A Multihop Mobile Networking Testbed for Telematics

1
A Multi-hop Mobile Networking Test-bed for
Telematics

• Carnegie Mellon University Rahul Mangharam
  (Speaker)
• Jake Meyers, Raj Rajkumar, Dan Stancil
• rahulm, jjmeyers, raj, stancil_at_ece.cmu.edu
• General Motors Jay Parikh, Hariharan Krishnan,
  Chris Kellum
• jayendra.s.parikh, hariharan.krishnan,
  christopher.kellum_at_gm.com
• April 2005

2
What this talk is about
Top-Down Approach
Vehicular Networking Applications
Network Protocol Design
Multi-hop Wireless Test-bed
Wireless Channel Protocol Performance
Bottom-Up Approach
3
O u t l i n e

• The Need for Wireless Vehicular Networking
• Unique Multi-hop Protocol Design Challenges
• GrooveNet - Geographic Routing
• Vehicular Networking Test-bed
• On-road Communication Performance Analysis
• Multi-hop Wireless Vehicular Applications

4
Experimental Ad Hoc Test-bed
5.9 GHz DSRC Dedicated Short Range
Communications Between vehicles
GPS
Differential GPS reference station beacons
Mobile Node

• Vehicle-to-Vehicle Multi-hop
• Vehicle-to-Mobile Gateway
• Vehicle-to-Infrastructure

5
Vehicular Networking Application Categories

• Safety Alerts
• Sudden Breaking
• Airbag deployment
• Skidding
• Traffic Congestion Probing
• Travel Time
• Dynamic Route Planning
• Road Condition Notification
• Interactive Applications
• Social Networking
• Multimedia Content Exchange
• Advertising

6
Geographic Routing Overview

• Why Geographic Routing?
• Events are locally relevant
• Path is more important than Destination
• Identity Position Reference
• Relevance Messages for specific target vehicles
• Direction Messages for on-coming vehicles only
• Zone-based Map is overlaid with a grid and
  divided into zones
• Navigation-based Route along specific road
  segments
• Packet Exchanges
• KISS DATA and ACK only
• No Handshaking Sequence
• Accept / Reject / Forward based on message and
  state info
• Restricted Flooding
• Triggers
• EVENT ? SEND_MSG ? RECV ACK
• RECV_MSG ? DECISION ? ACCEPT FWD_MSG DROP_MSG

7
Broadcast Scenarios
Highway Driving City
Driving Rural Driving

• Path with Intermediate points
• Static Source Routing

• Radial Broadcast

• Bounding Box
• Controlled Flooding

8
Routing Bounding Boxes
Requires a systematic study
9
GrooveNet Protocol Details

• Network foundation - caters to each application
  category
• Packet Types and Exchange Sequences
• Addressing
• Packet Formats
• Route Search Route Persistence Rules
• Route Table Representation
• State Information Neighbor Interaction

10
GrooveNet Network Architecture
Linux Application Layer
Delay A
Linux Networking Services
Delay B
Linux Networking Stack RT Filter
Delay C
Linux - Device Driver
MAC - Firmware
Delay D
PHY - Wireless Interface

• Fast Path Critical Messages
• Control Path Route Update Filtering

11
Packet Types Exchange Sequences

• DATA Packet with MESSAGE field
• Critical Alerts
• Non-Critical Updates
• Subscribed Services
• ACK Packet (Optional)
• No Route Request, No Route Reply, No Route!
• Broadcast, Multicast and Unicast DROP Rules
• Parse Message Maintain State Information

12
GrooveNet Vehicle Addressing

• Message Type Broadcast, Multicast, Unicast
• Filter Array
• Logical I want to talk to Buick FJF2323
• Geographic Message for Downtown Warren
• Non-Geographic If you have this capability.
• Hop count Message for - All cars within 6 hops
  radius
• Navigation Message for - All cars headed for
  Exit 22
• Direction Message for - All neighbors
  driving east on I-90
• Speed Message for - All cars moving
  within /- 8mph of my speed
• Connectivity Message for ANY mobile gateway

13
Route Search and Route Persistence Rules

• Scope Region of Validity for each packet
• Path with intermediate hops (optional)
• Greedy Routing
• Message Lifetime
• Retransmission frequency and event
• Nodes moving in opposite direction Forward
  packets
• Periodic HELLO Messages for Neighbor
  identification

14
GrooveNet Scalability Performance

• On Road Testing
• Implement in Linux as a kernel module
• Congestion Analysis
• Route Virtual Vehicles over Map with actual
  protocols
• Test for (a) Safety Alerts and (b) Congestion
  Probes
• Basic Connectivity Analysis
• Test-bed with straight road
• Determine end-to-end throughput delay
• Given speed distribution, max num nodes, density
  distributions

15
Simulation Demo
16
Simulation Demo
Over 1000 simultaneous vehicles simulated!
17
Experimental Ad Hoc Test-bed
5.9 GHz DSRC Dedicated Short Range
Communications Between vehicles
GPS
Differential GPS reference station beacons
Mobile Node

• Vehicle-to-Vehicle Multi-hop
• Vehicle-to-Mobile Gateway
• Vehicle-to-Infrastructure

18
GrooveNet Test Kit
LAPTOP COMPUTER w/ RADIO CARD
GPS RECEIVER
GPS ANTENNA
POWER CORD (12 DC)
5.8 GHz ANTENNAS
WEB CAM
HEADPHONES W/ MICROPHONE

• Driven 5 vehicles over 400 miles Urban, Rural
  and Highway
• Over 625,000 link measurements

19
Average Signal Attenuation vs. Distance

• Residential (Suburban) Environments
• (Pittsburgh, PA)

• Log-Distance Path Loss Model (PR(d) dBm
  PR(do) 10nlog10(d/do))
• Current data analysis suggests a path loss
  exponent of approximately 2

20
Packet Error Rate vs. Distance

• Residential (Suburban) Environments
• (Pittsburgh, PA)

• Packet Error Rate generally less than 10 at
  distances approaching 100 m
• Large PER spike at 10 m currently not
  understood

21
Packet Error Rate vs. Absolute Speed

• Residential (Suburban) Environments
• (Pittsburgh, PA)

• Packet Error Rate generally less than 10 at
  speeds up to 30 MPH
• No observed difference between effects of
  transmitter and receiver speed

22
Packet Error Rate vs. Relative Speed

• Residential (Suburban) Environments
• (Pittsburgh, PA)

• Packet Error Rate generally less than 10 at
  speeds up to /- 30 MPH
• Relative speed does not appear to have a
  significant effect on PER values

23
Summary and Conclusions

• The Time has come for Vehicular Networking
• Several Killer Apps! Safety, Congestion
  Probing, Collaborative Driving, Ads
• Robust Vehicular Networking Test-bed
• Extensive on Road Testing with over 400 miles
  logged
• GrooveNet Software Simulator
• Realistic Scalable with both real and virtual
  vehicles
• Network Wireless Channel Performance Analysis
• Encouraging Results with DSRC equipment!

24
BACKUP SLIDES
25
SocialNets ApplicationUser Profile Page
26
SocialNets ApplicationUser Filters