Transcript Investigation of Certification ofUsabilty/UCD Professionals

MAC Performance Analysis for
Vehicle to Infrastructure
Communication
Tom H. Luan*, Xinhua Ling§ , Xuemin (Sherman) Shen*
*BroadBand Communication Research Group
University of Waterloo
§ Research In Motion
Outline
1. Introduction to Vehicular Network
2. Model of MAC in V2I communication
3. Simulation
4. Conclusion
22
Why Vehicular Networks ?
 Internet becomes an essential part of our daily
life
 Watch video on Youtube; order literature on Amzone;
catch the final moments of an eBay auction …
 Americans spend up to 540 hours on average a
year in their vehicles (10% of the waking time)
 Internet access from vehicles is still luxury
 Vehicular Network
 To provide cheap yet high throughput data service for
vehicles on the road
33
V2V and V2I Communications
Vehicle to RSU
(V2R or V2I)
Vehicle to
Vehicle (V2V)
RSU (roadside unit)
 Infotainment: Internet access, video streaming, music
download, etc.
 MAC throughput performance evaluation of V2I
communication
44
Standard and Research Efforts
 IEEE drafts 802.11p standard to permit
vehicular communication
 802.11a radio technology + 802.11e EDCA MAC
 Multi-channel: 6 service channels + 1 control channel
 Drive-thru Internet
 Using off-the-shelf 802.11b
hardware, a vehicle could maintain
a connection to a roadside AP for
500m and transfer 9MB of data
at 80km/h using either TCP or
UDP
Image from http://www.drive-thru-internet.org/
[1] J. Ott and D. Kutscher, "Drive-thru Internet: IEEE 802.11 b for 'automobile' users," in IEEE INFOCOM, 2004
55
Standard and Research Efforts (cont’d)
 CarTel in MIT [2]
 City-wide experiment showing the intermittent and
short-lived connectivity, yet high throughput while
available
 Small scale network
without considering MAC
 Link layer and transport
layer performance
 What if a great number of
vehicles moving fast?
[2] V. Bychkovsky, B. Hull, A. Miu, H. Balakrishnan and S. Madden, "A measurement study of vehicular
internet access using in situ Wi-Fi networks," in ACM MobiCom, 2006
66
Problem Statement
 MAC performance evaluation for fast-moving
large scale vehicular networks
 We consider 802.11b DCF
 Used by most trail networks, e.g., Drive-thru
 Compatible to WiFi device (e.g., iPod Touch)
 The basis of 802.11p MAC
77
Network Model
 Perfect channel without packet loss and errors
 Saturated case: nodes always have a packet to transmit
 Multi-rate transmission according to the distance to RSU
 Spatial zones: the radio coverage of one RSU is divide into
Z = {0, 1, …, N} zones according to node transmission rate
 p-persistent MAC: nodes transmit with a constant probability
pz for different zone n in Z
RSU
 Mobility Model
n
Mirror zones
along RSU
nmap
Received SNR (dB)
 Sojourn time of vehicles in each
zone n is geometrically
distributed with mean tn
 Within a period  , vehicle moves
from zone n to n+1 with the
probability  /tn, and no change
with the left probability
RSU
Markov
chain
1
2
1
2


N-1
N
N-1
N
Zone
88
Markov Model of Vehicle Nodes
Back off Interval
Countdown
 Each node can be represented
by {z(t), b(t)}
 z(t): zone the vehicle is current
in at time t
 b(t): the value of backoff
counter of the node at time t
Geometric
distribution (p1)
1,0
1,1
1,2

1,W-1

Movement
of Vehicles
2,0
2,1
2,2





2,W-1
N-1,0
Geometric
distribution (pN)
N,0
N,1
N,2

N,W-1
 2D Markov chain embedded at the commencement of the
backoff counter countdown
 Upon the decrement of backoff counter, vehicle may either
move to the next zone or stay in the original zone
 When coming into a new zone, different transmission
probability is applied
99
Simulation Setup
Zone 2
ps
ps
Zone 1
1 1 Mb
5.5
Mb
Zone 0
...
2M
b
ps
ps
 By default, 50 vehicles move at
constant speed with v = 80 km/h
RSU
Mb
11
 Radio coverage of RSU is 250m,
which is divided into 8 zones
...
Zone N
Zone 0
 When arriving at the end of the road session (zone N),
vehicles reenter zone 0 and start a new iteration of
communication
 Two schemes
 Equal contention window (transmission probability p) in all zones
 Differential contention window in zones
10
10
Nodal Throughput in Each Zone n
Nodal Throughput in Each Zone
sn =
Average pkt length in each trans.
Mean interval between consecutive trans.
Integrated Throughput
S = ∑ Xn sn
n
Where Xn is the node population in zone n
 Using equal CW in all zones would suffer from
performance anomaly
11
11
Increasing Velocity
 With enhanced node velocity, nodes in front zones have
higher throughput than the back zones
 The small CW in zone 4 benefits the following zones
 System throughput reduces when velocity increases
12
12
Conclusion
 Throughput performance evaluation of DCF in the
vehicle to infrastructure communication
 Increase the velocity would reduce the system
throughput
 Future work
 Optimal design of DCF (contention window)
 QoS provision with call admission control etc.
13
13
Question and Answers ?
Thank you !
bbcr.uwaterloo.ca/~hluan
14