Transcript Document

CS 672 Paper Presentation
“A Survey of Inter-Vehicle Communication”
Jun Luo, Jean-Pierre Hubuax
EPFL, Switzerland
Technical Report IC/2004/24
Presented By
Saif Iqbal
IVC - The Big Picture
About
•Component of Intelligent Transportation System (ITS)
•One of the concrete applications of MANETS
•Attracted research attention in US, EU, and Japan
Motivation
•Improves road safety and efficiency by increasing the horizon of drivers
and on-board devices
•Transmission of road-side information about emergencies, congestion,
etc.
•Ability for inter-driver communication
•Existing ad hoc networks protocols and experiences can actually be put
to practice
IVC - The Big Picture
Groups & Applications
•Association of Electronic Technology for Automobile Traffic and
Driving (JSK), Japan - early 1980’s
•CarTALK, EU - 2000
•FleetNet, Germany - 2000
•PATH, California
•Chauffeur, EU
•DEMO 2000, Japan
IVC – Main Applications
Information and Warning Functions
Dissemination of road information to distant vehicles
Communication-based Longitudinal Control
Exploiting “look-through” capacity to avoid accidents, platooning
vehicles, etc.
Co-operative Assistance Systems
Coordinating vehicles at critical points
Added-value Applications
Internet access, Location-based services, Multiplayer games
Paper
•Surveys IVC with respect to key-enabling technologies
•Focuses the discussion on the various MAC protocols that are really
important for IVC
•Analyzes application requirements and protocols and comes up with
suggestions with regards to the direction of future work
•Does not describe applications, but mentions them when their enabling
mechanisms are discussed
Radio Frequency Spectrum
•Both infrared and radio waves have been studied and employed
•Radio waves: VHF, micro, and millimeter waves
•VHF and microwaves are of broadcast type
•Dedicated Short Range Communication (DSRC) spans 75MHz of
spectrum in the 5.9 GHz band
•DEMO 2000, Chauffeur used 5.8 GHz DSRC
•CarTALK, FleetNet use ULTRA TDD
•JSK, PATH, CarTALK have used infrared, typically for cooperative
driving
MAC/PHY Layer
WLAN approach
•802.11, Bluetooth
•Plus: inherent support for distributed coordination in ad hoc mode
•Minus: low flexibility in radio resource assignment and transmission rate
control
3G approach
•CDMA
•Plus: high granularity for data transmission and flexible assignment of
radio resources
•Minus: complexity in designing the the coordination function in ad hoc
mode
WLAN
•Can directly use WLAN standards for RVC, but does not work as well
for IVC
•For migrating to IVC the following problems need to be tackled
-Resistance to more severe multipath effects
-Time synchronization between rapidly moving nodes
-Distributed resource allocation
Token Ring Protocol - PATH
•Mechanism to construct, recover, join, and leave a ring
•Token circulation, recovery, and multiple token resolution
•Solves contention for radio resources
•Need further experiments to see if this suitable for IVC
WLAN
Location-based Channel Access (LCA)
•Divides the geographical area into cellular structure
•Each cell has a unique channel associated with it
•Any multiple access scheme like CSMA, CDMA, TDMA can be used
within the cell
•Adaptability to high mobility in IVC is a question mark
Non-Persistent CSMA – DOLPHIN/DEMO 2000
•DOLPHIN – Dedicated Omni-purpose inter-vehicle communication
Linkage Protocol for Highway Automation
•Non-persistent CSMA outperforms p-persistent one with respect to
packet loss in IVC
3G
•Cannot directly employ it as it is meant for centralized cellular networks
•For extending to IVC the following problems need to be addressed
-Distributed radio resource management
-Power control algorithms
-Time synchronization
•Solution should rely on distributed media access control
Reservation ALOHA – CarTALK/FleetNet
•Use R-ALOHA for distributed channel assignment
•High throughput, as a node which catches a slot can use it in subsequent
frames as long as it has packets to send
•Frequent reservation attempts due to short packet trains
•Due to hidden terminal problem, destructive interference with
established channels may occur
•Reliable R-ALOHA lets all nodes know the status of the slots
3G
Random Access CDMA
•Combines CDMA with random channel access – start transmission
immediately irrespective of the state of the channel
•Can avoid primary collisions (two nodes with the same code try to
access the channel together) using code assignment, spread-coding
schemes
•Multi-access interference (MAI) results in secondary collisions (near-far
problem) at the receiver
Controlled Access CDMA
•Uses a modified RTS/CTS mechanism to solve the problem
•Split the channel for control and data – RTS/CTS is transferred over
control channels to let the interfering nodes be aware of the status
•Use knowledge of power levels to alleviate the near-far problem
MAC/PHY Layer
Summary
•Number of MAC protocols proposed, but not all of them are put into
practice
•802.11b is used for demonstration (FleetNet)
•802.11a is chosen by ASTM (American Society for Testing and
Materials) as the basis for DSRC
•MAC Protocol based on ULTRA TDD, used in CarTALK, could be
another solution especially in the EU
Network Layer
•Any existing position-based routing protocol for ad hoc networks can be
applied to IVC
•Can optimize by using location and surrounding awareness information
•Mostly group-oriented communications rather than pair-wise
Unicast Routing
Contention-Based Forwarding
•Proposes a forwarding scheme avoiding the use of beacons and hence
higher efficiency
Reactive Location Service - FleetNet
•Reactive routing protocol by requesting the location of the destination
when sending a packet
•Afterwards, a greedy geographical forwarding technique is used
Network Layer
Broadcast Routing
•Used to disseminate traffic information
•Can be made adaptive – change the inter-transmission interval
•Use a randomized interval
•Multi-resolution data structure to express information in the message
Summary
•Unicast routing is superfluous in most cases
•Broadcast routing seems to be a necessary supporting mechanism of
IVC applications
•Broadcast routing could also be optimized in various ways
Group Communication
•Group communication primitives are important as reliability required for
critical situations, where group information dissemination may not work
•Required by two important applications – platooning and cooperative
driving
Localized Group Membership Service (LGMS)
•Reduce the group membership service to the local environment of a
node
•Only tracks memberships of neighbors and installs a local view at each
node
•Works for congestion area detection
•Does not support any functions with reliability requirement due to lack
of a global view of the group
Group Communication
Event-based middleware
•Supports cooperative mobile applications
•Underlying membership service is costly and not really needed
•Implemented only on RVC scenarios
The Driving Philosopher Problem
•Sharing resources among a group of vehicles
•Proposed algorithm solves the model in a synchronous model
•Impossible to achieve fairness and concurrency at the same time
•Impossible to solve the problem in an asynchronous model
Summary
•Important component of IVC
•Build the system directly upon the MAC layer
•Lightweight membership tracking instead of doing it globally
Security
Driver Ad Hoc Networking Infrastructure (DAHNI)
•System mounted on each vehicle which includes both processing and
wireless communication facilities
•Each car constitutes a local communication network around itself
•Come to the conclusion that no confidentiality is needed, and ignore
privacy concerns
•To securely estimate the distance between vehicles, establishment of
symmetric keys is required
Electronic License Plates
•Certified identity that a vehicle provides via a wireless link
•Usage: Dynamic toll charge, identify culprits, distance estimation
•Attacks: Disable system, impersonation, denial of service
•Supports cooperative driving
Security
Summary
•Security of IVC has been ignored so far by the research community
•Some ideas proposed but not followed through and implemented
•Emerging and potentially important research topic
Mobility Model
•Mobility pattern in IVC is quite different from the “random waypoint”
model used for ad hoc networks simulation
•Simulations for MAC protocols should also take mobility into account
which is not necessarily the case with traditional MAC
•There are proposed tools for traffic simulation to help extend the
network simulators
•Application context has to be taken into account before choosing a
mobility model to evaluate certain protocols
Mathematical Modeling for Traffic
•Microscopic – suitable for simulating group communications
•Macroscopic – for routing protocols discussion
•Statistical – real mobility pattern is in 2-d or 3-d space, approach based
on Markov chain theory
Conclusion
•Design of communication protocols in IVC is extremely challenging
•Ad hoc routing protocols, and group communication primitives
migrated from wired networks might not be efficient
•Most of the proposed work is in routing algorithms and MAC
•Routing protocols are unnecessary in most cases
•Local distributed coordination functions sitting directly upon the
MAC would be more efficient
•As vehicles become “smarter”, security and privacy gain importance
•Mathematical models for road traffic are important for simulations