Content uploaded by Alessandro Bazzi
Author content
All content in this area was uploaded by Alessandro Bazzi on Dec 14, 2016
Content may be subject to copyright.
Cellular Aided Vehicular Named Data Networking
Alessandro Bazzi, Barbara M. Masini,
Alberto Zanella, Cristina De Castro
CNR-IEIIT, Bologna, Italy
Email: {alessandro.bazzi, barbara.masini,
alberto.zanella, cristina.decastro}@ieiit.cnr.it
Carla Raffaelli, Oreste Andrisano
DEI, University of Bologna, Italy
Email: {carla.raffaelli, oreste.andrisano}@unibo.it
Abstract—1With the recent new paradigm for the future
Internet named data networking (NDN), contents are searched
by names and not by address, and any node storing a content
can also act as a source of information. Such a paradigm
appears to be of particular interest for vehicular ad hoc networks
(VANETs), to extend internet services to vehicles and support
new mobile applications. In VANETs, devices not only use, but
also generate contents; devices are generally not constrained
in power or memory, thus easily carrying large amount of
information; in addition, high mobility continuously creates new
opportunities for data exchange. In such scenario, the TCP/IP
networking paradigm shows some performance limitations and
NDN appears as a way to improve reliability and efficiency of
content distribution. However, when node density is limited, the
exclusive use of short range communications might reduce the
effectiveness of NDN. Motivated by this consideration, we propose
the use of cellular networks to carry the signaling part of NDN,
and the exploitation of short range wireless communications
for content distribution. We will show, through simulations
performed in an urban scenario with different vehicle densities,
that moving the signaling part to cellular networks significantly
improves the performance of NDN compared to the case when
only short range communications are used.
I. INTRODUCTION
Connected vehicles already represent a reality. Their number
is expected to grow threefold within five years and every new
car is expected to be connected in multiple ways by 2025
[1]. The most widespread solution is represented by cellular
systems, that will be embedded in about the 20% of cars
by 2018 [1]. Another long run solution is represented by
short range wireless communications enabled by WAVE/IEEE
802.11p [2], [3], with vehicles creating vehicular ad hoc
networks (VANETs). Cellular systems allow each vehicle to
communicate with the infrastructure (e.g., a remote control
center), but the cost of its use is not negligible [4]. On the other
hand, short range communications enable direct transmissions
among equipped vehicles, thus saving cellular resources, but
suffer from connectivity problems in low vehicular density
conditions [5], [6].
A wide set of applications will be enabled by connected
vehicles, ranging from infotainment services, safety, and effi-
1c
2014. This is the accepted version of the manuscript. The manuscript
will undergo copyediting, typesetting, and review of the resulting proof before
it is published in its final form; during the production process errors may be
discovered which could affect the content, and all disclaimers that apply to
the journal apply to this manuscript. A definitive version will be subsequently
published, DOI: 10.1109/ICCVE.2014.7297650.
ciency of transportation, to industrial and forensic applications.
VANETs will be characterized by rapidly changing topologies
and vehicles will be both users and sources of contents.
For these characteristics, the conventional network paradigm,
based on the TCP/IP model, does not appear to be the most
suitable for this new scenario.
The management of information requires a novel vision and
named data networking (NDN) (also known as content centric
networking (CCN)) represents the most promising answer [7].
In NDN, in fact, the application searches contents by name
and not by address, independently from which node generated
the data [8]. Clearly, this paradigm can be applied to any
content. Just as an example, if one searches for the updated
news from his favorite newspaper, he is not interested to reach
a server somewhere, but to obtain the content itself. In NDN,
communication is based on the broadcasting of Interest packets
including the name of the searched Data. Any node, owning
or caching the content, can reply with the Data.
In VANETs, nodes have no particular constrains on power
and memory, thus they can cache large amounts of information
and act as mules for a wide diffusion; the NDN paradigm
takes thus advantage of the high mobility and it overcomes
the problem of session setup and maintenance in a highly
dynamic and partially connected environment. Recent papers
started discussing and demonstrating the effectiveness of NDN
in VANETs, also known as vehicular NDN (VNDN) [8]; in
[8], a prototype implementation is described with preliminary
performance assessment; in [9], the focus is on the design of
naming for VNDN; in [10], the authors discuss the structure
and content of messages for content request and delivery;
authors in [11] focus on crowd-sensing in VNDN and pro-
pose a novel selective network coding method to increase
the reliability of request distribution; in [12], a solution to
allow interoperability between NDN and IPv6 in VANETs
is proposed; in [13], the authors focus on traffic information
diffusion in VNDN and design a set of timers to improve
content delivery rate.
It is worth noting that current proposals are focused only
on the use of short range wireless technologies for VNDN. As
shown in this paper, this approach might lead to a limited diffu-
sion of the Interest messages and cause the not reachability of
the available Data sources; the consequence is that available
contents are not spread to neighbors and following requests
cannot take advantage of new sources. This is particularly true
when the content is not very popular, that is, when requests
are not frequent. This problem will reduce when the density of
connected vehicles will increase, although this will be possible
as a long term solution and will open other challenges, such as
a heavier impact of collisions on delivery delay and delivery
ratio [14].
Thinking to short or medium term VNDN applications,
a novel approach is proposed based on the use of cellular
networks to better convey Interest messages in VNDN. The
load over cellular networks is limited, since Interest messages
are generally small and Data messages are sent through
WAVE/IEEE 802.11p.
Results, obtained in a realistic urban scenario with different
vehicle densities, show that the use of cellular networks
increases significantly the probability that the Data is delivered
to the requesting node through the VANET.
The paper is organized as follows: in Section II, the main
concepts of VNDN are described; in Section III, our proposal
is detailed; in Section IV the simulation settings are discussed;
in Section V results are given; in Section VI conclusions are
drawn.
II. VEHICULAR NA ME D DATA NETWORKING
In NDN, contents are searched by name instead of ad-
dress, and nodes are divided into producers,consumers, and
routers [7]. Any node requiring a content is a consumer that
sends an Interest for that content; any node storing that content
will act as producer and satisfy with the Data the received
request; all nodes involved in routing the Interest and Data act
as routers. Nodes in NDN store three structures to manage
pending requests and stored contents: a Content Store (CS)
to store Data packets, a Pending Interest Table (PIT) to store
unanswered requests with the requesting node or nodes, and a
Forward Information Base (FIB) to indicate where to convey
incoming requests for contents not yet stored in the CS. If
a node receives an Interest, it firstly checks the CS. If the
corresponding Data is present, it returns the Data, otherwise
it checks the PIT to verify if this request is new or can be
discarded. If the request is new, the FIB is used to route
the Interest. When a node receives some Data that was not
addressed to itself, then the PIT is checked; if an entry is
present for that Data, before removing the entry the node saves
the Data in the CS and forwards it to the node or nodes that
previously sent the corresponding Interest.
The described NDN architecture needs some modifications
to be efficient in a VANET environment, which is rather
peculiar. First of all, network topology modifies frequently
and quickly due to mobility and the FIB would be soon out to
date; for this reason, the FIB is generally not used in VNDN.
Secondly, memory and power consumption are not a primary
issue in VANETs, and nodes store all contents for as long as
possible, independently on whether they have pending requests
or not. Furthermore, requests have a nonzero probability to
be not satisfied due to limited connectivity, and a timer to
retransmit the Interest message must be foreseen [9].
III. PROP OS ED SOLUTION
In this paper we will refer to:
•vehicle-to-vehicle (V2V) for short range communications
among vehicles,
•vehicle-to-roadside (V2R) for short range communica-
tions between vehicles and road side units (RSUs), and
•vehicle-to-infrastructure (V2I) for communications
through cellular networks.
The use of V2V and V2R to enable VNDN is an attractive
opportunity to offload cellular networks. However, due to
the limited connectivity and high mobility, the exclusive use
of short range technologies might cause difficulties by the
consumers to reach the producers with the Interest messages;
if an Interest is not delivered, the corresponding Data cannot
be sent back and spread to other CSs; this strongly limits the
effect of NDN. Previous works, indeed, assume the presence
of few contents that are required by many nodes: in this
case, any successful exchange spreads the Data to the CS
of the involved routers, making the content available for the
following requests.
A different option is to use V2I for the Interest exchange,
and exploit V2V and V2R for the Data exchange. More
specifically, the Interest is sent through cellular networks to a
VNDN management center (VNMC), which is a functional
entity that has the only duty to forward the message in
downlink. In the downlink, multimedia broadcast multicast
services (MBMS) (in 2G and 3G systems) or evolved MBMS
(eMBMS) (in 4G systems) is used to broadcast the Interest to
all devices in the region. The use of MBMS/eMBMS allows
to address the nodes storing the requested Data, without
the need that VNMC takes trace of contents and producers.
Furthermore, MBMS/eMBMS allows to localize the request
to relevant areas by selectively transmitting the request only
in the neighboring cells.
Note that, Interest messages are small and the load over the
cellular networks is limited.2In addition, using cellular net-
works allows the Interest messages to be reliably and quickly
delivered to producers, thus improving Data spreading. In the
case no path from the consumer to the producer is available,
Data is forwarded by producers to their neighbors, and further
Interest attempts will find new producers, increasing the
probability of Data delivery.
IV. SIMULATION SETTINGS
1) Simulator and scenarios: To address the performance
of the proposed algorithm, simulations are made adopting the
simulation platform for heterogeneous interworking networks
(SHINE) [15]. The center part of the Italian city of Bologna is
considered as case study, with two different vehicle densities.
In the first case, hereafter denoted as Bologna A, fluent traffic
2In case of adoption of the universal mobile telecommunications system
(UMTS), Interest messages can be transmitted on the common random access
channel (RACH) [4], whereas with long term evolution (LTE) all messages are
transmitted on shared common channels. Hence, the cellular network load and
cost are limited. In any case, the user will not pay the cost of each transmitted
Interest message, but a sort of subscription to the service.
Fig. 1. Simulation scenario (center portion of Bologna, 1.6 km x 1.8 km)
with RSU position.
TABLE I
SCE NAR IO DE TAIL S.
Case Area Average number
of vehicles
Bologna A
2.88 km2
455
(Normal traffic in [6])
Bologna B 670
(Heavy traffic in [6])
is considered with 455 vehicles on average; in the second case,
hereafter denoted as Bologna B, 670 vehicles are present on
average and congestions are occurring in the main junctions.
An RSU is placed at the main junction. The characteristics of
the scenario are summarized in Table I and the road layout
with the RSU placement is shown in Fig. 1.
Only a variable portion δOBU of vehicles is equipped with
the on board unit (OBU), which implements VNDN and is
provided with both short range and cellular wireless commu-
nication interfaces.
2) NDN settings and benchmark cases: To stress the case
of not too popular contents, each OBU performs the request
for a different content; requests are generated in instants
randomly chosen over the simulation duration. All contents are
initially stored outside the simulated scenario and the Data is
retrieved by the RSU with negligible delay when it receives
the corresponding Interest. Except if otherwise specified, when
the Interest is sent by a consumer, a timer is started lasting
Tretry = 30 s. Except if otherwise specified, all OBUs store in
the CS all Data they receive.
In the further, results will be shown for the following cases:
•Only short range, single request: A benchmark case
where only V2V and V2R are used and each Interest is
sent only once, without setting the repetition timer. Since
different contents are requested by different OBUs, CS is
not used;
•Only short range, no CS: A benchmark case where only
V2V and V2R are used and OBUs do not have a CS.
Data is thus stored only by the RSU and the muling by
OBUs is not exploited;
•Only short range, with CS: A benchmark case where
only V2V and V2R are used and all OBUs have a CS.
Once an Interest reaches the RSU and Data is forwarded,
all OBUs receiving that Data become producers of that
content for possible new attempts;
•Cellular aided VNDN: V2I is exploited for the In-
terest delivery, whereas V2V and V2R are used for
Data transmission.3
3) Short range communications: The wireless access in
vehicular environment (WAVE)/IEEE 802.11p technology [2],
[3] is considered for V2V and V2R communications. In
particular, WAVE defines the system architecture and the set
of services and interfaces, whereas IEEE 802.11p, which is
an amendment to the IEEE 802.11 standard conceived for
vehicular scenarios at 5.9 GHz, describes the medium access
control (MAC) and physical layer protocols. A key amendment
introduced by WAVE/IEEE 802.11p is the WAVE mode, that
allows the transmission and reception of data frames with the
wildcard basic service set (BSS) identity and without the need
of belonging to a particular BSS. Other features are a smaller
channel bandwidth than previous IEEE 802.11 versions and
the presence of a channel for control purposes and others
for service purposes. Here, we assume the use of one service
channel for the addressed service.
The MAC protocol of IEEE802.11p is simulated in details,
with the sensing and random access procedures, also includ-
ing hidden terminals, exposed terminals, and capture effects.
Concerning the physical layer, a threshold model is assumed
for the packet error rate, with a hiding effect due to buildings:
a transmission between two devices is possible only if 1) the
virtual line connecting them does not cross any building, 2)
the received power is higher than the receiver sensitivity and
3) the signal-to-noise-plus-interference ratio is higher than a
threshold. Assuming the settings listed in Table II and an
attenuation PL(d) = 47.9 + 27.5 log10(d)[16] (where dis
the distance in meters), the maximum transmission range in
the absence of interferers is dtx = 200 m.
Flooding is used as routing protocol: each OBU receiving an
IEEE 802.11p message that was not addressed to it, includes
the message in the transmission queue and broadcasts it. A
list of received and forwarded messages is updated to avoid
forwarding the same message more than once. Although not
efficient in terms of channel occupation, flooding allows to try
all possible paths.
4) Cellular communications: LTE is assumed as cellular
technology. When the proposed solution is considered, an
OBU with an Interest to send must first obtain a resource in
the uplink through a random access mechanism with duration
tLTE-acc, then transmits the message in a time tLTE-UL. Such
delays are almost negligible in our scenario, since access delay
3In case of unavailability of cellular networks (e.g., in case of a disaster
or with low coverage), the system will of course automatically use the other
available technologies.
TABLE II
SIMULATION SETTINGS.
NDN settings Value
Interest message length 50 bytes
Data message length 1000 bytes
Interest repetion timer Tretry 30 s
IEEE 802.11p parameters Value
Equivalent radiated power 23 dBm
Receiver sensitivity -85 dBm
Receiver antenna gain 3 dB
Minimum SINR 10 dB
Maximum transmission range 200 m
and round trip time are upper bounded by LTE requirements to
300 ms and 10 ms, respectively [17]; in our simulations we as-
sume a worst case with tLTE-acc = 300 ms and tLTE-UL = 10 ms.
In the downlink, eMBMS is used to broadcast the Interest; the
worst case of one frame (10 ms) is assumed as LTE downlink
delay tLTE-DL = 10 ms. For the sake of simplicity, we assume
that errors introduced by cellular links are negligible.
5) Output metrics: Simulation results are given in terms of:
•Delivery rate DR, which is the rate of contents that are
delivered to the requesting OBUs
DR,γdeliv
γgen
(1)
where γdeliv is the number of contents delivered to the
requesting OBUs and γgen is the overall number of
contents requested. DR∈[0,1];
•Average number of attempts per satisfied request NA
NA,Pγdeliv
i=1 NAi
γdeliv
(2)
where NAiis the number of Interest packets that were
sent by the OBU requesting the ith content;
•Average delivery delay L
L,Pγdeliv
i=1 tdelivi−tgeni
γdeliv
(3)
where tdeliviis the instant when the ith Data reaches the
requesting OBU and tgeniis the instant when the first
Interest for that content was generated.
V. NUMERICAL RE SU LTS
Numerical results are shown in Fig. 2 for Bologna A (fluent
traffic) and in Fig. 3 for Bologna B (congested traffic). In
both traffic conditions, DR,NA, and Lare shown for the four
detailed cases, as a function of the portion of vehicles equipped
with OBU, δOBU.
In Bologna A, when δOBU is low, communication between
consumers and producers is difficult. Focusing on Fig. 2(a)
with δOBU = 0.25, when a single attempt is made by
consumers, less then 30% of Data is delivered. The use of
timers and new attempts increases DRsignificantly, but it
does not exceed 50%. Interestingly, the use of CSs does not
change significantly the results in terms of DRcompared to
the case without CSs. This can be explained by observing that
0 0.2 0.4 0.6 0.8 1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
δOBU
DR
Cellular aided VNDN
Only short range, single request
Only short range,
no CSs
Only short range,
with CSs
(a) Delivery rate vs. rate of equipped vehicles.
0 0.2 0.4 0.6 0.8 1
0
0.5
1
1.5
2
2.5
3
δOBU
NA
Cellular aided VNDN
Only short range,
single request
Only short range, with CSs
Only short range, no CSs
(b) Average number of attempts vs. rate of equipped vehicles.
0 0.2 0.4 0.6 0.8 1
10−2
10−1
100
101
102
δOBU
L [s]
Cellular aided VNDN
Only short range,
with CSs
Only short range, no CSs
Only short range, single request
(c) Average delivery delay vs. rate of equipped vehicles.
Fig. 2. Performance in scenario A.
Data delivery failures are mostly due to the absence of a route
between the producer and the consumer; under such condition,
many Interest messages also fail to reach their destination,
and the Data is not spread by producers to other nodes. The
presence or absence of CSs on board of OBUs is thus not
effective.
On the contrary, when cellular networks are used to carry the
Interest messages, each attempt causes Data to be forwarded
from producers to neighbor nodes, that can mule Data mes-
sages in various directions. The following attempts will thus
exploit more producers in new positions and the probability
that the Data reaches the consumerincreases. In the case with
lowest connectivity, i.e. in Bologna A with δOBU = 0.25, the
cellular aided VNDN brings to DR= 0.63, allowing 15%
more contents to be delivered compared to using only V2V
and V2R. Even if the gap among solutions reduces increasing
δOBU, the use of V2I allows the highest DReven when the
connectivity level increases. In normal traffic conditions, the
increased DRcomes with similar average attempts and similar
delays compared to the other cases that foresee the repetition
of requests, as shown through Figs. 2(b) and 2(c).
What happens in congested traffic conditions is shown in
Fig. 3. As observable in Fig. 3(a), a single attempt allows
the request to be satisfied in 40% to 80% of cases, depend-
ing on the rate of vehicles equipped with the OBUs. With
δOBU >0.75, the presence of a timer and the use of repeated
requests allows to increase DRto more than 0.9, regardless the
presence of CS or not; this behavior confirms that the increased
DRis mainly due to the mobility of vehicles, and that the
Data storing performed by nodes tends to be ineffective.
The use of cellular connections for Interest delivery allows
to further increase DR, reaching 0.98 when all vehicles are
equipped with the OBU. Besides an higher DR, in this case
cellular aided VNDN also allows lower values of NAand L
(Figs. 3(b) and 3(c)); in such case, avoiding the transmission
of Interest messages among vehicles reduces the number
of WAVE/IEEE 802.11p transmissions performed by OBUs
and allows Data messages to be delivered quicker and more
reliably.
VI. CONCLUSION
In this paper, a novel solution for VNDN messages ex-
changes is proposed. Instead of using WAVE/IEEE 802.11p
to convey all messages, we proposed to carry the signaling
part of VNDN over cellular networks. This solution still
allows offloading cellular networks of Data exchanges, yet
guaranteeing a higher probability that content requests reach
the interested nodes. Where connectivity cannot be guaran-
teed (i.e., in sparse networks), nodes storing the contents
are reached through the cellular connection and can forward
Data toward the requesting node; further requests may also
take advantage of the presence of more (and closer) content
sources. The performance of the proposed solution has been
assessed through simulations performed in an urban scenario,
with two different vehicle densities. Our proposal was shown
to significantly improve the delivered contents compared to
0 0.2 0.4 0.6 0.8 1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
δOBU
DR
Cellular aided VNDN
Only short range, single request
Only short range,
no CSs
Only short range,
with CSs
(a) Delivery rate vs. rate of equipped vehicles.
0 0.2 0.4 0.6 0.8 1
0
0.5
1
1.5
2
2.5
3
δOBU
NA
Cellular aided VNDN
Only short range,
single request
Only short range, with CSs
Only short range, no CSs
(b) Average number of attempts vs. rate of equipped vehicles.
0 0.2 0.4 0.6 0.8 1
10−2
10−1
100
101
102
δOBU
L [s]
Cellular aided VNDN
Only short range,
with CSs
Only short range, no CSs
Only short range, single request
(c) Average delivery delay vs. rate of equipped vehicles.
Fig. 3. Performance in the scenario B.
previous approaches, in particular when the node density is
low. When the density increases and the connectivity level
becomes high, the use of short range communications alone
allows high delivery probability; nevertheless, our proposal
was shown to further improve content delivery and reduce
transmission delays.
REFERENCES
[1] GSMA, “Connected car forecast: Global connected car market to grow
threefold within five years,” GSMA, Tech. Rep., Feb. 2013.
[2] “IEEE trial-use standard for wireless access in vehicular environments
(WAVE) - multi-channel operation,” pp. c1 –74, 2006.
[3] “Standard for information technology- telecommunications and informa-
tion exchange between systems- local and metropolitan area networks-
specific requirements part 11 - amendment 6: Wireless access in vehic-
ular environment,” IEEE, 2010.
[4] A. Bazzi, B. Masini, and O. Andrisano, “On the frequent acquisition
of small data through RACH in UMTS for ITS applications,” Vehicular
Technology, IEEE Transactions on, vol. 60, no. 7, pp. 2914–2926, 2011.
[5] S. Uppoor, O. Trullols-Cruces, M. Fiore, and J. Barcelo-Ordinas, “Gen-
eration and analysis of a large-scale urban vehicular mobility dataset,”
Mobile Computing, IEEE Transactions on, vol. PP, no. 99, pp. 1–1,
2013.
[6] A. Bazzi, B. Masini, and G. Pasolini, “V2V and V2R for cellular
resources saving in vehicular applications,” in Wireless Communications
and Networking Conference (WCNC), 2012 IEEE, 2012, pp. 3199–3203.
[7] V. Jacobson, D. K. Smetters, J. D. Thornton, M. F. Plass, N. H. Briggs,
and R. L. Braynard, “Networking named content,” in Proceedings of the
5th International Conference on Emerging Networking Experiments and
Technologies, ser. CoNEXT ’09. New York, NY, USA: ACM, 2009,
pp. 1–12.
[8] G. Grassi, D. Pesavento, G. Pau, R. Vuyyuru, R. Wakikawa, and
L. Zhang, “Vanet via named data networking,” in Computer Communi-
cations Workshops (INFOCOM WKSHPS), 2014 IEEE Conference on,
April 2014, pp. 410–415.
[9] L. Wang, R. Wakikawa, R. Kuntz, R. Vuyyuru, and L. Zhang, “Data
naming in vehicle-to-vehicle communications,” in Computer Communi-
cations Workshops (INFOCOM WKSHPS), 2012 IEEE Conference on,
March 2012, pp. 328–333.
[10] M. Amadeo, C. Campolo, and A. Molinaro, “Crown: Content-centric
networking in vehicular ad hoc networks,” Communications Letters,
IEEE, vol. 16, no. 9, pp. 1380–1383, September 2012.
[11] P. Talebifard and V. C. Leung, “Towards a content-centric approach to
crowd-sensing in vehicular clouds,” Journal of Systems Architecture,
vol. 59, no. 10, Part B, pp. 976 – 984, 2013, advanced Smart
Vehicular Communication System and Applications. [Online]. Available:
http://www.sciencedirect.com/science/article/pii/S1383762113001501
[12] S. Ata, H. Kitamura, and M. Murata, “Information-centric communica-
tion architecture for vehicular networking,” in Connected Vehicles and
Expo (ICCVE), 2013 International Conference on, Dec 2013, pp. 853–
854.
[13] L. Wang, A. Afanasyev, R. Kuntz, R. Vuyyuru, R. Wakikawa, and
L. Zhang, “Rapid traffic information dissemination using named data,”
in Proceedings of the 1st ACM Workshop on Emerging Name-Oriented
Mobile Networking Design - Architecture, Algorithms, and Applications,
ser. NoM ’12. New York, NY, USA: ACM, 2012, pp. 7–12. [Online].
Available: http://doi.acm.org/10.1145/2248361.2248365
[14] A. Bazzi, B. M. Masini, A. Zanella, and G. Pasolini, “On the use
of ofdma for next generation vehicular ad hoc networks,” in Personal
Indoor and Mobile Radio Communications (PIMRC), 2013 IEEE 24th
International Symposium on, Sept 2013, pp. 2223–2228.
[15] A. Bazzi, G. Pasolini, and C. Gambetti, “SHINE: Simulation platform
for heterogeneous interworking networks,” in Communications, 2006.
ICC ’06. IEEE International Conference on, vol. 12, 2006, pp. 5534–
5539.
[16] L. Cheng, B. Henty, D. Stancil, F. Bai, and P. Mudalige, “Mobile vehicle-
to-vehicle narrow-band channel measurement and characterization of the
5.9 ghz dedicated short range communication (DSRC) frequency band,”
IEEE J. Sel. Areas Commun., vol. 25, no. 8, pp. 1501 –1516, Oct. 2007.
[17] H. Holma and A. Toskala, LTE for UMTS. Wiley, 2009.
