Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System

Transcription

1 Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System Christian Lochert, Björn Scheuermann, Christian Wewetzer, Andreas Luebke, and Martin Mauve Heinrich Heine University Düsseldorf, Germany Volkswagen Group, Wolfsburg, Germany September, 15th 2008 San Francisco, CA, USA Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 1/22

2 Target of the study Global Targets Steps to proceed: Develop a navigation system based on current traffic information 1 Identify a suitable metric 2 Develop an aggregation scheme based on frequently updated floating data 3 Identify possible locations for supporting units Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 2/22

3 Target of the study Global Targets Steps to proceed: Develop a navigation system based on current traffic information 1 Identify a suitable metric 2 Develop an aggregation scheme based on frequently updated floating data 3 Identify possible locations for supporting units Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 2/22

4 Data and Metrics Assumptions Cars gather information by making observations about the time to pass a road segment Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 3/22

5 Data and Metrics Assumptions Cars gather information by making observations about the time to pass a road segment Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 3/22

6 Data and Metrics Assumptions Cars gather information by making observations about the time to pass a road segment Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 3/22

7 Data and Metrics Assumptions Cars gather information by making observations about the time to pass a road segment Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 3/22

8 Design Principle The Need for an Aggregation Scheme Quantity of information Information increases quadratically with the distance Bandwidth Packets can transport a limited amount of data only Network capacity has to be considered Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 4/22

9 Design Principle Implementation Multi level aggregation: 1 Hierarchical aggregation Cars have detailed knowledge about closer vicinity Cars have rough insight in regions far away 2 Aggregation based on the landmark principle Aggregation of street segments between two landmarks of the same hierarchy Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 5/22

10 Design Principle Landmark Aggregation 70s 20s 50s 40s 45s 25s 50s 35s 55s 125s Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 6/22

11 Design Principle Landmark Aggregation 70s 20s 50s 40s 45s 25s 50s 35s 55s 125s Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 6/22

12 Design Principle Landmark Aggregation 70s 20s 50s 40s 70s 45s 25s 50s 35s 55s 125s Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 6/22

13 Design Principle Hierarchical Aggregation P Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 7/22

14 Design Principle Hierarchical Aggregation P Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 7/22

15 Supporting Units Characteristics of a VANET Only few vehicles are equipped with VANET technology (especially during roll-out) Equipment density of VANET technology is very low Network of equipped cars is partitioned into multiple parts Supporting units Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 8/22

16 Supporting Units Functionality Behave like stationary cars Receive information by passing cars Supporting units are linked Share a common knowledge base Broadcast common knowledge Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 9/22

17 Supporting Units Placement Where to place supporting units? How many units are needed? How to test the effects of positions? Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 10/22

18 Placement Scenario Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 11/22

19 Placement Scenario Junctions Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 11/22

20 Placement Preparations for Evaluation Identify 100 possible locations for supporting units Vary the number of supporting units Test different subsets of these variations Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 12/22

21 Placement Random Placement of Supporting Units Junctions 2000 Streets Possible Positions for SUs Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 13/22

22 Genetic Algorithm Hard optimization problem Range of possible placements is huge: With 10 active supporting units: With 30 active supporting units: Fast optimization is needed! Genetic Algorithms Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 14/22

23 Genetic Algorithm Hard optimization problem Range of possible placements is huge: With 10 active supporting units: With 30 active supporting units: Fast optimization is needed! Genetic Algorithms Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 14/22

24 Genetic Algorithm Genetic Algorithm Solving the optimization problem by using genetic algorithms Representation of possible SU locations with 4 active SUs SU-vector SU 0 SU 1 SU 2 SU 3 SU 4 SU 5 SU 6 SU 7 SU 8 SU 9 A B C Fitness calculation and selection for recombination SU-vector Fitness Rank A B C Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 15/22

25 Genetic Algorithm Genetic Algorithm Solving the optimization problem by using genetic algorithms Representation of possible SU locations with 4 active SUs SU-vector SU 0 SU 1 SU 2 SU 3 SU 4 SU 5 SU 6 SU 7 SU 8 SU 9 A B C Fitness calculation and selection for recombination SU-vector Fitness Rank A B C Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 15/22

26 Evaluation Expected Results Low equipment ratio (5 %) reduces accuracy Important information is gathered primarily in the city center A small number of SUs provides a significant benefit Comparison between: Journey time without traffic information Journey time with disseminated traffic information Benefit is the objective function for the genetic algorithm Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 16/22

27 Results Location of ten active SUs Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 17/22

28 Results Location of thirty active SUs Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 18/22

29 Results Effects of Supporting Units Relative travel time Static Route Dynamic Route Number of supporting units Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 19/22

30 Results Effects of Supporting Units (CDF) SU 30 SU 10 SU 5 SU 0 SU Fraction of nodes Relative travel time (CDF) Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 20/22

31 Results Effects of Supporting Units (CDF) SU 30 SU 10 SU 5 SU 0 SU Fraction of nodes Relative travel time (CDF) Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 20/22

32 Summary Analysis of an navigation system Development of an aggregation scheme for 2D traffic information data based on Hierarchical aggregation Landmark routing Generation of a Genetic Algorithm to evaluate The number of supporting units The placement of supporting units Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 21/22

33 Thank you for your attention! Questions? Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 22/22

34 Example Car Junctions Streets Standard Route 2000 AppSim Route Global Route Std journey time: s Aggr journey time: s Global journey time: s Journey time savings: s (0.81) Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 23/22

35 Example Car Junctions Streets Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 23/22

36 Evolution of SU vector with 30 active SUs Number of vectors Location ID Generation Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 24/22

37 Evolution of SU vector with 10 active SUs Number of vectors Location ID Generation Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 25/22

38 Toolchain ns-2 VISSIM movement + traffic log file individual SU vector random initial SU vectors application simulator time savings (fitness) genetic algorithm optimal SU vector Data Aggregation and Roadside Unit Placement for a VANET Traffic Information System 26/22