Title: Connectivity Aware Routing for Vehicular Ad Hoc Networks
1Connectivity Aware Routing for Vehicular Ad Hoc
Networks
- Qing Yang, Alvin Lim, Prathima Agrawal
Auburn University
IEEE Wireless Communications Networking
Conference March 31 April 3, 2008, Las Vegas
2Outline
- Goals
- Basic concepts
- Related works
- Assumptions
- Model of the connectivity
- Validation with VanetMobiSim
- Routing strategy
- Simulations and result analysis
- Conclusions
3Goals
- Achieves routing between two ends
4Basic concepts
- Model the probability of connectivity of each
road segment - Find the route with the highest probability of
connectivity - Carry-and-forward the packet while facing network
partitions
5Related Works
- VADD (06Infocom)
- Vehicle-Assisted Data Delivery
- MURU (06MobiQuitous)
- Multi-Hop Routing for Urban Vanet
- GSR (05MC2R)
- Geographic source routing
- GPSR (00MobiCom)
6Assumption
- GPS on each vehicle
- Standard component
- Digital maps 1,2
- Vehicle density
- Vehicle speed
- Traffic light period
1. http//www.mapmechanics/ 2. http//www.yahoo.co
m/
7Model of connectivity
- One lane road segment
- road segment (length is L) is equally divided
into m cells - each cell can contain only one node
- communication range is size of n0 cells
8Multiple lanes case
- One lane road segment
- Number of empty cells is m-n
- Multiple lanes road segment
- Number of lanes is n
- Number of empty cells ranges m-n, m-n/n
9Problem formulation
- Problem formulation
- What is the probability that network is connected
- The probability that no gap in networks is larger
than the communication range - Random Allocation (RA) theory
1
0
1
1
0
0
1
0
1
1
1
1
2
10Probability of connectivity
- P1 (exists exactly k empty cells)
- P2 (exists more than n0 continuous empty cells)
- Pcon (Probability of connectivity)
11VanetMobiSim
- Feature
- multi-lane roads
- separate directional flows
- traffic signs at intersections
- Intersection management
- Lane changing
- VanetMobiSim3 mobility patterns have been
validated against TSIS-CORSIM - a well known and validated traffic generator
3. http//vanet.eurecom.fr
12Validation of connectivity model
- Length of road
- 1000m, 1800m
- Traffic light period
- 60sec., 120sec.
- Average velocity
- 7.5m/s, 10m/s
13Result of validation
14Result of validation (cont)
15Probability of connectivity of route
p3
p8
p1
p11
p6
p4
p9
p1p4p7p12Max?
p2
p7
p12
p5
p10
- Pi probability of connectivity of road segment i
- Pcon probability of connectivity of selected
route (path) - Defined as ?Pi
16Computation of Pr
- Define of Pr
- sequence of road segments from source to
destination - Modified Dijkstra Algorithm
- Final goal maximize the probability of whole
path - Each step add one more edge, the probability of
new path decrease - Can be computed
- Centralized
- Distributed
17Computation of Pr (cont)
- Centralized
- Path Pr was computed only by source node
- Size of packet header is proportionate to hop
number - Distributed
- Every node who received packet computes Pr
- More processing on vehicles
18Routing strategy
- Compute Pr
- Find hops on road segment along Pr
- Every node beacons its current position
- Predict neighbors position while choosing next
hop - Carry and forward
- Buffered packet if no available next hop
- Send buffered packet when new next hop is in range
19Carry and forward strategy
- Buffer packets which cannot be forwarded
- Send out buffered packet if new next hop is
available - Good for VANET
- Lots of holes in VANET
- On path Pr, higher probability of obtaining a new
next hop - Big buffer is feasible in cars
Wisitpongphan, N. Bai, F. Mudalige, P. Tonguz,
O. K., "On the Routing Problem in Disconnected
Vehicular Ad-hoc Networks," INFOCOM 2007. 26th
IEEE International Conference on Computer
Communications. IEEE , vol., no., pp.2291-2295,
May 2007
20Simulation
- Nodes movement
- VanetMobiSim
- Map
- TIGER/LINE
- Real map from Tennessee
- Network simulation
- Ns2 setup
- Result analysis
21Map information
- TIGER4
- Topologically Integrated Geographic Encoding and
Referencing - A format used by the United States Census Bureau
to describe land attributes - Roads, buildings, rivers, and lakes
- Can be read by Tivec5
4. http//www.census.gov/geo/www/tiger/ 5.
http//www.triusinc.com/latest.htm
22Map in simulation
- Centered at
- Long -84877562
- Lat 35162102
- Size
- Width 2000m
- Length 2000m
23Ns2 setup
- Simulation Area 2000m2000m
- Number of nodes 100, 200
- Communication range 250m
- Packet size 512 Byte
- CBR rate 0.1 1packet/sec.
- Random selected source and fixed destination
- Buffer size 64kBytes
- Beacon interval 1.0 sec.
- Velocity 15 35 MPH
24Data delivery ratio (100nodes)
25Data delivery ratio (200nodes)
26Networking delay (100nodes)
27Networking delay (200nodes)
28Throughput
29Conclusions
- Connectivity issue is very important in VANET
- CAR performs well and is independent on the
network density - Perimeter mode of GPSR suffers in frequently
disconnected networks
30Questions and comments