Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P....
Transcript of Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P....
![Page 1: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/1.jpg)
1 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Institute of Theoretical Informatics, Algorithmics II
Efficient Route Compressionfor Hybrid Route PlanningGernot Veit Batz, Robert Geisberger, Dennis Luxen, Peter Sanders, and Roman Zubkov{batz, luxen, sanders}@kit.edu
KIT – University of the State of Baden-Württemberg andNational Large-scale Research Center of the Helmholtz Association www.kit.edu
![Page 2: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/2.jpg)
2 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
MotivationWhy is hybrid route planning interesting?
Why is it a problem?
![Page 3: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/3.jpg)
Mobile vs. Hybrid Route Planning
3 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Mobile route planning:Routes computed by mobile device in the carHybrid route planning:Routes computed by server, then transmitted to car
route planning query
route
⇒ Hybrid route planning requires mobile radio communication
![Page 4: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/4.jpg)
Mobile vs. Hybrid Route Planning
3 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Mobile route planning:Routes computed by mobile device in the carHybrid route planning:Routes computed by server, then transmitted to car
route planning query
route
⇒ Hybrid route planning requires mobile radio communication
![Page 5: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/5.jpg)
Benefits of Hybrid Route Planning
4 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Advanced route planning algorithmshigh quality routes within millisecondssuited for server systemsdifficult to adapt to mobile devices
Hybrid route planning...makes server algorithms......available to car drivers
route planning query
route
![Page 6: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/6.jpg)
Server-Based Route Planning (1)
5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Time-dependent route planning:[TD SHARC 08] [TCH 09] [ATCH 10] [Brunel et al. 10] [Batz, Sanders 12]
exploit statistical data, e.g., rush houryields time-dependent edge weightsroute depends on time of day
Flexible route planning:[flexible CH 10]
dynamically fine tune cost function......with a parameter: ce + p · de
tradeoff, e.g., energy cost vs. travel time
![Page 7: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/7.jpg)
Server-Based Route Planning (1)
5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Time-dependent route planning:[TD SHARC 08] [TCH 09] [ATCH 10] [Brunel et al. 10] [Batz, Sanders 12]
exploit statistical data, e.g., rush houryields time-dependent edge weightsroute depends on time of day
Flexible route planning:[flexible CH 10]
dynamically fine tune cost function......with a parameter: ce + p · de
tradeoff, e.g., energy cost vs. travel time
![Page 8: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/8.jpg)
Server-Based Route Planning (1)
5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Time-dependent route planning:[TD SHARC 08] [TCH 09] [ATCH 10] [Brunel et al. 10] [Batz, Sanders 12]
exploit statistical data, e.g., rush houryields time-dependent edge weightsroute depends on time of day
Flexible route planning:[flexible CH 10]
dynamically fine tune cost function......with a parameter: ce + p · de
tradeoff, e.g., energy cost vs. travel time
![Page 9: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/9.jpg)
Server-Based Route Planning (2)
6 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Customizable route planning:[Delling et al. 11]
dynamically change cost functione.g., unexpected traffic situations
Multi-criteria route planning:[Delling, Wagner 09]
deal with multiple incomparable costse.g., travel time vs. inconveniency
![Page 10: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/10.jpg)
Server-Based Route Planning (2)
6 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Customizable route planning:[Delling et al. 11]
dynamically change cost functione.g., unexpected traffic situations
Multi-criteria route planning:[Delling, Wagner 09]
deal with multiple incomparable costse.g., travel time vs. inconveniency
![Page 11: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/11.jpg)
Server-Based Route Planning (2)
6 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Customizable route planning:[Delling et al. 11]
dynamically change cost functione.g., unexpected traffic situations
Multi-criteria route planning:[Delling, Wagner 09]
deal with multiple incomparable costse.g., travel time vs. inconveniency
![Page 12: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/12.jpg)
Server-Based Route Planning (3)
7 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Alternative routes:[Abraham et al. 10] [Luxen, Schieferdecker 12]
a handful of alternativesnearly as good as optimal routereasonable different
Hub-labeling algorithms:[Abraham et al. 11] [Abraham et al. 12]
running times < 1 µs implemented with C++fast route planning with database serversmore complicated queries, e.g., route via POI
![Page 13: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/13.jpg)
Server-Based Route Planning (3)
7 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Alternative routes:[Abraham et al. 10] [Luxen, Schieferdecker 12]
a handful of alternativesnearly as good as optimal routereasonable different
Hub-labeling algorithms:[Abraham et al. 11] [Abraham et al. 12]
running times < 1 µs implemented with C++fast route planning with database serversmore complicated queries, e.g., route via POI
![Page 14: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/14.jpg)
Server-Based Route Planning (3)
7 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Alternative routes:[Abraham et al. 10] [Luxen, Schieferdecker 12]
a handful of alternativesnearly as good as optimal routereasonable different
Hub-labeling algorithms:[Abraham et al. 11] [Abraham et al. 12]
running times < 1 µs implemented with C++fast route planning with database serversmore complicated queries, e.g., route via POI
![Page 15: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/15.jpg)
Problems with Radio Communication
8 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Low bandwidth, but complex routesTransmission costs
route planning query
route
⇒ Do data compression!
![Page 16: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/16.jpg)
Problems with Radio Communication
8 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Low bandwidth, but complex routesTransmission costs
route planning query
route
⇒ Do data compression!
![Page 17: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/17.jpg)
Problems with Radio Communication
8 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Low bandwidth, but complex routesTransmission costs
route planning query
route
⇒ Do data compression!
![Page 18: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/18.jpg)
Problems with Radio Communication
8 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Low bandwidth, but complex routesTransmission costs
route planning query
route
⇒ Do data compression!
![Page 19: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/19.jpg)
9 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
ContributionEfficient lossless compression of routes.
![Page 20: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/20.jpg)
Compression of RoutesTo Make Hybrid Route Planning Convenient to Use
10 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Fast: User experiences no delayDriving directions start within ≤ 0.1 sec.Fast compression / decompression / transmission
Lossless:Reconstructed route = server-provided route
Resulting Requirements:Fast algorithmsGood compression rates
route lanning query
route
![Page 21: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/21.jpg)
Basic Setup......Needed by our Approach
11 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Client has basic and fast route planning capabilityClient and Server use same road networkClient uses fixed cost function known to the server
route planning query
route
server client
![Page 22: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/22.jpg)
Basic Idea: Use Via Nodes
12 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
server client
high qualityroute
s t
not optimalwith respect
to client
segments unique optimalwith respect to client
connectionpoints “via nodes”
![Page 23: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/23.jpg)
Basic Idea: Use Via Nodes
12 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
server client
high qualityroute
s t
not optimalwith respect
to client
segments unique optimalwith respect to client
connectionpoints “via nodes”
![Page 24: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/24.jpg)
Basic Idea: Use Via Nodes
12 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
server client
high qualityroute
s t
not optimalwith respect
to client
segments unique optimalwith respect to client
connectionpoints “via nodes”
![Page 25: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/25.jpg)
Basic Idea: Use Via Nodes
12 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
server client
high qualityroute
s t
not optimalwith respect
to client
segments unique optimalwith respect to client
connectionpoints “via nodes”
![Page 26: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/26.jpg)
Basic Idea: Use Via Nodes
12 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
server client
high qualityroute
s t
not optimalwith respect
to client
segments unique optimalwith respect to client
connectionpoints
“via nodes”
![Page 27: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/27.jpg)
Basic Idea: Use Via Nodes
12 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
server client
high qualityroute
s t
not optimalwith respect
to client
segments unique optimalwith respect to client
connectionpoints “via nodes”
![Page 28: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/28.jpg)
Compression with Via NodesExact Definition
13 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Road networkDirected graph G = (V ,E)
Edge weights client: c(u, v) ∈ R≥0 (simple!)Edge weights server: not needed
Path 〈s, . . . ,ui1 , . . . ,ui2 , . . . . . . . . . ,uik , . . . , t〉
Compressedrepresentation 〈〈ui1 ,ui2 , . . . ,uik 〉〉with via nodes
![Page 29: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/29.jpg)
Compression with Via NodesExact Definition
13 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Road networkDirected graph G = (V ,E)
Edge weights client: c(u, v) ∈ R≥0 (simple!)Edge weights server: not needed
Path 〈s, . . . ,ui1︸ ︷︷ ︸unique
shortestpath
,
uniqueshortest
path︷ ︸︸ ︷. . . ,ui2 , . . .︸ . . .
︷. . . ,uik , . . . , t︸ ︷︷ ︸
uniqueshortest
path
〉
Compressedrepresentation 〈〈ui1 ,ui2 , . . . ,uik 〉〉with via nodes
![Page 30: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/30.jpg)
Compression with Via NodesExact Definition
13 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Road networkDirected graph G = (V ,E)
Edge weights client: c(u, v) ∈ R≥0 (simple!)Edge weights server: not needed
Path 〈s, . . . ,ui1︸ ︷︷ ︸unique
shortestpath
,
uniqueshortest
path︷ ︸︸ ︷. . . ,ui2 , . . .︸ . . .
︷. . . ,uik , . . . , t︸ ︷︷ ︸
uniqueshortest
path
〉
Compressedrepresentation 〈〈ui1 ,ui2 , . . . ,uik 〉〉with via nodes
![Page 31: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/31.jpg)
Compression with Via NodesMethod for Decompression
14 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Compute shortest paths between via nodesMobile CH do this in 0.1 sec. [ESA 08]Uniqueness⇒ correctness
〈〈u, v ,w〉〉
s t
≤ 0.1 sec
≤ 0.1 sec ≤ 0.1 sec
≤ 0.1 sec
⇒ Driver experiences no delay!⇒ Convenient to use.
![Page 32: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/32.jpg)
Compression with Via NodesMethod for Decompression
14 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Compute shortest paths between via nodesMobile CH do this in 0.1 sec. [ESA 08]Uniqueness⇒ correctness
〈〈u, v ,w〉〉
s t
≤ 0.1 sec
≤ 0.1 sec ≤ 0.1 sec
≤ 0.1 sec
⇒ Driver experiences no delay!⇒ Convenient to use.
![Page 33: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/33.jpg)
Compression with Via NodesMethod for Decompression
14 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Compute shortest paths between via nodesMobile CH do this in 0.1 sec. [ESA 08]Uniqueness⇒ correctness
〈〈u, v ,w〉〉
s t
≤ 0.1 sec
≤ 0.1 sec ≤ 0.1 sec
≤ 0.1 sec
⇒ Driver experiences no delay!⇒ Convenient to use.
![Page 34: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/34.jpg)
Compression with Via NodesMethod for Decompression
14 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Compute shortest paths between via nodesMobile CH do this in 0.1 sec. [ESA 08]Uniqueness⇒ correctness
〈〈u, v ,w〉〉
s t
≤ 0.1 sec
≤ 0.1 sec
≤ 0.1 sec
≤ 0.1 sec
⇒ Driver experiences no delay!⇒ Convenient to use.
![Page 35: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/35.jpg)
Compression with Via NodesMethod for Decompression
14 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Compute shortest paths between via nodesMobile CH do this in 0.1 sec. [ESA 08]Uniqueness⇒ correctness
〈〈u, v ,w〉〉
s t
≤ 0.1 sec
≤ 0.1 sec ≤ 0.1 sec
≤ 0.1 sec
⇒ Driver experiences no delay!⇒ Convenient to use.
![Page 36: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/36.jpg)
Compression with Via NodesMethod for Decompression
14 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Compute shortest paths between via nodesMobile CH do this in 0.1 sec. [ESA 08]Uniqueness⇒ correctness
〈〈u, v ,w〉〉
s t
≤ 0.1 sec
≤ 0.1 sec ≤ 0.1 sec
≤ 0.1 sec
⇒ Driver experiences no delay!⇒ Convenient to use.
![Page 37: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/37.jpg)
Compression with Via NodesMethod for Decompression
14 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Compute shortest paths between via nodesMobile CH do this in 0.1 sec. [ESA 08]Uniqueness⇒ correctness
〈〈u, v ,w〉〉
s t
≤ 0.1 sec
≤ 0.1 sec ≤ 0.1 sec
≤ 0.1 sec
⇒ Driver experiences no delay!
⇒ Convenient to use.
![Page 38: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/38.jpg)
Compression with Via NodesMethod for Decompression
14 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Compute shortest paths between via nodesMobile CH do this in 0.1 sec. [ESA 08]Uniqueness⇒ correctness
〈〈u, v ,w〉〉
s t
≤ 0.1 sec
≤ 0.1 sec ≤ 0.1 sec
≤ 0.1 sec
⇒ Driver experiences no delay!⇒ Convenient to use.
![Page 39: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/39.jpg)
Compression with Via NodesFrame Algorithm for Compression
15 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Compress path PQ := 〈〈〉〉Repeatedly
remove the maximal unique shortest prefix from Peach appending its last node to Q
s t
1 2
2 1
2 1
1 1
2 2
3
,
Q = 〈〈 〉〉
⇒ Finds minimal number of via nodes
![Page 40: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/40.jpg)
Compression with Via NodesFrame Algorithm for Compression
15 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Compress path PQ := 〈〈〉〉Repeatedly
remove the maximal unique shortest prefix from Peach appending its last node to Q
s t
1 2
2 1
2 1
1 1
2 2
3
,
Q = 〈〈 〉〉
⇒ Finds minimal number of via nodes
![Page 41: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/41.jpg)
Compression with Via NodesFrame Algorithm for Compression
15 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Compress path PQ := 〈〈〉〉Repeatedly
remove the maximal unique shortest prefix from Peach appending its last node to Q
s t
1 2
2 1
2 1
1 1
2 2
3
,
Q = 〈〈 〉〉
⇒ Finds minimal number of via nodes
![Page 42: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/42.jpg)
Compression with Via NodesFrame Algorithm for Compression
15 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Compress path PQ := 〈〈〉〉Repeatedly
remove the maximal unique shortest prefix from Peach appending its last node to Q
s t
1 2
2 1
2 1
1 1
2 2
3
,Q = 〈〈 〉〉
⇒ Finds minimal number of via nodes
![Page 43: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/43.jpg)
Compression with Via NodesFrame Algorithm for Compression
15 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Compress path PQ := 〈〈〉〉Repeatedly
remove the maximal unique shortest prefix from Peach appending its last node to Q
s t
1 2
2 1
2 1
1 1
2 2
3
,Q = 〈〈 〉〉
⇒ Finds minimal number of via nodes
![Page 44: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/44.jpg)
Compression with Via NodesFrame Algorithm for Compression
15 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Compress path PQ := 〈〈〉〉Repeatedly
remove the maximal unique shortest prefix from Peach appending its last node to Q
s t
1 2
2 1
2 1
1 1
2 2
3
,Q = 〈〈 〉〉
⇒ Finds minimal number of via nodes
![Page 45: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/45.jpg)
Compression with Via NodesFrame Algorithm for Compression
15 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Compress path PQ := 〈〈〉〉Repeatedly
remove the maximal unique shortest prefix from Peach appending its last node to Q
s t
1 2
2 1
2 1
1 1
2 2
3
,Q = 〈〈 〉〉
⇒ Finds minimal number of via nodes
![Page 46: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/46.jpg)
Compression with Via NodesFrame Algorithm for Compression
15 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Compress path PQ := 〈〈〉〉Repeatedly
remove the maximal unique shortest prefix from Peach appending its last node to Q
s t
1 2
2 1
2 1
1 1
2 2
3
,Q = 〈〈 〉〉
⇒ Finds minimal number of via nodes
![Page 47: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/47.jpg)
Find Maximal Unique Shortest PrefixModified Dijkstra – simple but slow
16 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
s u
u
d [u] = ∞
d [u] = 6d [u] = 5
2
3
4
4
3
1
Relaxing an edge:Mark node when non-uniquely reachedUnmark node when reached by shorter path
Settling a node:settling a marked node of P or......prefix not a shortest path
⇒ Unique maximal prefix found!s u
u
4
10
6
1P
![Page 48: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/48.jpg)
Find Maximal Unique Shortest PrefixModified Dijkstra – simple but slow
16 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
s u
u d [u] = ∞
d [u] = 6
d [u] = 5
2
3
4
4
3
1
Relaxing an edge:Mark node when non-uniquely reachedUnmark node when reached by shorter path
Settling a node:settling a marked node of P or......prefix not a shortest path
⇒ Unique maximal prefix found!s u
u
4
10
6
1P
![Page 49: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/49.jpg)
Find Maximal Unique Shortest PrefixModified Dijkstra – simple but slow
16 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
s
u
u
d [u] = ∞
d [u] = 6
d [u] = 5
2
3
4
4
3
1
Relaxing an edge:Mark node when non-uniquely reachedUnmark node when reached by shorter path
Settling a node:settling a marked node of P or......prefix not a shortest path
⇒ Unique maximal prefix found!s u
u
4
10
6
1P
![Page 50: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/50.jpg)
Find Maximal Unique Shortest PrefixModified Dijkstra – simple but slow
16 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
s u
u d [u] = ∞d [u] = 6
d [u] = 5
2
3
4
4
3
1
Relaxing an edge:Mark node when non-uniquely reachedUnmark node when reached by shorter path
Settling a node:settling a marked node of P or......prefix not a shortest path
⇒ Unique maximal prefix found!s u
u
4
10
6
1P
![Page 51: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/51.jpg)
Find Maximal Unique Shortest PrefixModified Dijkstra – simple but slow
16 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
s u
u d [u] = ∞d [u] = 6
d [u] = 5
2
3
4
4
3
1
Relaxing an edge:Mark node when non-uniquely reachedUnmark node when reached by shorter path
Settling a node:settling a marked node of P or......prefix not a shortest path
⇒ Unique maximal prefix found!s uu
4
10
6
1P
![Page 52: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/52.jpg)
Find Maximal Unique Shortest PrefixModified Dijkstra – simple but slow
16 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
s u
u d [u] = ∞d [u] = 6
d [u] = 5
2
3
4
4
3
1
Relaxing an edge:Mark node when non-uniquely reachedUnmark node when reached by shorter path
Settling a node:settling a marked node of P or......prefix not a shortest path
⇒ Unique maximal prefix found!s u
u
4
10
1
1P
![Page 53: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/53.jpg)
Contraction Hierarchies (CH)[Geisberger et al. 08]
17 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Construct a hierarchy in a preprocessing step:Order nodes by importanceObtain next level by contracting next nodePreserve shortest paths by inserting shortcuts
x →x
level i
level i + 1
![Page 54: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/54.jpg)
Contraction Hierarchies (CH)[Geisberger et al. 08]
17 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Construct a hierarchy in a preprocessing step:Order nodes by importanceObtain next level by contracting next nodePreserve shortest paths by inserting shortcuts
x →x
level i
level i + 1
![Page 55: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/55.jpg)
Contraction Hierarchies (CH)[Geisberger et al. 08]
17 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Construct a hierarchy in a preprocessing step:Order nodes by importanceObtain next level by contracting next nodePreserve shortest paths by inserting shortcuts
x →x
level i
level i + 1
![Page 56: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/56.jpg)
Contraction Hierarchies (CH)[Geisberger et al. 08]
17 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Construct a hierarchy in a preprocessing step:Order nodes by importanceObtain next level by contracting next nodePreserve shortest paths by inserting shortcuts
x →x
level i
level i + 1
![Page 57: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/57.jpg)
Contraction Hierarchies (CH)[Geisberger et al. 08]
18 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Construct a hierarchy in a preprocessing step:Order nodes by importanceObtain next level by contracting next nodePreserve shortest paths by inserting shortcuts
G = (V ,E) →
s
t
...
⇒ There is always a shortest up-down-path from s to t .
![Page 58: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/58.jpg)
Contraction Hierarchies (CH)[Geisberger et al. 08]
18 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Construct a hierarchy in a preprocessing step:Order nodes by importanceObtain next level by contracting next nodePreserve shortest paths by inserting shortcuts
G = (V ,E) →
s
t...
⇒ There is always a shortest up-down-path from s to t .
![Page 59: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/59.jpg)
Representing Paths with CH Uniquely
19 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
s P
t...u v
x
3; x
2 1
Shortcuts can be expanded recursivelyUp-down-paths contain several shortcutsExpand these shortcuts completely⇒ An up-down-path represents an original path
Question:Is there exactly one
shortest up-down paththat represents P?
We say: P is uniquelyCH-representable
![Page 60: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/60.jpg)
Representing Paths with CH Uniquely
19 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
s P
t...u v
x
3; x
2 1
Shortcuts can be expanded recursivelyUp-down-paths contain several shortcutsExpand these shortcuts completely⇒ An up-down-path represents an original path
Question:Is there exactly one
shortest up-down paththat represents P?
We say: P is uniquelyCH-representable
![Page 61: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/61.jpg)
Representing Paths with CH Uniquely
19 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
s P
t...u v
x
3; x
2 1
Shortcuts can be expanded recursivelyUp-down-paths contain several shortcutsExpand these shortcuts completely⇒ An up-down-path represents an original path
Question:Is there exactly one
shortest up-down paththat represents P?
We say: P is uniquelyCH-representable
![Page 62: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/62.jpg)
Representing Paths with CH Uniquely
19 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
s P
t...u v
x
3; x
2 1
Shortcuts can be expanded recursivelyUp-down-paths contain several shortcutsExpand these shortcuts completely⇒ An up-down-path represents an original path
Question:Is there exactly one
shortest up-down paththat represents P?
We say: P is uniquelyCH-representable
![Page 63: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/63.jpg)
CH-Based Representationof a path with Via Nodes
20 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Path 〈s, . . . ,ui1 , . . . ,ui2 , . . . . . . . . . ,uik , . . . , t〉
CH-basedcompressed 〈〈ui1 ,ui2 , . . . ,uik 〉〉
representationwith via nodes
![Page 64: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/64.jpg)
CH-Based Representationof a path with Via Nodes
20 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Path 〈s, . . . ,ui1︸ ︷︷ ︸uniquely
CH-repre-sentable
,
uniquelyCH-repre-sentable︷ ︸︸ ︷
. . . ,ui2 , . . .︸ . . .︷
. . . ,uik , . . . , t︸ ︷︷ ︸uniquely
CH-repre-sentable
〉
CH-basedcompressed 〈〈ui1 ,ui2 , . . . ,uik 〉〉
representationwith via nodes
![Page 65: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/65.jpg)
CH-Based Representationof a path with Via Nodes
20 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Path 〈s, . . . ,ui1︸ ︷︷ ︸uniquely
CH-repre-sentable
,
uniquelyCH-repre-sentable︷ ︸︸ ︷
. . . ,ui2 , . . .︸ . . .︷
. . . ,uik , . . . , t︸ ︷︷ ︸uniquely
CH-repre-sentable
〉
CH-basedcompressed 〈〈ui1 ,ui2 , . . . ,uik 〉〉
representationwith via nodes
![Page 66: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/66.jpg)
Compression with Via NodesFrame Algorithm for Compression
21 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Compress path PQ := 〈〈〉〉Repeatedly
remove the maximal uniquely CH-representable prefix from Peach appending its last node to Q
![Page 67: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/67.jpg)
Find Uniquely CH-RepresentablePrefixInspired by Binary Search
22 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Find uniquely CH-representable prefix of path 〈u1, . . . ,un〉(`,m, r ) := (1,n,n)while `+ 1 < r do
if 〈u1, . . . ,um〉 is uniquely CH-representable then ` := melse r := mm := b`+ 1 + rc
odreturn 〈u1, . . . ,u`〉
⇒ Checks O(log n) times whether uniquely CH-representable.
![Page 68: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/68.jpg)
Find Uniquely CH-RepresentablePrefixInspired by Binary Search
22 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Find uniquely CH-representable prefix of path 〈u1, . . . ,un〉(`,m, r ) := (1,n,n)while `+ 1 < r do
if 〈u1, . . . ,um〉 is uniquely CH-representable then ` := melse r := mm := b`+ 1 + rc
odreturn 〈u1, . . . ,u`〉
⇒ Checks O(log n) times whether uniquely CH-representable.
![Page 69: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/69.jpg)
Find Uniquely CH-RepresentablePrefixInspired by Binary Search
22 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Find uniquely CH-representable prefix of path 〈u1, . . . ,un〉(`,m, r ) := (1,n,n)while `+ 1 < r do
if 〈u1, . . . ,um〉 is uniquely CH-representable then ` := melse r := mm := b`+ 1 + rc
odreturn 〈u1, . . . ,u`〉
⇒ Checks O(log n) times whether uniquely CH-representable.
![Page 70: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/70.jpg)
How to Find aUniquely CH-Representable PrefixInspired by Binary Search
23 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
...
s t
NOYES NOYES
![Page 71: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/71.jpg)
How to Find aUniquely CH-Representable PrefixInspired by Binary Search
23 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
...
s t
NOYES NOYES
![Page 72: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/72.jpg)
How to Find aUniquely CH-Representable PrefixInspired by Binary Search
23 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
...
s t
NO
YES NOYES
![Page 73: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/73.jpg)
How to Find aUniquely CH-Representable PrefixInspired by Binary Search
23 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
...
s t
NO
YES NOYES
![Page 74: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/74.jpg)
How to Find aUniquely CH-Representable PrefixInspired by Binary Search
23 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
...
s t
NOYES
NOYES
![Page 75: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/75.jpg)
How to Find aUniquely CH-Representable PrefixInspired by Binary Search
23 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
...
s t
NOYES
NOYES
![Page 76: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/76.jpg)
How to Find aUniquely CH-Representable PrefixInspired by Binary Search
23 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
...
s t
NOYES NO
YES
![Page 77: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/77.jpg)
How to Find aUniquely CH-Representable PrefixInspired by Binary Search
23 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
...
s t
NOYES NO
YES
![Page 78: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/78.jpg)
How to Find aUniquely CH-Representable PrefixInspired by Binary Search
23 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
...
s t
NOYES NOYES
![Page 79: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/79.jpg)
How to Find aUniquely CH-Representable PrefixInspired by Binary Search
23 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
...
s t
NOYES NOYES
![Page 80: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/80.jpg)
Find Uniquely CH-RepresentablePrefixInspired by Binary Search
24 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Find uniquely CH-representable prefix of path 〈u1, . . . ,un〉(`,m, r ) := (1,n,n) Theorem:
Found prefix contains unique shortest prefix or more.while `+ 1 < r do
if 〈u1, . . . ,um〉 is uniquely CH-representable then ` := melse r := mm := b`+ 1 + rc
odreturn 〈u1, . . . ,u`〉
⇒ Checks O(log n) times whether uniquely CH-representable.
![Page 81: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/81.jpg)
How to Check Whether Path isUniquely CH-Representable
25 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
s t
Perform modified bidirectional upward Disjktra search
![Page 82: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/82.jpg)
How to Check Whether Path isUniquely CH-Representable
25 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
s t
Perform modified bidirectional upward Disjktra search
![Page 83: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/83.jpg)
How to Check Whether Path isUniquely CH-Representable
25 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
s t
Perform modified bidirectional upward Disjktra search
Detect multiple top nodes
![Page 84: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/84.jpg)
How to Check Whether Path isUniquely CH-Representable
25 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
s t
Perform modified bidirectional upward Disjktra search
Detect non-unique subpaths
![Page 85: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/85.jpg)
How to Check Whether Path isUniquely CH-Representable
25 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
s t
Perform modified bidirectional upward Disjktra search
This is very fast: Hierarchy is flat and sparse.
![Page 86: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/86.jpg)
Compression Rate with CH
26 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Lemma:Unique shortest path ⇒ uniquely CH-representable.
:
Consequence:In theory, CH-based representation can have less via nodes.
![Page 87: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/87.jpg)
Compression with Via NodesCH-based Compression – Summary
27 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Algorithm:Frame-Algor. + binary Scheme + bidir. search in CH
Fast:– perform bidir. search O(# via nodes · log |P|) times– upward search in flat and sparse hierarchy
Compression rate:(in theory) even better
![Page 88: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/88.jpg)
28 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
ExperimentsCompression Time and Rate.
![Page 89: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/89.jpg)
Experimental Setup
29 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
German road networkNodes: 4.7 millionEdges: 10.8 million, 7.2 % time-dependent
Simulating of Client and ServerDifferent metrics......for route planning and compression
Evaluated compression algorithmsFrame-Algor. + modified DijkstraFrame-Algor. + binary Scheme + bidir. search in CHProblem: Our CH-based implementation pessimistic
![Page 90: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/90.jpg)
Four Metrics (1)Metric = Edge-Weights + Objective
30 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
time-dependentEdge-Weights: Functions fe : time 7→ ∆ travel timeObjective: Find earliest arrival route for departure time τ0
free flowEdge-Weights: Time-independent travel times min feObjective: Find shortest path
![Page 91: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/91.jpg)
Four Metrics (2)Metric = Edge-Weights + Objective
31 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
distanceEdge-Weights: Driving distance dde
Objective: Find shortest path
energyEdge-Weights: Estimate energy cost dde + γ min feObjective: Find shortest path
with γ := 4 1 km costs 0.1e1 hour costs 14.4e
![Page 92: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/92.jpg)
Results (1)Server-Metric: Time-Dependent
32 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
# route client via nodes timenodes metric method # max. rate[%] [ms]
996
free flow Dijkstra 0.071 3 0.006 1 500.78CH-based 0.068 3 0.006 0.36
distance Dijkstra 9.771 26 1.045 481.60CH-based 9.677 25 1.036 20.98
energy Dijkstra 1.103 6 0.125 1 326.17CH-based 1.094 6 0.124 1.72
Fast with CH: less than 21 msSlow with Dijkstra: up to 1.5 sGood compression: max 26 via nodes, avg < 102 bit/edge need ≈ 248 byte, 10 via nodes need ≈ 104 byte
![Page 93: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/93.jpg)
Results (2)Server-Metric: Distance
33 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
# route client via nodes timenodes metric method # max. rate[%] [ms]
1 763free flow Dijkstra 29.312 76 1.689 162.49
CH-based 29.284 76 1.688 12.56
energy Dijkstra 24.902 69 1.434 182.87CH-based 24.876 69 1.433 15.63
Fast with CH: less than 16 msSlow with Dijkstra: up to 183 msStill good compression: max 76 via nodes, avg < 30Dijkstra faster with more via nodes
![Page 94: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/94.jpg)
Results (3)CH-Based: Running Time
34 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
●●●●●●●●●●
●
●
●
●
●
●●
●●●●
●●●
●
●
●
●●
runn
ing
time
[ms]
Dijkstra rank
26 28 210 212 214 216 218 220 222
0.01
0.1
1
10
100
●●●●●●
●
●
●●
●
●●●
●●●
●
●
●●●●●●●●●●●
●
●●
●●●●
●
●●●
●
●
●●
●●
●●●●●
●●●
●
●
●●
●
●●
●●●●
●●●●●●
●●●
●●
●●●●
●●
●
●●●
●●●●●
●●●●
●●●●●●●●●●●●●●●●●
●
●●●●●●●●●●●●●●●
●
●
●
●●
●●●●
●●●
●
●●●●●●●●●●●●●
●●
●●●
●●●
●●●●
●●●●
●●
server metric / client metric
time−dependent / free flowtime−dependent / energytime−dependent / distancedistance / energydistance / free flow
Runnig time increases with route lengthEven longest routes < 100 msRanks 212 – 214: middle sized German town, mostly < 1 ms
![Page 95: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/95.jpg)
Results (4)CH-Based: # Via Nodes
35 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
● ●●●
●●●●
●
●
●●
●
●
# vi
a no
des
Dijkstra rank
26 28 210 212 214 216 218 220 222
≤ 1
3
10
30
100
● ●● ●●●●●●●● ●●●●●●●●●
●
●
●
●●●●●●●
●
●
●
●●●●●●●●●
●
●●
●
●
●●
●
● ●●● ●●●●● ●●●●●
●
●●●
●●
●●●● ●●
●●
●●
●
●
●
●
●
●
●
●●●
●
●
●
●●●●●
●
● ●●●●
●
●●
●●
●●●●●●
●●
●
●●●
●●
●
●●●●●
●● ●●●●
●
●●
●
●●●●
●●
●●●●●●●●●●●
●●
●●
●●
●●●●
●●●●
server metric / client metric
time−dependent / free flowtime−dependent / energytime−dependent / distancedistance / energydistance / free flow
# via nodes increases with route lengthThere are outliers, but not seriosRunning time roughly increases with via nodes...... remember O(log |P| · # via nodes) bidir. searches
![Page 96: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/96.jpg)
Conclusions
36 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Experimental results:Fast (de)compression with CH: < 100 ms + 21 msFew via nodes: avg ≤ 30, max 76Compression with CH practically not better
Convenience of use:Driver experiences no delay......except for latency of mobile radio network
![Page 97: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/97.jpg)
Future Work
37 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Speedup Dijkstra-based compression (use ALT or ArcFlags)Faster CH-based without repeated bidir. searches?Non-pessimistic CH-based better?
Questions?
![Page 98: Efficient Route Compression for Hybrid Route Planning · 5 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov: Efficient Route Compression for Hybrid Route Planning Department](https://reader033.fdocuments.in/reader033/viewer/2022041512/5e291d74511084610d53ea0f/html5/thumbnails/98.jpg)
Future Work
37 G.V. Batz, R. Geisberger, D. Luxen, P. Sanders, and R. Zubkov:Efficient Route Compression for Hybrid Route Planning
Department of InformaticsInstitute of Theoretical Informatics, Algorithmics II
Speedup Dijkstra-based compression (use ALT or ArcFlags)Faster CH-based without repeated bidir. searches?Non-pessimistic CH-based better?
Questions?