database support for mobile computing applications

3.6.2000 R. Bayer, DB for Mobile Comp. 1

Database Support for Mobile Computing Applications

R. BayerInstitut für Informatik

TU München3.6.2000

Rudolf/Vorlesungen/DWH-SS2000/DWH-MobileComp


Future Mobile Phones

• contain WEB browser• have UTMS bandwidth 1.2 Mb/s• integrated with GPS, i.e. exact position of

phone and owner is known

entirely novel applications


Car Traveler on Road

• Car knows destination, gas consumption, gas remaining

• Questions and services:– reachable cheapest gas station along route?– next BMW repair station?– restaurant with in 100 km serving asparagus or

lobster?


Car Traffic Management in MunichAssumptions:

106 cars registered

105 cars traveling

Route-DB for preplanned routes, 10 per car, 104 B/route

106 cars * 10 routes/car * 104 B/route = 1011 B = 100 GB

Drive-DB

105 cars *1 route/car * 104 B/route = 109 B = 1 GB


Reads and UpdatesUpdates to Drive-DB: 1 update/(car*min)

105 cars*1 update/(car*min) =

105 updates/min ~ 1.700 t/s

Reads from Route-DB: travel time = 30 minutes

(105 cars/30 min)*1 read/car =

105 reads/(30*60) sec ~ 56 t/s


Route CalculationsInitial Routes: compute one optimal route on departure of car based on present traffic situation in Drive-DB

56 routes/s

Adaption of Routes: with every update of own position, car wants to know optimal route, in most cases the same as before?

1.700 checks/s


City Tourist• What is this? ( I am standing in front of?)

I want to know more, what is the URL?

• Where is the next bus or subway station, taxi point, public toilet, please guide me to it (load down part of city map with direct route inserted or text: )

• Where is medium priced fish restaurant within 500 m? Show it, make a reservation.

• I need an antihistamine, where is the next open pharmacy?


Air Traveler

• T is approaching airport Munich by car, server knows flight F and car or subway.

• „Proceed to parking P1, level 2, space 437“• „Mr. Bayer, you are checked into flight LH451 to

Cologne, leaving from gate A17, boarding time is 7:45, go towards gate in 7 minutes

• On return: parking ticket paid automatically, gate opens, when car approaches.


Truck Management

• What is position of trucks near Cologne with at least 1 ton loading capacity and slack of 1 hour in their delivery schedule?

• Show me planned travel route and stops of truck 327


Shoppers: Info pull or push?• On entering Kaufhof, select WEB server• Information pull:

– guide me to the perfume dept.– bestsellers in detective stories?

• Information push (to close shoppers):– McDonald coming up 120 meters to right– H&M has bikini special– WOM: Madonna‘s new single just arrived


Nightlife

• Which friends are in Schwabing now?• Who is DJ at P1?• Live video shot with soundtrack from P1.

ODODO, Kunstpark Ost• „My position is ...“• „Anybody close to share taxi to ...?


Sales Representative

• On approaching customer C: download the relevant marketing materials, sales and delivery data for C, combined with DWH.

• Show homepage and picture of person I meet• Did C have any reclamations recently?• Any significant changes in buying pattern of C

recently?• Question: B2B applications ???


Taxi Service

• Please pick me up to go to Kentvale apartments

• Taxi server finds optimal taxi: „Mr. Bayer, please stay where you are, a blue Comfort Taxi with plate SHA-488 368C will pick you up in approximately 3 minutes


Emergency Service

• Pushing the panic button calls police or ambulance

• mobile phone transmits position and medical data of owner, voice of attacker


Hotel Guest

• Theatre in walking distance showing movie with Meryl Streep starting around 20 h? Show map with route.

• reserve and pay ticket• nearby bar serving Heineken?• ...


The Database ProblemTypes of Subjects:

• Fixed location

• constant state: buildings, statues, paintings in museums, with a lot of additional information

• variable state: restaurants with seats

• Mobile• constant state, if subject refuses to disclose state • variable state


Attributes and Dimensions1. Location: 2 or 3 dimensions

2. Time: for tractable subjects like trucks

3. Classification of Subjects: hierarchies and MHC ~ subject ID, e.g.

automotive (cars, gas stations, parking, repair stations)eatingshopping arts 4 to 5 dimensions

4. State: movie playing now, number of free seats, ...probably modeled like features of GfK


Database Size (Mengengerüst)1. Assumptions about Size

• 107 mobile subjects in Bavaria * 1000 B = 10 GB

• 105 fixed subjects, which are constant, but with 105

Bytes for image = 1010 B = 10 GB

• 105 variable subjects * 10 KB = 1 GB

11 GB with high update frequency !


2. Assumptions about Updates and QueriesA person is moving at most 10% of the time,

i.e. 2.5 hours per day?

Update rate 1 per minute for variable subjects

11*105 variable subjects * 1 update per minute

= 11*105 /60 U/s ~ 20.000 updates /s

achievable with parallel DBMS !


Crosscheck for Airport Munich20 flights /h* 300 passengers/flight =

6.000 passengers / h ~ DB size 60 MB

10 updates or queries per passenger?

60.000 t/h = 1000 t/m ~ 20 t/s

1 Server handles problem, second server for redundancy and mirroring DB!


Hurdles• Data Acquisition

– Mobile subjects: Telekom companies– Fixed constant subjects:

105 * 10 DM = 1 million DM– Variable subjects: free data via advertising

• Business Model– network providers– content providers – SW and technology providers

database support for mobile computing applications

Documents