VANETs. Marc Torrent-Moreno, Prof. Hannes Hartenstein Decentralized Systems and Network Services Institute for Telematics, University of Karlsruhe

Transcription

1 VANETs Marc Torrent-Moreno, Prof. Hannes Hartenstein Decentralized Systems and Network Services Institute for Telematics, University of Karlsruhe April 15 th 2005 Marc Torrent Moreno 1

2 Main Motivation (In all American presentations ;-)) According to the U.S. National Highway Traffic Safety Administration (2003):» ~6.3 million police-reported traffic accidents» ~43,000 people were killed» ~3 million persons injured» Annual economic impact of traffic-related accidents: >$230 billion Automotive safety systems trend go from passive to intelligent and active Preventive Marc Torrent Moreno 2

3 Where communications can help * Source: DaimlerChrysler AG main focus:safety Marc Torrent Moreno 3

4 Technology Final Goals» Safety on Roads (huge list of situations/applications considered) Reducing accidents Alleviating accident damages» Traffic Conditions Improve transport efficiency Monitor traffic demand» Environment Reduce traffic congestion Reduce pollution» Driving Comfort Driver assistance Infotainment applications Marc Torrent Moreno 4

5 Overview» VANETs, Introduction Hot topic all over the world VANET ACM Workshop Network on Wheels Project» What our view is What we are focusing on Marc Torrent Moreno 5

6 VANETs, Intro» Vehicular Ad-Hoc Network comes from well known concept MANET, but considers: Vehicle-to-vehicle communications Vehicle-to-infrastructure communications» Getting relevance since Oct 02: FCC allocates 75MHz of spectrum to improve safety, efficiency and comfort on roads» Currently both industry and academia involved in ambitious projects motivated by governmental sponsorships Marc Torrent Moreno 6

7 Related Conferences/Workshops» IEEE Vehicular Technology Conference (VTC)» IEEE Intelligent Vehicles Symposium (IV)» IEEE Intelligent Transportation Systems (ITS)» ACM International Workshop on Vehicular Ad Hoc Networks (VANET)» International Workshop on Intelligent Transportation (WIT) Marc Torrent Moreno 7

8 1st ACM workshop on Vehicular Ad-Hoc Networks (VANET 2004)» Ultimate goal: to explore the development of wireless vehicular ad hoc networking technologies» Main interest: safety and commercial applications enabled by short to medium range communication systems and/or networks (vehiclevehicle or vehicle-roadside)» Call for papers: new applications, networking protocols, security paradigms, network management technologies, power control, modulation, coding, channel modeling, etc.» Impression from attendance: Emerging and motivated research community Lots of interest from industry 2005: Marc Torrent Moreno 8

9 Ongoing Projects Over The World» USA: Vehicle Safety Communications Consortium (VSCC) DSRC/WAVE Technology (all info, up to date) (standarization)» Europe: Car to Car Communication Consortium PReVENT CarTalk Network on Wheels (Germany) Japan: ITS Japan Marc Torrent Moreno 9

10 Projects Main Goals» Assess today s and future applications» Specify technology/system» Push standardization process» Find adequate market introduction strategies Marc Torrent Moreno 10

11 Network on Wheels Project» Picking up FleetNet project results (Demonstrator deployed)» Founded by German Ministry of Education and Research (BMB+F)» Started in June 2004, it will last until May 2008» Main goals: Solve technical challenges related to robustness, performance, scalability and security for inter-vehicle communications Design base architecture and platform for VANET technology Explore applications for market introduction Marc Torrent Moreno 11

12 VANETs Remarks (I)» OFDM decided as the lower layer technology (~802.11a)» USA dedicated 75MHz at 5,9Ghz band. 7 Channels of 10Mhz» Europe (C2CCC) in the process of getting some bandwidth» For car manufacturers costs are a major concern: All cars will have GPS receiver Only one transceiver? (channel hopping?) Base system without Digital Map? (benefit for safety, non-safety?)» USA they are taking decisions faster: FCC defined 3 priorities in its report : Safety-of-Life, Safety, Non-Safety 1 Control channel where safety applications should reside 6 Service channels to support non-safety traffic Standardization bodies involved: IEEE, ASTM... Marc Torrent Moreno 12

13 VANETs Remarks (II): Data traffic estimation req. Application Inters. Collision Warning/Avoidance Coop. Collision Warning Work Zone Warning Transit Vehicle Signal Priority Toll Collection Packet Size (B)/Bwth ~100 ~100 ~10Kbps ~100 ~1Kbps ~100 ~100 Allowable Latency (ms) ~100 ~100 ~1000 ~1000 ~50 Network Traffic Type Event Periodic Periodic Event Event Comm. Range (m) Still, applications requirements are not standardized Priority Safety of Life Safety of Life Safety Safety The gap between Applications and Communications must be filled Non-Safety Service Announc. ~100 ~2Kbps ~500 Periodic 0 90 Non-Safety Movie Download (2 hours of MPEG 1) /10 min. down. time > 20Mbps N/A N/A 0 90 Non-Safety * Source: California PATH Marc Torrent Moreno 13

14 Marc Torrent Moreno 14

15 How do we see Uni-Karlsruhe? VANETs are characterized by:» Wide spectrum of applications» Type of communication (end-to-end notion revised)» Self-organization and self-management (fully decentralized)» Packets vs. Information (state / data aggregation)» Network protocol requirements (efficient geo-casting/flooding)» Adverse medium conditions (congestion and radio channel) Marc Torrent Moreno 15

16 What is our Our motivation:» It is demonstrated that real-world conditions have significant impact on protocol performance» There is no proof of well (idealistically) studied protocols to work in real-world» Unfortunately, VANETs will face worst-conditions scenarios Our Goal:» Find best strategies for VANETs protocols to fight against unreliable wireless channels Marc Torrent Moreno 16

17 What elements do we need to perform this work? (simulation)» Realistic radio propagation models» Realistic traffic distributions and movement patterns» Verified simulator What do we have now?» Some realistic highway movement patterns (Courtesy of DC Germany)» Realistic radio propagation model at highway scenarios using Nakagami distribution (Courtesy of DC Palo Alto)» Verified ns MAC and PHY layer (Uni-Karlsruhe) adjusted to real a Atheros Chips (values courtesy of Atheros) but we need more Marc Torrent Moreno 17

18 First Problem study: Broadcast Reception Rates and Effects of Priority Access in Based Vehicular Ad-Hoc Networks Marc Torrent-Moreno, University of Karlsruhe Daniel Jiang, DaimlerChrysler RTNA, Inc. Hannes Hartenstein, University of Karlsruhe 1st ACM workshop on Vehicular Ad-Hoc Networks, Oct 04 Marc Torrent Moreno 18

19 Problem Statement Broadcasts emergency message What is the probability of reception for this car? But:» All vehicles send data contending for the channel» Hidden terminal problem» Channel characteristics Marc Torrent Moreno 19

20 Channel modeling: the standard way Unit Disc Graph Model c.r. If no interference: Cars inside communication range receive the packet Cars outside communication range do NOT receive the packet Marc Torrent Moreno 20

21 In reality Reality c.r. If no interference: Cars inside communication range CAN receive the packet Cars outside communication range MAY receive the packet Reception Probability (d)? Marc Torrent Moreno 21

22 Dynamic Scenario (mobile and TRG or Nakagami) 1 50m distance 100m distance 200m distance Distribution m Received power (dbm) Nakagami received power distribution - 8 lanes, 2 direcctions (912 cars) - Comm. Range: 100m, 200m - Constant speed different lanes - Packet size: 200B, 500B - Distance between cars: 20m - 10 pckts/s Marc Torrent Moreno 22

23 Dynamic Scenario Metrics Safety Applications Broadcast Scenario Delay Effectiveness» Metrics: Channel access time Probability of reception Marc Torrent Moreno 23

24 Channel Access Time Scenario Channel Access Time Com. Range Pckt Size Load Mbps Priority Non-Priority 100m 200 B ms 0.9 ms 100m 500 B ms 4.8 ms 200m 200 B ms 3.9 ms 200m 500 B ms 16.4 ms 200m (Nak) 500 B ms 26.5 ms» Clearly the Priority node has always shorter Channel Access Time! Marc Torrent Moreno 24

25 Probability of Reception (not saturated scenario)» 200 bytes packets» 200 m intended communication range Marc Torrent Moreno 25

26 Result of the study» Results Priority access provides improvement in channel access time (AIFS and CW can be adjusted) When saturation occurs, nodes experience severe performance degradation (probability of reception can be as low as 20% at half com. range) Non-deterministic radio model degrades performance of both types of nodes (worse effect on probability of reception of the prioritized ones)» Conclusions Need better understanding of distributed MAC on global scale Priority access methods as well as relay/repetition strategies are very important aspects of future safety-critical applications * For more detail, please take a look at the paper Marc Torrent Moreno 26

27 Our research right now 4 ongoing research studies» Highway routing with realistic radio propagation models» Effect on reception rates of new capture effect» Robust strategies for data dissemination (flooding)» Fair strategies to limit load of safety applications Marc Torrent Moreno 27

28 Thank you very much for your attention! Marc Torrent Moreno 28