(Slides) A Method for Sharing Traffic Jam Information Using Inter-Vehicle Communication
Download as PPTX, PDF2 likes3,120 views
The document presents a method for sharing traffic jam information through inter-vehicle communication, addressing economic losses caused by traffic congestion. It outlines a system for autonomous data collection using GPS, enabling vehicles to estimate arrival times and navigate around congested areas without relying on costly infrastructure. The proposed method aims to improve the accuracy of traffic information by minimizing redundant data processing and prioritizing data transmission between vehicles.
![A Method for
Sharing Traffic Jam Information
using Inter-Vehicle Communication
Naoki Shibara, Takashi Terauchi*, Tomoya Kitani*,
Keiichi Yasumoto*, Minoru Ito* and Teruo Higashino**
Shiga University, JAPAN
*Nara Institute of Science and Technology (NAIST), JAPAN
**Osaka University, JAPAN
[ Invited Paper ]
The 2nd International Workshop on
Vehicle-to-Vehicle Communications 2006 (V2VCOM 2006), 21 July, 2006](https://image.slidesharecdn.com/060725-shibata-v2vcom06-091112191049-phpapp01/85/Slides-A-Method-for-Sharing-Traffic-Jam-Information-Using-Inter-Vehicle-Communication-1-320.jpg)
![Shibata, N., Terauchi, T., Kitani, T., Yasumoto,
K., Ito, M., Higashino, T.: A Method for
Sharing Traffic Jam Information Using Inter-
Vehicle Communication, The 2nd International
Workshop on Vehicle-to-Vehicle
Communications (V2VCOM) (Mobiquitous2006
Workshop), pp. 1-7．
DOI:10.1109/MOBIQ.2006.340428 [ PDF ]
32July 21, 2006 V2VCOM 2006 - N. Shibata et al.](https://image.slidesharecdn.com/060725-shibata-v2vcom06-091112191049-phpapp01/85/Slides-A-Method-for-Sharing-Traffic-Jam-Information-Using-Inter-Vehicle-Communication-32-320.jpg)
(Slides) A Method for Sharing Traffic Jam Information Using Inter-Vehicle Communication
- 1. A Method for Sharing Traffic Jam Information using Inter-Vehicle Communication Naoki Shibara, Takashi Terauchi*, Tomoya Kitani*, Keiichi Yasumoto*, Minoru Ito* and Teruo Higashino** Shiga University, JAPAN *Nara Institute of Science and Technology (NAIST), JAPAN **Osaka University, JAPAN [ Invited Paper ] The 2nd International Workshop on Vehicle-to-Vehicle Communications 2006 (V2VCOM 2006), 21 July, 2006
- 2. 2July 21, 2006 V2VCOM 2006 - N. Shibata et al. Background $100 billion/year of economic loss by traffic jam Existing services for alleviating traffic jam VICS (Vehicle Information and Communication System) A service provided by Japanese government Provides traffic jam information of trunk roads Cyber-Navi （Pioneer） An advanced car-navigation system Predicts traffic jam from information in the past Weaknesses of the existing methods Service area Operation cost Time lag / not up-to-date information
- 3. 3July 21, 2006 V2VCOM 2006 - N. Shibata et al. Purpose of the study Autonomous collection of statistical traffic info. Using GPS and inter-vehicle communication No need for costly infrastructures Prediction of arrival time using the gathered info. Provides basic car-navigation system functions Route navigation avoiding congested areas
- 4. 4July 21, 2006 V2VCOM 2006 - N. Shibata et al. A1 A2 A3 A6A5A4 A7 A8 A9 Overview of the method (1/2) The map is divided into areas. Granularity of area is decided according to the amount of traffic and road density. Cars measure time to pass through each area using GPS. Traffic info. can be collected by exchanging info. between cars.
- 5. 5July 21, 2006 V2VCOM 2006 - N. Shibata et al. Overview of the method (2/2) Arrival time can be estimated by summing up time to pass through areas A1 A2 A3 A6A5A4 A7 A8 A9 1’20’’ 1’40’’ S D tA7 tA4 tA5 tA2 tA7’ tA8 tA5’ tA2’
- 6. 6July 21, 2006 V2VCOM 2006 - N. Shibata et al. Outline Step 1. Collection of traffic info. Collecting time to pass through each area Propagation and retention of collected info. Avoiding redundant processing of same data Exchanging data with priorities Step 2. Estimating arrival time
- 7. 7July 21, 2006 V2VCOM 2006 - N. Shibata et al. Challenges for accuracy Distinguishing time between turning right/left and going straight is crucial Especially, time needed to turn right is largely influenced by the opposite lane
- 8. 8July 21, 2006 V2VCOM 2006 - N. Shibata et al. Problem of per-link measurement • How can time to turn right/left and go straight be distinguished? Collecting info. on a per-lane basis is hard Too much data by collecting on a per-link basis
- 9. 9July 21, 2006 V2VCOM 2006 - N. Shibata et al. Per-area measurement Collecting data for each combination of incoming and outgoing link Time to turn at crossings is reflected Area border 24 sec 38 sec
- 10. 10July 21, 2006 V2VCOM 2006 - N. Shibata et al. Time to pass through area Measuring time for each combination of two links crossing area border G E A H I B D C F a e d c b Blue links cross area border Red dotted lines are area borders Links Measured time ( a , b ) 150 sec ( a , e ) 220 sec .... For the area at the center, time is measured for each two combination of blue links
- 11. 11July 21, 2006 V2VCOM 2006 - N. Shibata et al. Outline Step 1. Collection of traffic info. Collecting time to pass through each area Propagation and retention of collected info. Avoiding redundant processing of same data Exchanging data with priorities Step 2. Estimating arrival time
- 12. 12July 21, 2006 V2VCOM 2006 - N. Shibata et al. Area border Exchanging info. between cars When a car crosses area border, it broadcasts its info When another car receives this info, it adds the info into its memory The car continues to broadcast data for some time
- 13. 13July 21, 2006 V2VCOM 2006 - N. Shibata et al. Adjacent areas The original area the info is generated at Retention and disposal of data • Info. for an area adjacent to the area a car is running at is retained.
- 14. 14July 21, 2006 V2VCOM 2006 - N. Shibata et al. Retention and disposal of data Here, this car disposes info. • Info. for an area adjacent to the area a car is running at is retained. • When a car goes off from these areas, it disposes the corresponding data
- 15. 15July 21, 2006 V2VCOM 2006 - N. Shibata et al. Outline Step 1. Collection of traffic info. Collecting time to pass through each area Propagation and retention of collected info. Avoiding redundant processing of same data Exchanging data with priorities Step 2. Estimating arrival time
- 16. 16July 21, 2006 V2VCOM 2006 - N. Shibata et al. Redundant counting of same data Average area passage time to be obtained Data from multiple cars have to be collected Data originated from one car is exchanged through many cars, and treated as multiple data A A (A+B)/2 (A+C)/2 (A+A+B+C)/4 ?!
- 17. 17July 21, 2006 V2VCOM 2006 - N. Shibata et al. Avoiding the problem Avoiding redundant processing of area passage time Measured time by each car is not processed until a car gets a sufficient number of data Then, data are converted to statistical data Ignore received data if it has a same set of data as data retained. Avoiding redundant processing of statistical data Distinguish redundant data by hash value
- 18. 18July 21, 2006 V2VCOM 2006 - N. Shibata et al. Avoiding redundant counting 5 sets of data7 sets of data AA
- 19. 19July 21, 2006 V2VCOM 2006 - N. Shibata et al. Avoiding redundant counting AA If a same set of data is included in the received data, it's ignored.
- 20. 20July 21, 2006 V2VCOM 2006 - N. Shibata et al. Exchanging data between cars The case when threshold C = 10 5 sets of data 7 sets of data
- 21. 21July 21, 2006 V2VCOM 2006 - N. Shibata et al. Exchanging data between cars The case when threshold C = 10 5 sets of data7 sets of data C sets of data is converted to one statistical data 10 Remainder
- 22. 22July 21, 2006 V2VCOM 2006 - N. Shibata et al. AA Avoiding redundant processing of stat. data Each stat. data include hash value calculated from car IDs Stat. data A A Compare hash value, and discard if it has already the same data
- 23. 23July 21, 2006 V2VCOM 2006 - N. Shibata et al. Outline Step 1. Collection of traffic info. Collecting time to pass through each area Propagation and retention of collected info. Avoiding redundant processing of same data Exchanging data with priorities Step 2. Estimating arrival time
- 24. 24July 21, 2006 V2VCOM 2006 - N. Shibata et al. Exchanging information All information cannot be exchanged due to bandwidth limitation Data is prioritized based on where the car is running at, and how the areas are congested Mid High High HighMidLow Low Low Mid Position of the car
- 25. 25July 21, 2006 V2VCOM 2006 - N. Shibata et al. Evaluation Experiment 1 Gain of accuracy by avoiding redundant processing Experiment 2 Gain of accuracy by prioritized data transmission Experiment 3 Overall accuracy of estimated arrival time
- 26. 26July 21, 2006 V2VCOM 2006 - N. Shibata et al. Common configurations Area size 300 m×300 m Radio range 100 m Max speed of cars 16.6 m/s (60 km/h) Threshold C 5 Size of packet 1500 byte Max # of packets sent by a car 0.3 packets / sec Duration of each packet 0.01 sec Duration of simulation 1 hour Time to live of info. 30 minutes Number of cars 300 at maximum
- 27. 27July 21, 2006 V2VCOM 2006 - N. Shibata et al. Traffic simulator ：NETSTREAM by TOYOTA Map size ：1.2km×1.2km The number of nodes ： 21 The number of links ： 78
- 28. 28July 21, 2006 V2VCOM 2006 - N. Shibata et al. Gain of accuracy by avoiding redundant processing 0 20 40 60 80 100 120 A B C D E F G H I Link Pair PassageTime(sec.) No avoidance Our method Actual time If no avoidance, obtained time is much closer to the mode
- 29. 29July 21, 2006 V2VCOM 2006 - N. Shibata et al. Gain of accuracy by prioritized data transmission 0 20 40 60 80 100 A B C D E F G H I Link Pair PassageTime(sec.) No prioritizing Our method Actual time Data is lost Only two cars have data Loss of information is avoided by prioritizing
- 30. 30July 21, 2006 V2VCOM 2006 - N. Shibata et al. 0 20 40 60 80 100 120 140 (A,O) (B,P) (C,Q) (D,R) (E,S) (F,T) (G,U) (Area 1, Area 2) PassageTime(sec.) Estimated time vs actual time Estimated Actual A1 A2 A O B C P Q
- 31. 31July 21, 2006 V2VCOM 2006 - N. Shibata et al. Conclusion We proposed an autonomous collection method for traffic information We evaluated our method using the traffic simulator NETSTREAM Future works include Dissipating information to further area Making it works correctly with very few cars on the road
- 32. Shibata, N., Terauchi, T., Kitani, T., Yasumoto, K., Ito, M., Higashino, T.: A Method for Sharing Traffic Jam Information Using Inter- Vehicle Communication, The 2nd International Workshop on Vehicle-to-Vehicle Communications (V2VCOM) (Mobiquitous2006 Workshop), pp. 1-7． DOI:10.1109/MOBIQ.2006.340428 [ PDF ] 32July 21, 2006 V2VCOM 2006 - N. Shibata et al.
Editor's Notes
- #2: My name is Naoki Shibata. I'd like to have a talk titled ...
- #3: Traffic jam is a huge problem for the whole world. In Japan, we are having more than 100 billion dollar of economic loss for each year,. Traffic jam also causes lots of extra CO2 emission. There are some existing services for alleviating traffic jam. Japanese government is providing a service called VICS, which stands for Vehicle Information and Communication System. This provides traffic jam information of trunk roads, by means of FM radio and optical beacons. Some Japanese companies are selling advanced car-navigation systems utilizing VICS. For example, Pioneer sells a car-navigation system called Cyber-Navi. This system has a traffic jam information collected in the past in its HDD, and predicts traffic jam from the information in the past. But, these existing methods have some weaknesses. First, VICS doesn’t provide information other than trunk roads, since VICS collects information from sensors placed roadside. Second is the operation cost. VICS requires the sensors to be placed and maintained. Cyber-Navi requires update of statistical data. The last one is time lag. VICS collects data from sensors to a central server, and dissipates the information. So, there is some time lag. Cyber-Navi predicts traffic jam from the information in the past, so it can’t handle traffic jam caused by road construction works.
- #4: So, we want to alleviate this huge problem. The purpose of this study is to realize autonomous collection of statistical traffic information. Today, GPS receiver and wireless LAN interfaces are very cheap. Using GPS and inter-vehicle communication, traffic information can be collected without costly infrastructures on the ground. And, by using the collected information, we want to realize basic car-navigation system functions including route navigation avoiding congested areas.
- #5: Ok, now I explain the overview of the method. We first divide the map into areas, like this. Granularity of the areas are decided according to the amount of traffic and road density. Cars measure time to pass through each area using GPS. Traffic information can be collected by exchanging information between cars.
- #6: And, once we got time to pass through each area, arrival time can be estimated by summing up time to pass through areas.
- #7: Here is the outline of the proposed method. I’ll now explain collecting time to pass through each area
- #8: In order to estimate the time needed to pass through each area, we have to distinguish cars turning right/left and going straight. Especially, time needed to turn right is largely influenced by condition of the opposite lane. So, distinguishing the cases of turning right, left and going straight is crucial.
- #9: But, how can time of turning right, left and going straight distinguished? The simplest way is to collect information for each lane. But, it’s hard to distinguish which lane the car is running on due to GPS positioning error. Also, collecting information on a per-lane or a per-link basis is not sufficient. Besides these problems, if we collect data on a per-link basis, we will have too much data.
- #10: In the proposed method, we adopt per-area measurement. In this figure, these red dotted lines are area border. If a car goes straight, it needs 24 seconds to pass through this area. If another car turns right, it needs 38 seconds to pass through this area. By measuring time to pass an area for each combination of incoming and outgoing link, we make time to turn at crossings reflected in the data.
- #11: I now give another example. On this map, these red dotted lines are area borders. So, these five blue links named alpha to epsilon cross area borders. We measure time to pass through area for each combination of incoming and outgoing links. So there are 5 times 4 equals 20 combinations to measure time for the area at the center.
- #12: OK, now I explain propagation and retention of the collected information.
- #13: 読む
- #17: If we exchange data between cars by the simplest way, we will have the problem explained here. We have to collect data from multiple cars. And, data originated from one car is exchanged through many cars, and it can be treated as multiple data. For example, this red car first disseminates it’s measured time A. This data is received by these blue car and black car. They also have their measured time B and C. And then, they disseminate the averaged time (A+B)/2 and (A+C)/2. When these two data are received by this white car and averaged, the first data A is counted twice. We found this largely worsen the accuracy of the final estimated time.
- #18: In the proposed method, we avoid the problem in two steps. In order to avoid redundant processing of the area passage time, the measured time are not processed until a car gets a sufficient number of data. If the car receives same data as it possesses, the car just ignore it. If a car gets a sufficient number of data, it converts the data into statistical data. Redundant processing of the statistical data can be avoided using hash values.
- #19: For example, this black car has 7 sets of measured time. The red car has 5 sets of measured time. When these cars exchange data,
- #20: If a same set of data is included in the received data, all received data are ignored.
- #21: If there is no redundancy, the black car concatenates the received data with data in its memory, and it converts part of them into a statistical data.
- #24: Now, I explain exchanging data with priorities
- #25: 読む
- #26: We evaluated our method from three viewpoints. First of them is observing gain of accuracy by avoiding redundant processing. Second is the gain of accuracy by prioritizing data transmission. The third is overall accuracy.
- #27: These are common configurations for the experiments.
- #28: We used a traffic simulator NETSTREAM developed by TOYOTA
- #29: We observed ( タイトル )
- #30: Next, we observed (title) We observed lost data, and cases where the data are not dissipated as we expected.
- #31: We observed overall accuracy of estimated arrival time. Here, we observed actual time to pass these two areas, and estimated time to pass these areas by our method. These data shows that the error of estimation is within 10 seconds.
- #32: I now conclude my talk. 読む