Software Engineering for Connected Car

  • View

  • Download

Embed Size (px)


Effective software engineering for developing connected car systems is discussed by Prof. June Sung Park at KAIST.



  • 2



    Ohio State University, Computer Science Industrial Engineering


    University of Iowa, Information Systems (1989-2000)

    Institute for Operations Research and Management Sciences (INFORMS), Technical

    Section on Telecommunications (1998-2000)

    SDS, CTO/ (1998-2009)

    Hon Company, HP Software, Rockwell Collins, LG, , , , , IT


    Marquis Whos Who in Science and Engineering

    Marquis Whos Who in Media and Communications


    KAIST, / (2010-)

    KAIST, Smart Cloudlet Research Program ( 5G )

    SW (SEMAT: Software Engineering Method and Theory), San Rafael, CA, U.S.A.;

    OMG Essence (EssenceKernel and Language for Software Engineering Methods)

    Information Technology and Management Associate Editor

    Telecommunication Systems Associate Editor

    SW(KOSTA), SW , , /

    SaaSTF , , , , ,

  • , , , , IoT IT SW , .

    (Service-Oriented Architecture) (Model-Driven Agile Development) SW .


  • 4


  • 5

    Built and run on

    Drive use of

    Store and process



    Big Data Analytics Cloud



  • Among 100 largest American companies

    in 1917 only 39 remained, and only 18

    managed to stay in the top 100 in 1987.



  • (Value Chain) SMACI (Use Case) (value Proposition)

    SMACI Connected Car (Reference Architecture), , , /,

    / (UX Scenario), (Business Process) (Information Semantics)

    SMACI (Interoperability)



  • 8

    The new breed

    of customer

    learns about

    products and

    services from

    new media, is

    more influenced

    by peer reviews,

    and is much

    more informed.

    Businesses must

    sell to the

    customer via a

    multiplicity of

    channels, and

    also meet the


    demand for

    increasing levels

    of choice and


    Dealers need to

    provide the

    training and tools

    to sell the add-on

    technologies and

    services, and

    collect payment

    from the

    customer for

    each component.

    With more and

    more connected

    services that are


    customers need a

    simple way to

    quickly get

    started using


    The high usage

    and price of

    services hinges on

    the supporting

    infrastructure in

    place to help

    customers learn

    how to derive

    value from their

    services, and sign

    up and manage


    Businesses should

    invest in building

    customer self-

    service, since it

    pays dividends

    both in terms of


    profitability as

    well as customer


    Businesses should

    ensure that renewal

    and upgrades for

    service are not only


    but also

    incentivized, and

    cater to customer

    needs for choice,


    and self-sufficiency.

  • 9

  • 10

    AR Navigation

    Internet Connection

    Stream Media

    Smart Phone Integration

    Driving Assistance

    Health Monitoring

    Remote Vehicle Tracking

  • Navigation: Live traffic information Points of interest displayed in the car Finding parked car Intelligent parking - pinpointing available parking spots Counterevidence for speeding tickets Areally/ timely extension of navigation area Booking of parking spots in advance Intermodal route planning Automatic traffic sign recognition Locally based information about events Offers of available parking spots Electronic logbook

    Safety: eCall Wrong-way driver warning Prevention of accidents (M2M communication) Health check of driver

    Infotainment: WLAN in the car Purchasing and downloading music Entertainment streaming into the cars displays Synchronization via the airwaves SMS messages - reading and sending Spotify and Internet radio Location-based ads Business functions such as calendars, address books Social networking in the car Location-sharing and tracking of friends


  • Remote telematics: Remote control Stolen vehicle recovery Surveillance of the car Analyzing driving behavior/ optimizing fuel efficiency

    Diagnose: Self-diagnosis including data cloud Used car check

    Insurance: Usage-based insurance Combined insurance services (e.g. bCall) Cross-selling offers

    Ad hoc carpooling: Combined booking of cars and parking spots Private car sharing without physically exchanging keys

    Other: Reminder of forgotten mobile devices in the car Mobile payment of car tolls Concierge services Automatic information on delays Location-based memory function Leasing rate based on driving behavior Current car residual value sent to the consumer


  • 13

  • 14




    Cloud /






    Cloud /



  • 15

    Enable with APIs and Control with Identity

    Onboard Internal Developers

    Composite Web & Mobile Apps

    APIs & External Developers

    Partner & Payment APIs

    Standards & Regulations

    Auto Supply Chain APIs

    Traffic Mgmt APIs

    Fleet Tracking APIs

  • 16

  • Use Case Cloud

    Analytics IoT Mobile

    SaaS PaaS IaaS LTE WLAN

    Live traffic

    information Private Private Public Yes Yes Yes Yes


    parking Private Public Public Yes Yes Yes

    Automatic traffic

    sign recognition No No No Yes Yes No Yes

    eCall Private Public Public Yes Yes No

    Prevention of

    accidents No No No No Yes No Yes

    Internet radio Public Public Public No No Yes No

    Self-diagnosis Private Private Public Yes Yes Yes


  • 18

  • 19

  • 20

    Service-Oriented Architecture

  • 21

  • 22

    Intel set up the cloud adoption

    strategy and process based

    on the ODCA framework:

    Conducted an environment


    Created cloud definitions,

    attributes and taxonomy.

    Identified potential benefits

    and risks of cloud services.

    Developed a cloud use

    case model.

    Revised the enterprise

    architecture to

    accommodate cloud


    Developed cloud adoption


    We aligned IT priorities to Intels key focus areas. We extend our investments in SMACI to accelerate Intel products TTM, grow revenue and improve operational efficiency. Kim Stevenson, CIO

  • 23

  • 24

  • 25

    Ford expands connected services for

    customers around the world with the

    cloud-based Ford Service Delivery

    Network, powered by Microsoft Azure.

  • 26

    IoT means mass

    adoption of ubiquitous

    computing causing

    industry-wide business


    50 billion things will

    be connected by IoT

    by 2020. (Cisco)

  • 27

    Thing-centric architecture:

    Industrial machines (e.g. transportation, construction, utilities, medical machines)

    having sensors, storage, processing

    capacity and connection to the Internet

    Gateway-centric architecture (Fog


    Gateway (e.g. smart mobile devices, IoT gateways) aggregating data from many

    things and running applications and

    connecting to the Internet (e.g.

    fitness/healthcare wearables, smart home,

    building, utilities, smart cities)

  • 28

    Cloud-centric architecture (Cloud


    Consumer-based IoT, Office machines

    Enterprise-centric architecture (On-

    Premise Computing; Intranet of Things)

    Things and computing behind the enterprise firewall (e.g. things in a

    hospital or a factory connected by a local


  • 29

  • From the dawn of civilization until 2003, humankind generated 5 exabytes (1018) of data. Now we produce 5 exabytes every two days, and the pace is accelerating.

    A disk drive that can store all the music in the world (a few TBs) is only $600. However, it takes an average of 2.5 hours to read 1TB (1012).

    With sensors monitoring everything from tire pressure to engine RPM to oil temperature and speed, cars can produce anywhere from 5 to 250 gigabytes of data an hour.

    Advanced concept cars go even higher; Googles autonomous vehicle, for example, generates about 1 gigabyte of data every second.


  • Big data is structured and

    unstructured, static and streaming

    data of large volumes on the order of

    petabytes (1015) which relational

    database and data warehouse

    technologies cannot efficiently store

    and process

    Hadoop Distributed File System (HDFS)

    and MapReduce opened new

    possibilities allowing to scale out

    with low-cost commodity hardware.

    If you distribute 1TB on 100 disk

    drives, it takes 1.5 minutes to read.