Data-intensive systemer data overalt og masser af det! Professor Torben Bach Pedersen Center for Data-intensive Systemer Aalborg Universitet
Daisy Forskere Rene Hansen <rhansen@cs.aau.dk> postdoc 25 forskere Christian Thomsen Dalia Kaulakienė Nerius Tradišauskas Darius Sidlauskas <chr@cs.aau.dk> <daliak@cs.aau.dk> <nerius@cs.aau.dk> <darius@cs.aau.dk> postdoc Ph.D. student Ph.D. student Ph.D. student Dingming Wu <dingming@cs.aau.dk > Ph.D. student Simon Kongshøj Nadeem Iftikhar Laurynas Šikšnys <sk@cs.aau.dk> <nadeem@cs.aau.dk> <siksnys@cs.aau.dk Ph.D. student Ph.D. student Ph.D. student 1.2 professorer 6 lektorer. 1 adjunkt 3 postdoc 14 PhDer 1 VA Rico Wind <rw@cs.aau.dk> Ph.D. student Saulius Samulevičius <sauliuss@cs.aau.dk> Ph.D. student Laurynas Biveinis <lauras@cs.aau.dk> Ph.D. student Kostas Tzoumas <kostas@cs.aau.dk> Ph.D. student Carmen Ruiz <carmrui@cs.aau.dk> Ph.D. student Liu Xiufeng <xiliu@cs.aau.dk> Ph. D. Student Kurt Nørmark <normark@cs.aau.dk> associate professor Christian S. Jensen Hua Lu <csj@cs.aau.dk> <luhua@cs.aau.dk> professor (part time) associate professor Torben Bach Pedersen Simonas Šaltenis <tbp@cs.aau.dk> <simas@cs.aau.dk> professor associate professor Bent Thomsen <bt@cs.aau.dk> associate professor Lone Leth Thomsen Kristian Torp Ralf Rantzau <lone@cs.aau.dk> <torp@cs.aau.dk> <ralf@cs.aau.dk> associate professor associate professor assistant professor BrainMatch 27. marts 2012 2
Daisy Forskning Programming Technology Programming paradigms and specific language facilities and concepts for general- and special-purpose languages, embedded, mobile, data intensive, web and high-performance applications. Data Warehousing and Business Intelligence Separate talk Spatio-Temporal Data Management Spatio-temporal indexing, query processing algorithms, and location privacy. The research is performed in the context of unconstrained space, spatial networks, and in indoor settings. Location-Based Mobile Services Infrastructure for location-based mobile services, including indoor Wi-Fi-based positioning, indoor moving object management based on RFID-type positioning, seamless indoor-outdoor positioning, and location-related spatio-textual queries. BrainMatch 27. marts 2012 3
Daisy Forskning: Programmering Goal: influence next generation mainstream programming languages Experimenting with advanced programming Functional and OO integration Programmatic Program Construction Parallel and Concurrent Programming Developing a new programming language ASL Abstraction Step Language BrainMatch 27. marts 2012 4
Daisy Forskning: Programmering Real-time Java Definition and implementation of Predictable Java profile Verification of RT Java programs via UPPAAL DPT and DES/CISS collaboration Java Objects project (lots of local industry partners) EU COST Action IC0701 Formal Verification of Object-Oriented Software Interactive collection of tests while programming BrainMatch 27. marts 2012 5
Spatio-Temporal Data Management Support for workloads with very frequent updates Tracking of moving objects Modeling, indexing, and query processing proposals for spatial-network constrained objects Particular attention given to indexing techniques Several new indexing techniques proposed Benchmarking of indexes has been pursued. BrainMatch 27. marts 2012 6
TPR-Tree Time Parameterized R-tree an index for the present and predicted future positions of continuously moving objects Future linear trajectories are indexed. Spatial (and spatio-temporal) queries are supported. R1 p6 Query p1 R3 p2 R4 p4 R5 p 7 R1 R2 p3 p8 R3 R4 R5 R6 R7 p5 R2 R7 p12 R6 p9 p11 p10 p13 p1 p2 p3 p4 Pointers to data items p5 p6 p7 p8 p9 p10 p11 p12 p13 BrainMatch 27. marts 2012 7
TPR Tree Successors BrainMatch 27. marts 2012 8
Update-intensive workloads Advances in wireless networks and sensing devices Location-based services Specific example tracking cars in the whole country Total number of vehicles: 10 Million Average speed: 10 m/s (36 km/h) Required tracking accuracy: 10 m Result: 10 million updates per second + support for concurrent queries! BrainMatch 27. marts 2012 9
PGrid PGrid a main-memory spatial indexing technique for update-intensive workloads exploiting the parallelism of modern multi-core processors Scales well with the number of threads Returns fresh query results. Number of physical threads in the processor PGrid Techniques, that do not fully exploit the available parallelism BrainMatch 27. marts 2012 10
Mobile Services/Indoor Seamless Outdoor AND indoor integration Streamspin framework Be for mobile services what Youtube is for video Umbrella for integration Indoor Modeling, positioning, tracking, indexing, query processing Location-related privacy for LBS and geo-social networks Spatial web objects: mining, ranking, and querying BrainMatch 27. marts 2012 11
Hvad er Streamspin? En platform til at levere og modtage lokations- og kontekstbevidste mobile services (Youtube for Mobile Services) = limen der binder udbydere og modtagere af services sammen Service udbydere kan fokusere på indhold. BrainMatch 27. 12 marts 2012
BrainMatch 27. marts 2012 13
GPS Data og ITS BrainMatch 27. marts 2012 14
ITS Open Source Architecture Project specific software pygrametl Psycopg Netwo rkx LatLon- UTM- Conversi on Shape ly Rtree Spatial Index M-GEMMA (64 bit) PostgreSQL 8.4 Python 2.6.6 (64 bit) / 2.7 (32 bit) 64 bit Operativing system BrainMatch 27. marts 2012 15
Speed map A complete speed map for Denmark Using only GPS data Usages Identify and quantify congestion Road segments where vehicles drive to fast ( black spots ) Detect rush hours Find fastest route Done (60% coverage in DK)! 30 20 40 55 50 45 35 45 Unit is km/h 25 30 30 30 70 50 BrainMatch 27. marts 2012 16
06:00 06:30 07:00 07:30 08:00 08:30 09:00 09:30 10:00 10:30 11:00 11:30 12:00 12:30 13:00 13:30 14:00 14:30 15:00 15:30 16:00 16:30 17:00 17:30 18:00 18:30 19:00 19:30 20:00 20:30 21:00 21:30 Østre Alle (ring 2), Aalborg Sønderbro/Østre Alle og Hobrovej/Østre Alle 60 55 50 45 40 35 30 25 20 Simple punkt Punkt norm. hast. Tur hast. Mastra hast. BrainMatch 27. marts 2012 17
Eco Routes University 20 10 20 10 20 Airport 5 5 30 10 Shortest distance: Consumption = 80 Most Eco-Friendly: Consumption = 70 BrainMatch 27. marts 2012 18
Result: University to Airport Smallest Consumption Least Time Shortest Distance Method Distance (km) Cons. Time Smallest consumption 13,75 1875 14:04 Least time 13,75 1875 14:04 Shortest distance 11,50 2343 21:21 BrainMatch 27. marts 2012 19
BagTrack styr på bagagen Bagagestrimler med RFID chips i kan læses på afstand Nummerplade (ID), rute, datp Vision: verdensomspændede bagage information i realtid! AAU forskning Rensning af data (få mening ud af RFID læsninger) Kontinuerte forespørgsler (får svar i realtid) Business intelligence: DW+data mining: find problemerne Arkitektur: store datamængder + tusinder af lokationer Statistik: håndter at ikke alt får RFID på Departure Transfer Arrival BrainMatch 27. marts 2012 20
Interesseret i mere? Daisy hjemmeside www.daisy.aau.dk APPNET netværk Applikationsudvikling http://www.brainsbusiness.dk/dk/ictnorcom_- _business_forum/fokusnetvaerk/appnet/appnet.htm Daisy Innovation Vækstforum projekt http://www.daisy.aau.dk/collaboration/dain.php Kontakt: Christian Thomsen chr@cs.aau.dk BrainMatch 27. marts 2012 21