EVALUATION OF MOBILE WIMAX PERFORMANCE IN A METROPOLITAN VEHICULAR APPLICATION by TONG JIN A Thesis submitted to the Graduate School-New Brunswick Rutgers, The State University of New Jersey in partial fulfillment of the requirements for the degree of Master of Science Graduate Program in Electrical and Computer Engineering written under the direction of Professor Marco Gruteser and approved by ________________________ ________________________ ________________________ ________________________ New Brunswick, New Jersey May, 2011
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
EVALUATION OF MOBILE WIMAX PERFORMANCE IN A METROPOLITAN
VEHICULAR APPLICATION
by
TONG JIN
A Thesis submitted to the
Graduate School-New Brunswick
Rutgers, The State University of New Jersey
in partial fulfillment of the requirements
for the degree of
Master of Science
Graduate Program in Electrical and Computer Engineering
written under the direction of
Professor Marco Gruteser
and approved by
________________________
________________________
________________________
________________________
New Brunswick, New Jersey
May, 2011
ii
ABSTRACT OF THE THESIS
Evaluation of Mobile WiMAX Performance in a Metropolitan Vehicular Application
By TONG JIN
Thesis Director:
Professor Marco Gruteser
Intelligent transportation systems (ITS) will enhance on-road safety, improve road
utilization, fuel efficiency and, last but not the least, allow occupants an entertainment
filled ride. They will include a plethora of applications requiring information
dissemination that will use vehicle-to-vehicle (V2V) communication, which has vehicles
form an ad hoc network to communicate with each other, and also vehicle-to-roadside
(V2R) communication, which amongst other things will provide connectivity to the
internet. The offered load in a vehicular network can be significant because of the
possibility of large vehicle densities and a variety of supported applications. Also, the
wireless propagation environment can be harsh, especially in a metropolitan area like New
York City and Los Angeles, and can significantly impact network performance. The load
requirements together with the high mobility in a vehicular network make mobile WiMAX
a suitable technology for vehicular networks. However, before WiMAX deployments are
used by vehicular networks, the performance of typical vehicular applications over
WiMAX needs to be evaluated. Also, propagation characteristics of WiMAX, especially in
relation to application performance, for example, the impact of intermittent connectivity on
data transfer over WiMAX, needs evaluation. In this work we first introduce the ParkNet
iii
system, in which vehicles collect and disseminate, over WiMAX, information regarding
roadside parking availability in a city. We implement the system and evaluate the
performance of a ParkNet deployment in Brooklyn, New York City, which includes
multiple cars and a WiMAX base station. The deployment monitors and measures the
WiMAX signal strength and data dissemination status and associated temporal and spatial
information. The data collected over multiple experiment runs conducted over many days
is used for application performance evaluation. The signal strength readings are further
used to derive a path loss model, which can be used by simulations of a mobile WiMAX
network.
iv
Acknowledgement and Dedication
I would like to acknowledge my thesis advisor Prof. Marco Gruteser providing me this
opportunity to work in such an interesting and exciting project. Thanks for his inspirational
instruction and guidance during my research in Wireless Information Network Laboratory
(Winlab) at Rutgers University. I would like to thank my co-workers, Ivan Seskar, Suhas
Mathur and Sanjit Kaul, both of whom indeed encouraged me and helped me a lot during
my research in Winlab. I feel so fruitful and enjoyable to work with them. I am also so
grateful to my colleagues Bin Zan, Sangho Oh, Nikhil.K, Janani.C, and Wenzhi Xue, for
their support and help during the whole project. Then, I would like to thank all the staffs in
Winlab and Electrical and Computer Engineering Department for their kind assistance. In
addition, I wish to thank all my friends and other people who look at me, care about me,
and emotionally support me. Finally, I would like to sincerely thank my parents and other
family members for their understanding, support, continuous inspiration and eternal love. I
cannot complete this job without their assistance, tolerance, and enthusiasm.
v
Table of Contents
ABSTRACT OF THESIS ................................................................................................... ii
Acknowledgement and Dedication .................................................................................... iv
Table of Contents ................................................................................................................ v
List of Tables .................................................................................................................... vii
List of Illustrations ........................................................................................................... viii
related downlink RSSI values, we can infer the WiMAX coverage, which reflects the
communication range of the base station, and also understand the distribution of the RSSI
values.
(a) (b)
Figure 5.1: (a) Experiment areas vs. Coverage of WiMAX signal. (b) The experiment data
distribution vs. received RSSI data distribution.
Figure 5.1(a) shows the experiment information of our running system on Google Maps.
As is shown, the experimental vehicles responsible for data covered the light orange area.
The dark red area near base station indicates the unique region over which WiMAX RSSI
could be sensed and measured. In the rest of the area that the vehicles covered, no link
related information was obtainable by them. Only when the RSSI value is greater than -
98dBm, the Intel WiMAX adapter embedded in client nodes could detect the signal,
connect to the WiMAX base station, and then transfer measured RSSI values. Figure
5.1(b) represents the real RSSI measurements accumulated during the experiment.
When compared with the entire 2000m * 1800m experiment area, WiMAX signal
coverage only occupies a small area of 500m * 1km around WiMAX base station that
31
has a directional antenna with 22dBi gain. Owing to the effect of tall building clusters in
urban experiment area, the Non Line-Of-Sight (NLOS) propagation results in the shorter
coverage range than the general coverage results on mobile WiMAX indicated in [19]
[22]. In the next subsection, we will discuss the statistic of the obtained RSSI
measurements and quantitively analyze the results.
5.1.2 RSSI Statistic Information and Connection Study
We can calculate the distance between the current measurement location and the base
station through the coordinate information provided by GPS data, and then map the RSSI
measurement with corresponding distance. This could be obtained by using the formula
below:
, (3.1)
where the Latitude’ and Longitude’ are the latitude and longitude that have been
translated to UTM standard with distance measured in meters, the subscripts represent the
base station and measurement location, and H0 is the height of the building height of the
building on which the base station is mounted, which is 16 meters in our case. According
to formula (3.1), we may plot the distribution of RSSI measurements against
corresponding distance values, as shown in Figure 5.2.
32
(a) (b)
Figure 5.2: (a) Downlink RSSI data over distance to Base Station.
(b) The RSSI data distribution over corresponding coordinates.
Figure 5.2 (a) shows the downlink RSSI data over distance to WiMAX base station, the
range of which varies from -40 dBm to -98 dBm. Figure 5.2 (b) depicts the area where
the RSSI data is distributed and could be received by client nodes in the car. When the
signal strength is weaker than the threshold -98 dBm, its RSSI value cannot be detected
and measured. At this detection sensitivity level, the WiMAX coverage extended up to
590 meters, which is smaller than kilometer or so of propagation distance revealed in
previous research results [2] and [22]. In [22], researchers from Norway ran the
vehicular experiment over mobile WiMAX on 3.5GHz, with 28dBm transmitter power
and 14dBi antenna gain. The result showed that the coverage could reach 1000m in urban
area and 2km in suburban area. Compared with these research outcomes, reduction in
coverage distance mainly results from the difference in the transmission environment,
which in our case is a metropolitan area. Terrain characteristic in New York City, such as
dense constructions, negatively impacts WiMAX propagation and has a significant effect
on implementation of applications using mobile WiMAX. Also, some outliers in the red
33
rectangle don’t match the trend of measurements distribution. By projecting these
measurements on the map, we may find that they are collected under the Line of Sight
condition, where has the higher signal strength than those from other space with same
distance.
Minimum
Connection Time
Maximum
Connection Time
Mean Connection
Time
Profile #1 79.2 s 1145.1 s 402.5 s
Profile #2 128.6 s 276.7 s 2027 s
Profile #3 36.9 s 1527.9 s 389.5 s
Profile #4 151.6 s 1193.3 s 522.0 s
Profile #5 156.8 s 410.3 s 2283.5 s
Profile #6 310.7 s 475.2 s 393.0 s
Profile #7 3.83.9 s 3.83.9 s 3.83.9 s
Profile #8 529.5 s 529.5 s 529.5 s
Profile #9 207.7 s 292.8 s 250.2 s
Profile #10 1183.2 s 1183.2 s 1183.2 s
Total 36.9 s 1527.9 s 402.6 s
Table 5.1: Experimental Vehicles’ Profiles on Connection Duration
34
Table 5.1 shows the duration of each connection, the minimum of which is only 36.9
seconds. These duration values could help us plan the WiMAX base station deployment
in future implementation.
5.1.3 Metropolitan Terrain Characteristics and Effects
Vehicle Mobility:
Because of the inherent wide coverage and mobility support of mobile WiMAX, it
minimizes the rate of handover and data loss. The effect is more severe in suburban area
rather than urban area, owing to the higher speed limit and lower traffic density. In
Brooklyn area at NYC, frequent traffic light, speed limits, and congested traffic flow
reduces vehicular mobility and hence the Doppler spread.
Minimum Speed Maximum Speed Mean Speed
Profile #1 0 m/s 10.29 m/s 2.03 m/s
Profile #2 0 m/s 9.09 m/s 1.54 m/s
Profile #3 0 m/s 12.21 m/s 1.39 m/s
Profile #4 0 m/s 11.89 m/s 1.35 m/s
Profile #5 0 m/s 13.78 m/s 1.87 m/s
Profile #6 0 m/s 13.06 m/s 1.28 m/s
Profile #7 0 m/s 13.27 m/s 1.67 m/s
Profile #8 0 m/s 10.72 m/s 1.84 m/s
Profile #9 0 m/s 13.24 m/s 2.66 m/s
Profile #10 0 m/s 9.17 m/s 1.07 m/s
Total 0 m/s 13.78 m/s 1.67 m/s
Table 5.2: Experimental Vehicles’ Profiles on Velocity Information
35
Table 5.2 shows the car profiles of testing vehicles. As it can be seen, during the entire
experiment, none of the vehicles reach 14 meter/second and the mean speed is limited to
1.67 meter/second, which is less than that in a usual urban area [28].
Terrain Effects:
Figure 5.3 (a) below depicts the heat map of RSSI values, which represents all the RSSI
values obtained through interpolation over the experiment area by using different colors.
The received signal strength decreases gradually from warm colors to the colder ones.
(a) (b)
Figure 5.3: (a) The RSSI heat-map. (b) Building distribution inside RSSI coverage area
The RSSI values at same levels are contoured with dash lines, showing the general RSSI
distribution over the selected environment. As we can see, the RSSI attenuation gradually
becomes larger with the increasing distance from base station; but the variance is
relatively slower in vertical and horizontal direction. In contract, on the left flank area,
dense contour lines represent the faster fading that shortens the coverage radius. This
phenomenon could be explained through the comparison with Figure 5.3 (b), which
36
displays the distribution of constructions projected on Google Map, over the same area,
in Figure 5.3 (a). The dark black regions represent the buildings and the light blue lines
represent the testing routes where RSSI values are measured. The measurements with
different colors depict various RSSI levels. In our experiment configuration, the base
station is placed on the top of a 16 meters building, which is surrounded by nearby tall
buildings. Hence, when the vehicular client nodes run following the blue routes, the paths
of wave propagation are blocked by the construction clusters at most of time, resulting
into the NLOS propagation environment that greatly lowers the effective received power
and reduces the transmission distance. As a result of being impacted by continuous high
obstacles, signal strength sharply attenuates, especially on the lower left area. However,
since the area on the top left is next to the on which the base station is installed, the
measurements collected in horizontal direction experience Line-Of-Sight (LOS)
propagation conditions and perform better on transmission range and signal fading rate.
In addition, the directional antenna radiates greater power in vertical downward direction,
allowing for increased performance on transmit and receive on that area, as shown by the
heat map.
Therefore, from the analysis above, we can conclude that the metropolitan terrain
environment and geographical position characteristics introduce a mixing propagation
condition including both NLOS and LOS, the former being predominant.
5.2 Path Loss Model for Mobile WiMAX in Metropolitan Environments
In wireless communication systems, information is transmitted between the transmitter
and the receiver by electromagnetic waves. As electromagnetic waves propagate through
37
space, its continuous interaction with environment causes the path loss, which is the
reduction in its power density.
We want to derive a path loss model for a metropolitan environment like that in New
York City. The model can be used as a reference for future WiMAX experiments and also
to evaluate any vehicular applications that use WiMAX.
During most of the experiment time the vehicles observe NLOS conditions. The vehicles
observe LOS communication only along a few streets in the vicinity of the WiMAX base
station. Hence, the path loss model under NLOS condition has a crucial value on urban
network planning. We delete the RSSI collections from LOS streets and only consider the
measurements under NLOS conditions to obtain the path loss model. The pruned
measurements are shown in Figure 5.4.
Figure 5.4: NLOS downlink RSSI data over distance to Base Station.
We aim at getting a path loss model of the form:
38
PL(dB) = A+Blog10(d) (1)
where A is PLd0, which is the reference path loss value at d0=1 meter distance; d is the
distance from the transmitter to receiver; and B is 10n; n is the path loss exponent, which
indicates the rate of propagation path loss with respect to distance. When the environment
propagation characteristic is close to free space propagation or has fewer clusters, the
path loss exponent value is about 2. In urban environment, the path loss exponent is
between 2 and 4. Therefore, the path loss at a given location with respect to the reference
distance d0, the equation (1) could be expressed as:
PL(dB) = PLd0 + 10nlog10(d/d0) (2)
When a graph of path loss against logarithm of distance is plotted, the path loss exponent
n could be determined by calculating the slope of the line that corresponds to the linear
least squares fit that minimizes error in prediction by the fit. The path loss value at
reference distance d0 can be obtained as well.
To derive the path loss value at a given location with respect to the path loss at a
reference distance, we need make use of the relationship between path loss values and
existing measured plentiful of RSSI data with the same location. A path loss model can
be derived from link budget equation of communication system:
RSSI = Tx + Gt – PL + Gr (3)
where the left side of equation is the received signal strength (dBm); on the right side of
equation, Tx is the transmitter power (dBm); Gt is the transmitter antenna gain (dBi), Gr
is the receiver antenna gain (dBi), and PL is the path loss (dB) in communication
environment. Thus, from equation (3), the path loss mode equation could be written as:
39
PL = Tx + Gt + Gr - RSSI (4)
Figure 5.5: RSSI measurements against the logarithm of the distance
The RSSI related to the distance between WiMAX Base Station and the mobile client
nodes provides valuable information related to the power loss in this WiMAX
communication system. Figure 5.5 shows the RSSI measurements versus the logarithm of
the distance between Base Station and clients, excluding abnormal measurements. A
straight line should be drawn through the points in this figure so as to confirm the path
loss equation (4). We use Least Square (LS) regression analysis to determine the slope
and other parameters of the path loss line, which is given by:
RSSI = -7.84-30.08*log10(d) (5)
Where d is the distance between Base Station and client nodes, and the RSSI is denoted
in dBm.
40
Figure 5.6: Equation (5) with RSSI measurements against distance.
Also, we can calculate the prediction error between collected RSSI and the result on
equation (5) by using:
Error = RSSI measured – RSSI (6)
After using the statistical analysis approach of distribution fitting, we can get a normal
distribution curve fitting this error data. The normal distribution N (0, 57.1) is a good fit.
Its mean value is 0, and its variance is 57.1. In the figures 5.7 and 5.8 we show the
probability distribution function and the cumulative distribution function with
corresponding error data statistic quintiles.
41
Figure 5.7: Probability density of model error vs. Probability density function curve of fitting
Normal distribution
Figure 5.8: Cumulative probability of model error vs. Cumulative distribution function curve of
fitting Normal distribution
42
In our system, the transmission power is 32 dBm and the transmitter antenna gain is 22
dBi. According to path loss model equation (2), we may put the reference path loss value
at 1 meter point to form the full path loss model:
PL = 61.84 + 30.08*log10(d) (7)
Where d is the distance between base station and client node, and the path loss is denoted
dB. Also, the path loss exponent is obtained to be 3.008.
Figure 5.9: NLOS Path Loss Model (upper one),
vs. Path Loss Model on Free Space (lower one).
Figure 5.9 indicates the comparison between NLOS path loss model that we get from
experiment results and the one assuming free space. The path loss exponent in urban area
reflects the negative effects of high density tall constructions clusters between WiMAX
base station and onboard client receivers.
43
As the above analysis procedure shows, the derived empirical NLOS Path Loss Model of
metropolitan mobile WiMAX is presented on equation (7), whose error follows the
normal distribution N (0, 57.1).
5.3 Data Transfer Performance
In this mobile WiMAX implementation, when employing the micro-element file transfer
approach, the client node makes all the effort to transfer the small fixed-size file during
the network connectivity. After the connection times out or the TCP link breaks, it aborts
the current uploading assignment and keeps attempting to re-connect to the WiMAX
network until next successful connectivity is built up; and then it continues to upload the
remaining files. During our entire experiment, we successfully transferred 608 files each
of size 20KB.
Figure 5.10: Transferred File Numbers/Number of RSSI vs. RSSI Values
44
Figure 5.10 above illustrates the number of transferred files grouped based on received
RSSI values, comparing with corresponding number of RSSI measurements. As we can
see, the red columns represent the number of RSSI measurements falling into associated
RSSI value range; and the blue ones show the number of successfully transferred files
under a certain RSSI value. According to the figure, we find that most of successful file
transfers happened when the RSSI was above -75dbm, and a few files are transferred
between -79dbm and -75dbm. It is inferred that SCP operation would be able to build the
TCP connection when the signal strength is larger than a certain threshold value like -79
dBm.
Figure 5.11: File Transfer Area vs. RSSI Coverage Area
45
Figure 5.11 compares the RSSI coverage area (red) with file transferring area (black).
The maximum distance that file transfer succeeded reaches 396 m.
Time #1 Time #2 Time #3 Time #4 Time #5
time file # time file # time file # time file # time file #
Profile #1 133 s 6 7 s 1 109 s 26 11 s 5 296 s 61
Profile #2 163 s 7 257 s 111
Profile #3 128 s 5 24 s 5 37 s 7 1391s 115
Profile #4 43 s 2 127 s 39 1027s 66
Profile #5 161 s 7 116 s 38 150 s 55
Profile #6 80 s 4 297 s 50
Profile #7 309 s 148
Profile #8 244 s 94
Profile #9 214 s 118 113 s 9
Profile #10 746 s 52 145 s 22
Table 5.3: File Transfer Duration and Number
Although micro-element file transfer mechanism effectively avoids possible frequent
aborting of file uploading during large file transfers when using SCP, due to break in
network connectivity, it introduces a large amount of overhead caused by frequent
connection attempts and TCP handshake. Every time when an experiment vehicle passes
through WiMAX coverage area, it makes best effort to connect to the network and
transfer a certain number of files during that time, the results of which are depicted in
Table 5.3. As the table shows, the efficiency of file transfer is very low, which has the
46
average throughput of only 70kbps and maximum value of 91kbps. In other words, only
0.43 file is transferred per second.
Figure 5.12: Comparison Experiment over TCP
After doing a comparison experiments, we tested the data rate performance in different
transfer mechanisms using SCP operation. As is shown in Figure 5.12, the blue curve
represents the large file transfer over one SCP operation. Then, we split the file into small
pieces with 20KB size each to get the red curve. The black curve used the same
mechanism in Brooklyn test. We find that the data rate of black curve keeps at 70Kbps all
the time. This result is produced due to the dynamic behavior of TCP. It takes some
amount of time to increase sending rate to the maximum bandwidth. Hence, when the file
size is small, the file transfer is already completed before TCP reaching the maximum
link throughput. Although blue curve shows the better maximum link throughput, it costs
47
more time on file transfer and brings into higher risk of transfer abortion. Therefore, we
would like to achieve a tradeoff between the high throughput and risk of transfer abortion
caused by increasing transfer time. To obtain such tradeoff, a fast reaching throughput
with small pieces is preferable in the future design. For example, UDP is a good choice
instead of TCP, when the reliable transfer mechanism and congestion control could be
achieved on the upper layer.
48
Chapter 6
Conclusion and Future Work
In this work, we describe the implementation of ParkNet system, a WiMAX-based
vehicular application in Brooklyn area, New York City. Also, we analyze and evaluate
the physical performance of mobile WiMAX in the metropolitan environment using a
large amount of measurements collected over the course of extensive experimentation
using a real world WiMAX deployment and multiple cars.
Through the analysis of the terrain characteristics in Brooklyn, we find that the sensing
vehicles in ParkNet system predominantly observe Non Line of Sight (NLOS)
propagation conditions, due to the dense tall building constructions between the base
station and client receivers. In addition, the low speed limit and congested traffic
conditions in the experimental area reduce vehicular mobility; the speed never exceeds 14
meter/second; and the mean speed is constrained to 1.67 meter/second.
From the downlink RSSI measurements we collected, the range of which varies from -40
dBm to -98 dBm, we observed that WiMAX coverage extends up to 590 meters under
this terrain environment, which is smaller than kilometer-scale propagation distance
observed in other propagation environments. We also derived the analytical path loss
model describing the WiMAX signal propagation in the metropolitan environment. As
the analysis in Chapter 5 shows, the derived empirical NLOS Path Loss Model of
metropolitan mobile WiMAX is described by a complementary error CDF that has a
normal distribution ~ N(0,57.1). The path loss exponent is obtained to be 3.008.
49
During our entire experiment, we successfully transferred 608 files of which each was
20KB large. Most of these transfers happened when the RSSI was greater than -75dBm,
and a few were done when the RSSI was in between -79dBm and -75dBm. It is inferred
that secure copy process (SCP operation) would be able to build the TCP connection
when the signal strength is larger than a certain threshold value of about -80 dBm. File
transfer is hard to complete under an RSSI of -79dBm.
We analyze the effect of micro-element file transfer mechanism, and show that a file is
uploaded before TCP is able to reach maximum link throughput. Therefore, in spite of the
avoidance of temporal cost caused by data transfer abortion, the uplink throughput of
mobile WiMAX in this application is lower than most of previous work. Thus, in the
future, we plan to modify the file transfer mechanism to balance the data size and risk of
file transfer abortion. Also, we may use UDP to replace TCP in ParkNet data
dissemination process, and design reliable file transfer mechanism on upper layer. Last
but not the least, in addition to throughput, the latency performance of data transfer can
be measured and analyzed.
50
Reference
[1] Hartenstein, H. and Laberteaux, K. P. (2009) Front Matter, VANET: Vehicular
Applications and Inter-Networking Technologies, John Wiley & Sons, Ltd, Chichester, UK [2] Standard IEEE 802.16e-2005 available on site http://www.IEEE802.org/16 published on 28
February [3] Suhas Mathur, Tong Jin, Nikhil Kasturirangan, Janani Chandrasekaran, Wenzhi Xue, Marco
Gruteser, and Wade Trappe. Parknet: Drive-by sensing of road-side parking statistics. The 8th international conference on Mobile systems, applications, and services, MobiSys ’10, pages 123–136, New York, NY, USA, 2010. ACM.
[4] Jungkeun Yoon, Brian Noble, and Mingyan Liu. Surface street traffic estimation. In MobiSys 07, pages 220–232. ACM, 2007.
[5] Jörg Ott and Dirk Kutscher. Drive-thru internet: IEEE 802.11b for automobile users, Twenty-third AnnualJoint Conference of the IEEE Computer and Communications Societies, INFOCOM, 2004.
[6] Richard Gass and James Scott. Measurements of in-motion 802.11 networking. In Processing of IEEE WMCSA, pages 69–74, 2006.
[7] Vladimir Bychkovsky, Bret Hull, Allen Miu, Hari Balakrishnan, and Samuel Madden. A measurement study of vehicular internet access using. In Situ Wi-Fi Networks. In 12th ACM MOBICOM Conf, pages 50–61, 2006.
[8] Ratul Mahajan, John Zahorjan, and Brian Zill. Understanding wifi-based connectivity from moving vehicles. In IMC 07: Proceedings of the 7th ACM SIGCOMM conference on Internet measurement, 2007.
[9] S. Vasudevan, K. Papagiannaki, C. Diot, J. Kurose, and D. Towsley. Facilitating access point selection in IEEE 802.11 wireless networks. In IMC, 2005.
[10] Lei Xie, Qun Li, Weizhen Mao, Jie Wu, and Daoxu Chen. Optimal access point selection and traffic allocation in IEEE 802.11 networks.
[11] Lei Xie, Qun Li, Weizhen Mao, Jie Wu and Daoxu Chen. Achieving Efficiency and Fairness for Association Control in Vehicular Networks, 17th IEEE International Conference on Network Protocols, ICNP 2009.
[12] I. Amdouni and F. Filali. Intelligent strategies of access point selection for vehicle to infrastructure opportunistic communications, Vehicular Networking Conference (VNC), 2009 IEEE.
[13] Jakob Eriksson, Hari Balakrishnan, and Samuel Madden. Cabernet: Vehicular content delivery using wifi, Mobicom '08, San Francisco, California, USA, 2008. ACM.
[14] Hao Wu, Richard Fujimoto, Randall Guensler, and Michael Hunter. MDDV: a mobility-centric data dissemination algorithm for vehicular networks. In VANET ’04: Proceedings of the 1st ACM international workshop on Vehicular ad hoc networks, pages 47–56, New York, NY, USA, 2004. ACM.
[15] Robert Akl and Sangtae Park. Performance study of robust data transfer protocol for VANETS, WMSCI: Communication and Network System, Technology and Applications, 2005.
[16] Marios D. Dikaiakos, Tamer Nadeem, Saif Iqbal, and Liviu Iftode. Vitp: an information transfer protocol for vehicular computing. In Workshop on Vehicular Ad Hoc Networks, pages 30–39, 2005.
[17] Uichin Lee, Joon sang Park, Joseph Yeh, Giovanni Pau, and Mario Gerla. CodeTorrent: Content distribution using network coding in vanet, Mobishare '06, Los Angeles, California, USA, 2006.
[18] M. Tran, G. Zaggoulos, A. Nix, and Doufexi. Mobile wimax: Performance analysis and comparison with experimental results, IEEE 68th Vehicular Technology Conference, Fall 2008.
[19] Pål Grønsund, Paal Einar Engelstad, Torbjørn Johnsen, and Tor Skeie. The physical performance and path loss in a fixed wimax deployment. In Mohsen Guizani and Hsiao-Hwa Chen, International Wireless Communications and Mobile Computing Conference, pages 439 – 444. ACM Press, 2007.
[20] Jim Barthold. Wisconsin first responders use 3g for emergencies, ambulance service connects in-vehicle network and hospitals via mobile, Telecommunications Magazine, 2007.
[21] Pralhad Deshp, Xiaoxiao Hou, and Samir R. Das. Performance comparison of 3g and metro-scale wifi for vehicular network access, presented at IMC '10, 2010.
[22] Doug Gray. Mobile wimax: A performance and comparative summary, WiMAX Forum, 2006.
[23] R. Bonneau, M. Salazar-Palma, M. Wicks, and T. Sarkar. A survey of various propagation models for mobile communication, Smart Antennas, IEEE, 2005.
[24] Mardeni.R and T. Siva Priya. Optimised cost-231 hata models for wimax path loss prediction in suburban and open urban environments, 2010.
[25] J. Milanovic, Rimac-Drlje, S., K. Bejuk, and Osijek. Comparison of propagation models accuracy for wimax on 3.5 ghz, Electronics, Circuits and Systems, 2007. ICECS 2007. 14th IEEE International Conference on, 2008.
[26] Antnio J. Rodrigues, João Pedro Eira. Analysis of wimax data rate performance. [27] Damiano De Luca(b), Fabio Fiano(d), Franco Mazzenga(a), Cristiano Monti(b), Stefano
Ridolfi(e), and Francesco Vallone(c). Outdoor path loss models for IEEE 802.16 in suburban and campus-like environments, IEEE Communications Society subject matter experts for publication in the ICC 2007 proceedings, 2007.
[28] D. Aphiraksatyakul, Boon-Chong Seet, and Chiew-Tong Lau. Evaluation of terrain effects on mobile wimax in a vehicular environment, ITS Communication, 8th International Conference on, 2008.
[29] R. Wahl, O. Stabler, and G. Wolfle. Propagation model and network simulator for stationary and nomadic wimax networks, Vehicular Technology Conference, 2007. VTC-2007 Fall. 2007 IEEE 66th, 2007.
[30] D. Shoup. “Cruising for parking”, Access, vol. 30, pp.16-22, 2007. [31] “The Price of Parking on Great Streets”. In Planetizen Contemporary Debates in Urban
Planning, Washington, DC, Island Press, pp. 52-56, 2007. [32] Connie Ribeiro. Bringing Wireless Access to the Automobile: A Comparison of Wi-Fi,
WiMAX, MBWA, and 3G, presented at 21st Computer Science Seminar, Rensselaer at Hartford, 2005.
[33] D. Jiang, V. Taliwal, A. Meier, W. Holfelder, and R. Herrtwich. "Design of 5.9 ghz dsrc-based vehicular safety communication," Wireless Communications, IEEE , vol.13, no.5, pp.36-43, October 2006
[34] S. Eichler, "Performance Evaluation of the IEEE 802.11p WAVE Communication Standard" IEEE 66th Vehicular Technology Conference, 2007. VTC-2007 Fall 2007.
[35] Dongmyoung K, Hua Cai, and Sunghyun Choi. "Performance Measurement over Mobile WiMAX/IEEE 802.16e Network" In Processing of the 2008 WOWMOM ‘08, IEEE, 2008.