Professor Haris och hans forskare på KTH har ett samarbete med de som utvecklar InfoSphere Streams på IBM Research. I detta projektet analyseras trafikdata från Stockholmsområdet för att se hur man kan nyttja informationen på bästa sätt för att styra trafiken smartare och informera resenärerna om hur man tar sig fram på bästa sätt.
Förutom själva Stockholmsområdet, så analyserar man trafiken till/från Arlanda. Bland annat vill man prediktera sannolikheten för att man kommer i tid till sin flygavgång på Arlanda beroende på vilken tid man skall åka och beroende på vilket transportsätt man väljer. KTH är vinnare till priset för en smartare planet 2010.
Talare: Haris N. Koutsopoulos, Professor and Head of Transportation and Logistics Division, KTH
Denna presentation hölls på ett seminariepass för Business Analytics & Optimization under IBM Software Day 2010.
5. Intelligent Transportation Systems (ITS)
Sensor, communications, computing, and
IT technologies to improvethe efficiency
and safety of the transport system
Widespread adoption of data collection
technologies
5
6. Stockholm region data
• Traffic data
– loop detectors
– GPS (1500 vehicles)
– microwave detectors
– traveltimes
• Public transport
• Environmental data
• Weather
• Infrastructure and roadworks
• Parking
• Incidents and events
6
7. Data Overload
• Information has
gone from scarce
to superabundant.
That brings huge
new benefits but
also big headaches.
Economist, Feb. 2010
7
8. The ITS Laboratory at KTH
• NexTMC3: Next Generation Traffic
Management, Communications, and
Control Center for Sustainable Urban
Transport
• Support from IBM (Shared University
Research Award), Transport Administration,
Trafik Stockholm, KTH
8
9. The ITS Laboratory at KTH
• Real time streaming data
traffic + PT + environment + weather + ….
• Hardware
• IBM Blade Center
• 10 blade servers (HS22)
• 80 CPU cores 2.53 GHz
• 240 GB of memory
• 16 TB external storage
• Software
IBM Infosphere Streams and databases
Redhat Linux
9
12. Problem Characteristics
• Large quantities of continuous,
heterogeneous data streams in
motion
• Real time operations
• Performance and scalability
• Information on demand
• Traffic management centers
• Individuals
• Fleet monitoring
• Exceptions/deviations
• Complex analytics
12
13. Powered by InfoSphere Streams
Real time delivery
Streams delivers:
Ability to fuse structured and Powerful
unstructured data types Analytics
Scalability for large urban traffic
management centers
Millions of Microsecond
Intuitive programming model events per Latency
second
Example: GPS location mapping
4 x86 blade servers Traditional /
250,000 GPS probes per second Non-traditional
data sources
Mapped to 630,000 road segments
13
14. IBM InfoSphere Streams
• Scalability
• Modularity and Extensibility
A toolkit of basic stream-relational operators
and user defined operators (in C++ or Java)
• Stream adapters to ingest/publish data
• Fast processing of high volumes of data
Query on streams
Parallel/distributed platform
• Complex analytics
• High level programming language
14
16. InfoSphere Streams
Connections
High level
PE PE
PE
Job manager
Source
language source compiler
PE
PE Sink
PE
PE
PE
PE
PE Sink
Source PE PE PE Sink
Source PE PE Sink
PE
Processing Processing Processing Processing Processing
Element Element Element Element Element
Container Container Container Container Container
Streams Data Fabric
Physical Network
TCP-IP / Ethernet
x86 x86 x86 x86 x86
X86 X86 X86 X86 X86
Node Node Node Node Node
Blade Blade Blade Blade Blade
16
17. Stockholm data
• 1500 vehicle probes
• More expected in the future
• 10 million GPS points per month
• 1,000 GPS points per minute peak
• 600,000 road segments in a 80km x 80km area
17
21. Conclusion
• Increased availability of large
amounts of traffic data
• IBM Infosphere Streams provides the
real time stream processing
capabilities required to facilitate
applications and services targeting
serious traffic problems
21
22. Acknowledgments
• IBM Sweden and IBM Watson Labs
• Swedish Traffic Administration
• Trafik Stockholm
• Stockholm City
• KTH
22