows. As   a   matter   of   fact   the   developments   of   such   kind   are   currently   going   on   in   the   scope   of   Smart   Cities (although   multiple   references   exist   we   refer   to   a   policy   document   by   Euro   Commission  [8].)  In  the  scope  of this   document   Smart   Cities   are   referenced   not   only   due   to   a   similarity   of   technical   characteristics   and requirements   as   those   of   globally   integrated   industries,   but   also   due   to  availability  of  commercial   solutions from   leading   IT   vendors   such   as   CISCO   [9]   and   IBM   [10].   Another   relevant   activities   in   this   domain   are several  “City  Operating  Systems”  solutions,  including  those  from  Urbiotica  [11]  and  PlanIT  [12]. In   the   opinion   of   the   authors   these   and   similar   companies   will   shape   the   future   development   of   the computing   infrastructure   for   intelligent   industrial   processes.   Adopting   the   state-­of-­the-­art   techniques   from service   oriented   architectures   such   companies   will   eventually   standardize   the   SOA   interfaces   for   industrial interaction.

7

What   currently   is   not   fully   understood   and   it   is   where   the   universities   could   contribute   to   the   ongoing standardization   process   is   the   systematization   of   domain-­specific   raw   data   and   its   conceptualization.   The importance   of   this   is   proclaimed   by   Tim   Berners-­Lee,   the   inventor   of   the   world   wide  web  [13].   It   is  essential that   this   research   is   conducted   in   close   cooperation   with   the   industries   forming   parts   or   an   entire   supply chain.   This   is   mainly   to   bridge   the   cultural   gap  between  the   industries  and   the   world  of   information.   Although there   are   notable   examples   of   such   collaborative   projects   e.g.   [14]   there   is   a   clear   need   for   a   closer interaction  with  the  diverse  industrial  partners. Another   gap   on   this   level   of   “intelligence”   relates   to   data   acquisition   and   and   data   processing.   The   Cloud Computing  nowadays   is  already   mature  technology.   However,   so   far   the   virtualized  computing  infrastructures are   very   rarely   (if   not   at   all)  used   as  part   of   control   loops.  The   current  state  of  the  cloud  technology  could  be somewhat   compared   to   initial   versions   of   the   world-­wide   web   having   humans   as   final   consumers   of   the information.   By   connecting   the   “Intelligent   industries”   the   size   of   the   information   flows,   the   demand   on   the computational   power,   the   requirements   on   the   dependability   and   sustainability   of   virtualized   computing infrastructures   will   increase   dramatically.   In   the   following   sections   particular   gaps   in   the   areas   of   data acquisition  and  virtual  computing  infrastructures  are  discussed.

2.1.1 Data Acquisition The   clear   trend   is   widespread   penetration   of   wireless   sensors,   mobile   devices,   Cloud-­services   into   the factory   floor.   The   controllers   are getting   more   decentralized   and networked.   This   trend   will   certainly continue   in   the   years   to  come,   leading to   the   situation   when   all   sensors   and actuators   will   become   intelligent, network   connected   and   providing   rich semantic   data   rather   than   merely signals.   This   will   provide   seamless access   to   any   process   related information   (measurements)   and   its integration   with   process   models   in real-­time,   enabling   more   accurate control   and   reconfiguration   of processes.

The   authors   of   this   document   conjecture   that   the   new   forms   of   traffic   should   be   natively   supported   by operators   of   wireless   broadband   communications.   In   order   to   enable   this   it   is   of   primary   importance   to conduct   research   on   adaptation   of   LTE-­A   architecture   to   machine-­to-­machine   communications   in   general and   the   specifics   of   industrial   communications   in   particular.   On   a   high   level   of   abstraction  the  challenge  for the   developers   and   vendors  of   LTE   technology   is  to   enable   mobile  operators  as   the   first   tier   data  aggregation service   providers.   Specifically,   LTE   networks   should   natively   support   access   to   individual   machine components   either   by   including   technology-­specific   gateways   in   the   architecture   or   even   providing   direct broadband  access  to  the  individual  sensors.    A  set  of  particular  research  challenges  is  discussed  in  [21].

8

Figure   5.   Adopting   the   advances   in   across   application   domains   for   sustainable   automation   and including  the  quality  of  the  virtualized  services  as  another  dimension  of  the  optimization  task.

2.1.2 Virtualized computing infrastructures The   cross-­sectoral   gap   to   be   bridged   on   the   the   way   towards  enabling   mobile   operators   as   part   of   the   m2m data   processing   loop   is   the   historical   barrier   between   the   computing   and   telecommunication   industries. Traditionally   these   industries   followed   parallel   development   paths.   Today,   however,   the  borders  between  the computing   and   telecommunication   technologies   are   disappearing,   we   are   witnessing   a   convergences   of   the two   sectors.   Only   at   the   end   of   2010   large   mobile   operators   [22]   and   leading   vendors   of   telecom technologies   initiated   adoption   of   recent   trends   in   management   of   computing   infrastructure   for   optimizing own  infrastructures.  As  the   trend  is   still  new  there   is  a  great  demand  on   understanding   the   technological   and scientific   principles   for   future   converged   architectures.   Therefore   the   main   scientific   objective   before   2030 and   further   on   is   to   establish   and   foster   a   common   research   agenda   on  the   intersection  of   the   two  sectors. It   is   of   ultimate   importance   to   address   a   cultural   gap   between   different   branches   of   engineering   science contributing   collectively   to   the   development   of   large-­scale   distributed   computing   and   communication systems.   Traditionally   research  on   infrastructure   optimization,   quality  of  data  services  and   fault  management of   large-­scale   infrastructures   and   security   is   conducted  separately.   This  makes   the   resulting   solutions  often sub-­optimal   by   not   accounting   for   performance-­limiting   factors   introduced   by   the   adjacent   technologies. Referring   back   to   Figure   4   it   is   of   particular   importance   to   introduce   another   direction   into   the   optimization process.   The   automation   procedures   for   achieving   higher   degree   of   energy   efficiency   should  be   codesigned with   the   optimization   procedures   of   the   virtual   environments   which   have   an   objective   of   performance optimization  of  quality  of  virtual  services.

2.2 Flexible factories Flexibility   of   factories   in   general   and   their   production   floors   in   particular   should   be   enabled   already   at   the design   stage.   An   extensive   state-­of-­the-­art   survey   in   the   area   of   the   design   of   automation   of   industrial processes   is   in   [7].   The   most   important   impact   of   the   CPS   concept   is   cultural:   In   the   pre-­CPS   era,   there

9

always   has   been   a   substantial   cultural   gap   between   process   engineers   designing   the   physical   process  and designing  the  cyber  part  (control  and  automation). The   process   and   control   engineers   have   been   using   accurate   modeling   of   processes,   but   had   little understanding   of   computations   specifics   in   the   embedded   devices,   using   very   coarse   grain   computation models   (e.g.   PLCs).   On   the   other   hand,   embedded   systems   engineers   were   focusing   on   computational performance   and   dependability   of   embedded   controllers,   but   used   the   “environment”  abstraction   of   the   world outside  the  computer   that  lead   to   substantial   difficulties  in  achieving  the  required   system   properties,  such   as robustness. The   CPS   approach   implies   the   cross-­penetration   of   knowledge   and   design   approaches   between these,  previously  separated  domains. The  system  engineering  practices  of  industrial  automation  systems  experience  the  influence  from: -­              simulation  and  virtual  reality,  gaming -­              best  practices  in  software  engineering  domain  (UML,  SysML) -­              Internet  and  IoT

2.2.1 Simulation Modeling,   analysis,   and   simulation   are   essential   for   understanding   complex   systems  such   as   CPS.   This  is widely   recognized   by   Universities,   research  institutes   and  industrial  groups  around   the   world.  The   creation   of reliable   multi-­disciplinary   simulation   tools   that   can   be   used   to   support   the   entire   development   process   has been   identified   as   a   major   scientific   goal   in   several   research   roadmaps   and   agendas   for   the   coming   15 years.   The   research  efforts   in  this  domain   should  be  directed   towards   enabling  creation   of  simulation technology,   which   will   be   used   daily   throughout   the   engineering   life-­cycle   (e.g.,   research   and development,   marketing,   concept   study,   detailed   design,   testing,   operation,   product   updating,   problem solving,  maintenance,  operator  training). A   strong   evidence   of  the  cyber-­physical  approach   getting   recognized  and   adopted   in   industrial  automation  is the   growing   popularity   of   the   “simulation   in   the   loop”   approach   to  systems  validation.   This  approach  implies availability   of   accurate   simulate   models   of   the   uncontrolled  plant  behaviour   that  can  be   connected   to  control systems   through   open   interfaces   and   used   instead   of   the   real   plant   for   control   system   debugging   and performance  estimations. The   availability   of   the   simulation   models   as   “commodity”   will   be   a   great   enabler  of  this  approach,  leading  to higher   quality   designs   achieved   in   shorter   time   and   at   lower   costs.   It   may   also   lead   to   wider   application   of the   “parallel”   systems   approach,   when   a   simulation   model   of   the   plant   is   executed   in   parallel   with   its   real operation   and   can   be   used   for   finding   more   optimal   scenarios   of   operation   without   disturbing  the  production process.

2.2.2 IoT and data driven dynamic SOA The  impact  of  the  Internet  and  Internet  of  Things  on  the  engineering  process  is  seen  in  the  adoption  of Service-­oriented  Architecture  that  implies  design  of  systems  as  decentralized  nodes  offering  well-­defined service  interfaces.  IoT  is  seen  as  enabler  of  Intelligent  Product  concept,  of  higher  quality  and  traceability.

10

The  authors  of  this  document  remarks  that  a  Service-­Oriented  Architecture  by  itself  does  not  make  any process  more  flexible  if  the  underlying  process  is  not  flexible  by  itself.  A  high  degree  of  physical modularization  and  decentralization  in  control  procedures  should  be  developed  along  with  the definition  of  the  suitable  SOA  interfaces.

Figure  6.  The  latest  model  of  modular  mobile  telephone  by  Google.  This  analogy  is  used  as  an  example of  an  agility  of  a  complex  software-­hardware  system,  which  makes  its  reconfiguration  inherently  simple. Again   referring   to   projections   on   the   development   of   the   world   wide   web   by   its   creator   [13]   the   main challenge   is   facing   the   computing   systems   of   the   future   is   to   efficiently   derive   meaning   out   of   huge   amount of   raw   data.   In   the   case   of   Intelligent   Industrial   Processes   the   meaning   shall   be   defined  in  terms   of   models which   describe   complex   interactions   within   the   system   of   systems   of   different   scale.   In   this   respect classical   formulation   of   data   mining   as   finding   patterns   in   vast   amount   of   data   should   be   extended   by problems   of   inter-­relating   the   patterns   into   structures   suitable   for   automatic   taking   of   decisions.   During recent   years   the   advances   in   the   computing   platforms   and   communication   systems   boosted   the   interest   in applied  methods  of  artificial  intelligence  on  the  new  level. Artificial   intelligence   is   a   mature   area   of   science   and   engineering   with   multitude   of   methods   being   used   in practical   applications   [15].   In   the   context   of   the   IIP  three  major   classes   of   systems  are  of  particular   interest since   they   found   their   applications   for   management   of   processes   control   [16,   17,   18]   as   well   as   the management   of   large-­scale   telecommunication   networks   [19].   These   are   Rule-­based   systems,  Case-­based event  correlation,  and  Probabilistic  event  correlation. A   typical   rule-­based   system   contains   a   set   of   “IF-­THEN”   statements   of   diverse   hierarchy.   It   is   typically constructed   by   a   knowledge   engineer   to   cover   a   particular   domain,   e.g.   fault   management   for   a   particular plant.   An   obvious   drawback   of   this   approach   is   inflexibility   in   adapting   to   changing   conditions.   In management   domains   with   dynamically   changing   properties   such   systems   must   be   redesigned,   reflecting the  changes  in  the  new  set  of  rules. The   probabilistic   approach   performs   event   correlation   accounting   for   uncertainty   using   solid   mathematical foundation   of   Bayesian   reasoning.   One   of   the   main   criticisms   of   Bayesian   reasoning   is   the   difficulty   of coming  up  with  prior  probabilities  before  computation  begins. A   case-­based   (CB)   event   correlation   system   solves   a   new   problem   by   remembering   a   previous   similar experience   and   adapting   the   previous   solution   to   the   specifics   of   the   new   problem.   A   case-­based   event correlation   offers   several   distinct   operational   features.   Firstly,   a   case   in   CB  can  be   a   semantically  rich  data

11

structure,   thereby   making   it   ideal   for   management   tasks   dealing   with   complex   problems.   Secondly,   the case   adaptation   algorithm   used   in   CB   systems   makes   this   approach   more   suitable   for   solving   problems where   the   exact   solution   either   does   not   exist   or   is   too   costly.   Finally,   being   by   nature   a   learning   system, CB   allows   the   operational   behavior   of   the   event   correlation   process   to   be   improved   without   additional hard-­coding   [20].   The   main   technological   challenges,   which   so   far   prevented   the   usage   of   CB   methods   for temporal   reasoning,   are   problems   with   semantic   analysis   of   alarm   information   due  to  the  lack   of  a   common standard   for   semantics   of   alarm   representation   and   low   performance   of   the   current   event   correlation algorithms  for  performing  spatio-­temporal  analysis  on  masses  of  events. In   their   recent   publication   [20]   the   authors   overview   the   progress   of   the   event   correlation   techniques   and provide   a   set   of   recommendations   for   future   development   of   event   correlation   techniques   in   the   context   of system  management: 1. The  next  generation  event-­correlation  systems  must  be  able  to  deal  with  uncertain  knowledge. 2. Better  learning  techniques  to  improve  the  accuracy  of  case-­based  systems 3. Faster  algorithms  based  on  binary  vector  mapping  which  would  convert  a  problem  of  correlating spatiotemporal  events  from  complex  cross-­matching  of  “IF-­THEN”  rules  into  binary  vector  mapping operations.

2.3 Security is the key to IIP emergence Traditionally   the   importance   of   security   is   often   underestimated   both   by   the   operators   and   research   groups focusing   on   optimization   of   functional   block.   As   a   result   it   is   treated   as   a   costly   add-­on   rather   than   a mandatory   property.   The   security   challenges   of   modern   ICT   systems   are   complex   and   should   take   into account   complex   relationships  between   people,  data,  applications   and  the   infrastructure.  In   order  to  address these   challenges   radically   new   integrated   security   solutions   are   needed.   The   main   shift   of   paradigm   in designing   security   components   should   be   to   divert   from   “reactive”   approach   towards   more   “proactive” deploying   sophisticated   methods   for   situational   awareness.   A   good   overview   of   the   security   challenges   and a   roadmap   for   addressing   them   is   provided   in   IBM   Security   Strategy   [28].   One   of   the   key   theme   in   the agend   is   deeper   integration   of   security   intelligence   through   an  improved   usage  of  analytics.  The   analitycs  in turn  should  be  leveraged  by  the  advances  in  the  area  of  artificial  intelligence.

2.4 IIP are human-‐centric It   would   be   wrong   to   think   that   intelligent   industrial   processes   assume   diminishing   the   roles   of   human workers   in   operation.   Although   “Intelligence”   of   the   processes   implies   the   all-­increasing   degree   of automatization,   the   humans   will   not   be   excluded   from   the   loop   in   the   foreseeable   future.  Instead   IIP   places new   demands   on   the   skills   of  the  human  operators  and   work  forms   on  the  one   hand   and   should   address  the demands   and   expectations   of   the   human   resources   on   the   technology   and   the   new   worker-­industry relationships  on  the  other. Here   it   is   of   ultimate   importance   that   the   concept   of   IoT   and   CPS   are   not   purely   technical,   they   are centered   around   humans   and   treat   humans   as   the   native   part   of   the   cyber-­physical   socium.   The   main technological   enablers   for   a   deeper   integration   of   humans   into   a   CPS   are   embedded   devices,   mobile gadgets,  smart  clothes  and  service  oriented  architectures. Using  mobile  devices,  work  activities  can  be  computerized  and  automated  outside  of  the  actual  offices  and

12

independent   from   the   workers   location.   Mobility   of   resources,   from   this   point   of   view,   increases   the productivity   of   the   organizations   via   using   out-­of-­office   workplaces   and   thus   it   is   attractive   to   the   modern organizations.   Moreover,   mobile   devices   are   equipped   with   tools   such   as   camera,   voice   recorder,   etc.,  and are   capable   of   sending   and   receiving   data   over   communication   networks,   which   enables   them   to communicate   with   other   devices   (e.g.   sensors)   as   well   as  data   processing   and  data   analysis  systems  (e.g. image   processing,   data   mining   and   machine   learning   systems).   All   of   these   state   of   the   art   technologies and   equipment   can   be   utilized   to   acquire   and   analyze   real-­time   information   from   the   context   of   an environment,  which  in  our  case  is  an  industrial  plant  [26,  27]. The   main   challenge   to   address   in   the   years   to   come   is   to   demonstrate   how   control   and   supervision   of production   plants   can   be   enhanced   by   the   use   of   mobile   wearable   devices   utilizing   automated   knowledge engineering.   This   will   require   developing   the  critical   mass   of   technologies   enabling   more   efficient  supervision and   maintenance   of   industrial   plants   by   distributing   the   functionality   of  control  room   across  wearable  mobile devices   of   the   personnel   and   assisting   their  operation  with  a   set  of   services   based   on  automated   knowledge engineering.

2.3.1 Educating the next generation of CPS-‐enabled specialists While   in   general   modern   students   are  highly  technology-­aware,   a   more   systematic   approach  to   education  of future   specialist   of   Intelligent   Industries   is   needed.   Specifically   Intelligent   industries   will   call   both   for   the diverse   knowledge   of   the   CPS   ecosystems   as   well   as   a   quick   refactoring   their   skills.   The   traditional engineering  educations  needs  to  be  upgraded  in  order  to  leverage  the  specialists  of  the  required  quality. Dependability   of   a   computing   system   in   general   is   an   integrated   system   property   jointly   characterized   by: security,   reliability,   availability   and   manageability.   The   ultimate   importance   of   advancing   the   theories   and foundations   for   highly   dependable,   scalable   ICT   systems,   methods   for   verification,   validation   and implementation   of   such   systems   based   on   fault   tolerance,   controlled   degradation   and   self-­healing   is recognized   on   transnational   level   [25]   as   well   as   by   the   major   players  in   the   respective   industries.  Learning how   to   design   highly   dependable   distributed   systems   is   hard   due   to   the   inter-­disciplinary   nature   of   the problem.   Traditionally   education   in   Europe   rather   narrow-­focused   with   relatively   small   insights   into   a  “bigger picture”.   This   situation   is   not   by   any   means   unnatural   taking   into   account   a   large   range  of  adjacent  topics. Nevertheless,   it   becomes   more   and   more   evident  that   the   improvement   should  happen   through   more  evident links  to  the  particular  industrial  needs  and  proper  attention  to  the  innovative  side  of  the  education. The   problematic   of   educating   specialists   of   the   future   Internet   of   Things   is   nowadays   very   hot   in   industrial countries.   Amongst   recent   education   activities   targeting  this  objective   are  works  from  the  Open  University  in the   UK   [23]   and   Luleå   University   of   Technology   [24].   As  one  of   the   challenges  the  university  should  face  on the   way   towards   enabling   the   IIP   industries   the   authors   highlight   the   importance   of   development   dedicated BSc   and   MSc   specializations   as   well   as   entire   graduate   schools   consistently   focusing   on   providing   the holistic  picture  of  the  cyber-­physical  systems  and  the  Internet  of  Things.

13

2.4 Summary of research challenges The  discussion  on  challenges  the  universities  and  the  industries  are  confronting  towards  enabling  Intelligent Industrial  process  vision  is  summarized  below  on  the  per  high-­level  objective  basis. ●









Cross  domain  knowledge  adoption  and  transfer ○ military,  space,  consumer  electronics,  robotics,  automotive  electronics,  business computing,  gaming. High  level  objective  1:    Globally  Integrated    Industries ○ systematization  of  domain-­specific  raw  data  and  its  conceptualization. ○ provide   seamless   access   to   any   process   related   information   (measurements)   and   its integration   with   process   models   in   real-­time,   enabling   more   accurate   control   and reconfiguration  of  processes. ○ adaptation  of  LTE-­A  architecture  to  machine-­to-­machine  communications  in  general  and  the specifics  of  industrial  communications  in  particular.  On  a  high  level  of  abstraction  the challenge  for  the  developers  and  vendors  of  LTE  technology  is  to  enable  mobile  operators as  the  first  tier  data  aggregation  service  providers. ○ bridge  the  cross-­sectorial  gap  between,  data,  telecom  and  automation    industries  by homogenizing  the  design  and  development  principles  as  well  as    developing  methods  for joint  analysis  and  optimization. High  level  objective  2:  Flexible  factories  and  processes ○ enable  co-­design  of  processes  and  supporting  services. ○ observation  based  (learned,  deducted)  models  of  systems  of  systems  accounting  for operators’  actions. ○ dynamically  evolving  SOA  based  on  run-­time  deduced  models. ○ flexible  programming  languages  allowing  for  (semi-­)  automatic  re-­tasking  of  functional components. ○ simulation  technology,  which  will  be  used  daily  throughout  the  engineering  life-­cycle  . High  level  objective  3:  Secure  IIP ○ Enable  an  integrated  security  across  human,  data,  application  and  infrastructure  axis. ○ Wider  usage  of  advanced  methods  for  data  analytics  to  leverage  proactive  security. High  level  objective  3:  Human-­centric  IIP ○ Enable  the  “worker-­as-­a-­service”  vision. ○ Develop  tools  and  education  programs  for  CPS  aware  specialists  for  future  IIP  industries.

14

References 1.  Securing  the  future  of  German  Manufacturing  Industry:  Recommendations  for  implementing  the  strategic initiative  INDUSTRIE  4.0.  Final  report  of  the  Industrie  4.0  Working  group.  April  2013.  [Online.]  Available: http://www.acatech.de/fileadmin/user_upload/Baumstruktur_nach_Website/Acatech/root/de/Material_fuer_S onderseiten/Industrie_4.0/Final_report__Industrie_4.0_accessible.pdf    .  Last  Accessed:  December  2013. 2.  Swedish  production  2025  Strategic  research  and  innovation  agenda  to  meet  the  global  challenges. September  2011.  (Svensk  produktion  2025  Strategisk  forsknings-­  och  innovationsagenda  för  att  möta  de globala  utmaningarna.  Sept.  2011).  [Online.]  Available: http://www.teknikforetagen.se/Documents/Produktion/Svensk_produktion_2025.pdf  .  Last  Accessed: December  2013. 3.  Factories  of  the  future  PPP  strategic  multi-­annual  roadmap.  Prepared  by  the  Ad-­hoc  Industrial  Advisory Group.  2010.  [Online.]  Available: http://ec.europa.eu/research/industrial_technologies/pdf/ppp-­factories-­of-­the-­future-­strategic-­multiannual-­road map-­info-­day_en.pdf  .  Last  Accessed:  December  2013. 4.European  Roadmap  for  Industrial  Process  Automation.  ProcessIT.EU  2012.  [Online.]  Available: http://processit.eu/Content/Files/Roadmap%20for%20IPA_130613.pdf      .  Last  Accessed:  December  2013. 5.  A  Roadmap  for  U.S.  Robotics  From  Internet  to  Robotics  2013  Edition.  [Online.]  Available: http://robotics-­vo.us/sites/default/files/2013%20Robotics%20Roadmap-­rs.pdf  .  Last  Accessed:  December 2013. 6.   Edward   A.   Lee,   “Cyber-­Physical   Systems   -­   Are   Computing   Foundations   Adequate?”,   Position   Paper   for NSF   Workshop   On   Cyber-­Physical   Systems:   Research   Motivation,   Techniques   and   Roadmap   October 16-­17,  2006,  Austin,  TX 7.   V.   Vyatkin,   “Software   Engineering   in   Factory   and   Energy   Automation:   State   of   the   Art   Review”,   IEEE Transactions  on  Industrial  Informatics,  2013,  doi:  10.1109/TII.2013.2258165 8.   “COMMUNICATION   FROM   THE   COMMISSION   SMART   CITIES   AND   COMMUNITIES   -­   EUROPEAN INNOVATION   PARTNERSHIP”,   Available: http://ec.europa.eu/energy/technology/initiatives/doc/2012_4701_smart_cities_en.pdf 9.    The  Cisco  Unified  Service  Delivery. Available: http://www.cisco.com/web/IN/solutions/sp/sp_datacenter_virtualization/unified_service_delivery.html 10.  IBM  Intelligent  Operation  Center  for  Smart  Cities. Available:  http://www-­03.ibm.com/software/products/en/intelligent-­operations-­center/ 11.  Urbiotic.  Web  site.  Available:  http://www.urbiotica.com 12.  Living  PlanIT.  Web  Site.  Available:  http://living-­planit.com/UOS_overview.htm 13.  “Raw  data,  now!”,  Tim  Berners-­Lee.  Available:  http://www.wired.co.uk/news/archive/2012-­11/09/raw-­data 14.  The  Arrowhead  project. Available: http://www.ltu.se/research/subjects/Kommunikations-­och-­berakningssystem/Forskningsprojekt/Arrowhead-­ Project-­1.109826?l=en

15

15.    Ela  Kumar,  Artificial  Intelligence,  I  K  International  Publishing  House,  ISBN-­13:  978-­8190656665,  Sept. 2008. 16.  Douglas  H.  Rothenberg,  Alarm  Management  for  Process  Control,  2009,  Hard  Cover,  610  pg.,  published by  Momentum  Press,  Print  ISBN:    978-­1-­60650-­003-­3. 17.        Varanon  Uraikul,  Christine  W.  Chan,  and  Paitoon  Tontiwachwuthikul.  2007.  Artificial  intelligence  for monitoring  and  supervisory  control  of  process  systems.  Eng.  Appl.  Artif.  Intell.  20,  2  (March  2007) 18.  Macías-­Escrivá,  F.  D.,  Haber,  R.,  del  Toro,  R.,  &  Hernandez,  V.  (2013).  Self-­adaptive  systems:  A  survey of  current  approaches,  research  challenges  and  applications.  Expert  Systems  with  Applications,  40(18), 7267–7279. 19.  G.Prem  Kumar  and  P.  Venkataram.  1997.  Review:  Artificial  intelligence  approaches  to  network management:  recent  advances  and  a  survey.  Comput.  Commun.  20,  15  (December  1997) 20.          Jean  Philippe  Martin-­Flatin,  Gabriel  Jakobson,  and  Lundy  Lewis.  2007.  Event  Correlation  in  Integrated Management:  Lessons  Learned  and  Outlook.  J.  Netw.  Syst.  Manage.  15,  4  (December  2007) 21.  Gotsis,  A.G.;;  Lioumpas,  A.S.;;  Alexiou,  A.,  "M2M  Scheduling  over  LTE:  Challenges  and  New Perspectives,"  Vehicular  Technology  Magazine,  IEEE  ,  vol.7,  no.3,  pp.34,39,  Sept.  2012 22.  Z.  Zhu,  et  al.  “Virtual  Base  Station  Pool:  Towards  A  Wireless  Network  Cloud  for  Radio  Access Networks”,  Proceedings  of  the  8th  ACM  International  Conference  on  Computing  Frontiers,  2011.enges  and New  Perspectives,"  Vehicular  Technology  Magazine,  IEEE  ,  vol.7,  no.3,  pp.34,39,  Sept.  2012 23.  G.  Kortuem,  A.  Bandara,  N.  Smith,  M.  Richards,  and  M.  Petre,  “Educating  the  internet-­of-­things generation,”  Computer,  vol.  46,  no.  2,  pp.  53–61,  2013. 24.  Osipov,  E.;;  Riliskis,  L.,  "Educating  innovators  of  future  Internet  of  Things,"  Frontiers  in  Education Conference,  2013  IEEE  ,  vol.,  no.,  pp.1352,1358,  23-­26  Oct.  2013. 25.    Information  Society  Technologies  Advisory  Group,  “Orientations  for  EU  ICT  R&D  &  Innovation  beyond 2013:  10  Key  Recommendations  Vision  and  Needs,  Impacts  and  Instruments”  ,  ISTAG  Report,  [Online] Available:  URL: http://cordis.europa.eu/fp7/ict/istag/documents/istag_key_recommendations_beyond_2013_full.pdf    ,  2011 26.  Jörn  Nilsson,  Tomas  Sokoler,  Thomas  Binder,  Nina  Wetcke  (2000)  Beyond  the  Control  Room:  Mobile Devices  for  Spatially  Distributed  Interaction  on  Industrial  Process  Plants.  Handheld  and  Ubiquitous Computing,  Lecture  Notes  in  Computer  Science,  Volume  1927,  Springer  Berlin  Heidelberg,  pp  30-­45,  DOI: 10.1007/3-­540-­39959-­3_3 27.  Engin  Ozdemir,  Mevlut  Karacor  (2006)  Mobile  phone  based  SCADA  for  industrial  automation,  ISA Transactions,  Volume  45,  Number  1,  January  2006,  pages  67–75. 29.  IBM  security  Strategy.  Presentation.  [Online.]  Available: https://www-­304.ibm.com/partnerworld/wps/servlet/download/DownloadServlet?id=AOns_zTAPyXiPCA$cnt& attachmentName=ibm_security_strategy.pdf&token=MTM5NDIwMzc5ODczMA==&locale=en_ALL_ZZ

16