The role of ad hoc networking in future wireless communications

Abstract

This paper discusses the role of ad hoc networking in future wireless communications. Ad hoc networks are classified as isolated ad hoc networks with large and small sizes, integrated ad hoc networks in various scenarios and cellular ad hoc networks for the future mobile access networks. The very low traffic performance of large-scale ad hoc networks is shown by simulation results, indicating such a king of networks having little commercial potential. Small size ad hoc networks seem to be ubiquitous because of the availability of cheap wireless LAN technologies. Integration of small size ad hoc networks with the global Internet can be realized by ad hoc gateways, which are proposed in this paper. In contrast to large scale isolated ad hoc networks, traffic performance of a cellular ad hoc network is very promising, indicating that cellular ad hoc networking seems to be a promising solution to fulfill the requirements of future wireless communication systems.

Full text

Proceedings
of
ICCT2003
The Role
of
Ad
hoc Networking in Future Wireless Communications'
Bangnan
Xu,
Sven
Hischke
T-Systems, Technologiezentrum, Am Kavalleriesand
3,
D-64295
Darmstadt,
Germany
Email:
Bangnr~.Xu~.t-s~stems.conz Sven.HischkeG3,t.t-sstems.com
Bernhard Walke
Communication Networks, Aachen University of Technology,
Kopernikusstr.
16,
D-52074 Aachen,
Germany
Email:
Wfkc~~m~11.comncts-aachen.de
Abstract-This paper discusses the role of ad hoc
networking
in
future wireless communications. Ad hoc
networks are'classified
as
isolated ad hoc networks with large
and small sizes, integrated ad hoc networks in various
scenarios and cellular ad hoc networks for the future mobile
access networks. The very low traffic performance of large
scale ad hoc networks
is
shown
by
simulation results,
indicating such a kind
of
networks having little commercial
potential Small size ad hoc networks seem to
be
ubiquitous
because of the availability of cheap wireless
LAN
tech-
nologies. Integration of small size ad hoc networks with the
global Internet can
be
realized by ad hoc gateways, that are
proposed in
this
paper. In contrast
to
large scale isolated ad
hoc networks, traffic performance of a cellular ad hoc
network is very promising, indicating that cellular ad hoc
networking seems to
be
a promising solution
to
fulfill the
requirements of future wireless communication systems.
Keywords-ad
hoc
networking;beyond
3C;
mukihop;
ad
hoc
routing; mobile
IP;
gateway; traff7c performance.
I. INTRODUCTION
Ad hoc networking is
an
increasingly important topic in
wireless communications
and
has been regarded
as
one
of
the key features of beyond
3G
systems. In the
1970s
and
1980s
the
research
on ad hoc networking was mainly for
the military purpose
[l].
The typical scenario
of
ad hoc
networks at that time was to set up a communication
network in a battlefield, where no
infrastructure
could
be
available. Each node in
an
ad hoc network may work as a
router to relay connections or
data
packets to their
destinations. The key issues of
ad
hoc networking
are
MAC (Medium Access Control), that is used to share
common channel resources among wireless nodes, and
routing, that is to find a route between
source
and
destination nodes across a number
of
possible relay nodes.
Since
1990s
ad
hoc networking is increasingly important in
the commercial and residential
mas.
The
reasons
for that
are
as
follows:
(1)
With
the
increase of small
size
information processing devices, such
as
laptop, pocket PC,
PDA etc.,
the
need to exchange
digital
information, such
as
video, music and documents among people within a short
range is ever increasing.
(2)
The emerging wireless
technologies, such
as
Bluetooth,
IEEE
802.11,
ETSI
HiperLAN
2
and W-CHAME3 (Wireless Channel Oriented
Ad hoc Multihop Broadband Networks)
[2-11
J
enable ad
hoc networking among a number
of
wireless nodes.
(3)
Ad
hoc networking has been proven to be
a
promising solution
to increase the radio coverage of broadband wireless
systems
[ll].
(4)
Ad hoc networking is an important
method to extend the multimedia services
in
Internet
to
wireless environments
[15].
This paper presents the possible scenarios and applications
of
ad
hoc networks
in
future
wireless communications.
Advantages and disadvantages of various scenarios
are
discussed and verified by computer simulations.
Challenges
and
corresponding solutions to
realize
ad
hoc
networks concerning
MAC
protucols, ad hoc muting,
mobility management, security and business cases
are
also
discussed.
11.
ISOLATED
AD
HOC
NETWORKS
An isolated
ad
hoc network means that all nodes
communicate with each other within the same ad hoc
network. The isolated
ad
hoc network has no connection
with any infrastructure-based communication network,
such
as
the global Internet. Isolated ad hoc networks
can
be
classified as .two types: large scale isolated ad hoc
networks
and
small
size isolated ad hoc networks.
A.
Large
Scale Isolated Ad hoc Networks
A large scale isolated ad hoc network may have hundreds
to thousands or even more nodes. Some researchers have
proposed
to
apply
a
large
scale
ad hoc network to form a
radio
MAN
(Metropolitan Area Network) or even
a
WAN
(Wide
Area
Network) to compete with the wired
telecommunication networks.
This
idea
has
been proven
unpractical. The large scale ad hoc network
is
suited moxe
or less for
the
academic purpose as the related problems
are
very challenging., However, large
scale
ad hoc
networks have little commercial potential. It may be
used
to transport some very important messages with a small
amount
of
data in some special cases, such
as
in
a
battlefield to transport some commands or
on
highway to
transport a warning message to inform other vehicles about
an
accident or a traffic congestion. A large scale ad hoc
network is not suited to be used
as
a communication
network that needs to transport a large amount of
information. The reason
is
that a large ad hoc network
This
work
is
partly
supported by the German Federal
Ministry
for Education and Research.
1353

causes
severe security problems, extremely high network
organization costs and very low traffic performance.
It
is
very difficult to implement a security mechanism in a
large scale isolated ad hoc networks. How could the
involved nodes trust each other? The
usual
AAA
(Authentication, Authorization and Accounting)
mechanism is infeasible, if not impossible, to be
implemented
in
a large scale isolated ad hoc network due
to
the lack of
a
central control server.
The signaling cost of an ad hoc network increases
essentially with the increasing size of the ad
hoc
network.
In ad hoc routing protocols proposed as
Dnfts
in
IETF,
like AODV and DSR [12,13], the
source
node needs to
flood a route request to find a route to the destination node.
If
some nodes in an ad hoc network
are
mobile, flooding of
route requests happens frequently. In
a
large scale ad hoc
network, flooding of route requests may use
up
most of the
channel resources.
Even
if
we do not consider the security and routing
problems in
a
large scale isolated
ad
hoc network,
the
traffic performance in terms
of
delay and throughput
decreases signifcantly with the increase
pf
mean hops of
connections.
I)
Security problem
2)
Cost
of
routingpackets
3)
Traffic Performance
R:h.nnnitfrsosivaw
Figure
1.
An
isolated ad
hoc
network
To
study the impact of mean hops
of
connections on the
traffic performance, we simulate
a
network with 25 nodes,
shown in Fig. 1.
A
desired network connectivity
c,
that
is
reverse to the mean hops of connections, is achieved by
adjusting the transmit
power
of the wireless nodes
accordingly. The connectivity
c
is defined
as
the mean
number
of
neighbors, normalized
by
the number
of
the
maximum possible number of neighbors.
18
I
I
I I
16
16
QAM
ln
24
Mb/a
14
12
10
0.2
0.4
Ob
0.8
1
Traffic
Id
-0
Figure
2.
Throughput
with
an
isolated ad
hoc
network
0.001
LX-,
IEEE802.11a
-
I
I
I
:
8%
x::.
-
02
0.4
0.6
0.8
1
0.m1
Trnffic
Id
Figure
3.
Mean
delay with
an
isolated ad
hoc
network
where
n,
is the number of neighbors to node
i,
N
is the
number of nodes in the network.
A
fully connected
network
has
a connectivity of 1 [8-111.
Fig
2.
and
Fig. 3 show the traffic performance of W-
CHAMB and IEEE 802.11a
[5,
7-11] in a multihop
environment with various network connectivities. Realistic
packet sizes read
from
an
Ethernet trace file [14]
are
used
for traffic loads. Fig.
2
shows the traffic performance with
the transmission rate of 24 Mbit/s with a network
connectivity of 0.24, 0.34 and 1.0, respectively. For both
systems, W-CHAMB and IEEE 802.11a, the smaller the
network connectivity
is,
the lower is the throughput.
This
is
because at
a
smaller network connectivity more hops
are
needed to establish an end-toend connection. The
system
capacity is used up rapidly owing to the multihop
transmissions.
The benefit of frequency spatial reuse
in
multihop networks is reverse to that
of
cellular networks
where multiple access points (base stations) exist.
The superiority
of
the traffic performance of
W-
CHAMB is very significant
for
all connectivities. The
maximum throughput with W-CHAMB and with
IEEE
802.1 la is 14.2 Mbit/s and 11 Mbit/s with
c
=
1.0,9
MbWs
and
7
Mbit/s
with c=0.34, and
7.2
Mbit/s and
5
Mbit/s with
c=
0.24,
respectively. The efficiency
of
IEEE 802.11a
becomes extremely worse
with
a
small
network
connectivity as it suffers from the hidden
station
problem.
The impact
of
the network connectivity on the mean delay
performance can be seen in Fig.
3.
With a larger network
connectivity, a better delay performance can be achieved.
We
can
see
that IEEE 802.11a
has
a
better overall mean
1354

delay performance
than
W-CHAMB as long as it is not
saturated. The mon is that IEEE 802.11a can transmit a
packet
as
large
as
2304
bytes, whereas W-CHAMJ3
has
to
segment large packets into 102 byte fragments. The
fragments are transmitted over a number of W-CHAMB
MAC frames
as
a PDU train. But IEEE 802.1 la goes into
saturation with
a
much lower traffic load
than
W-CHAMB
can
carry.
It is
worth
mentioning that IEEE 802.1 la is not
able to differentiate between service classes and the mean
delay of all service classes together is a too rough measure
to
characterize
a
system.
From the simulation results we could conclude that a large
scale isolated
ad
hoc network having a small network
connectivity is not suited
to
be
used
as
a communication
network due to the lack of network capacity needed. Large
scale isolated ad hoc networks could
,
therefore, have little
commercial meanings. In comparison with W-CHAMB,
IEEE 802.11a
is
extremely inefficient in multihop
networks.
B.
Small Size Isolated
Ad
hoc Networks
In contrast
to
large scale isolated
ad
hoc networks, small
size ad hoc networks seems to be ubiquitous and have high
commercial potential in home environments, business
meeting places, hotspots, and even in personal areas. The
emerging wireless
LAN
technologies like ETSI HiperLAN
2,
IEEE 802.11, Bluetooth enable small size ad hoc
networks to be deployed easily. Small size ad hoc networks
will take
a
significant role in connecting home appliance,
exchanging documents during meetings, instant digital
presentations, and even playing a game among a group
of
people for the entertainment purpose.
To meet the need of a member of an ad hoc
group
to access
the global Intemet or to be reached over Internet, an ad hoc
network is connected to global Intemet using ad hoc
gateways [MI.
This
is the issue of integrated
ad
hoc
netwodcs introduced in
the
next Section.
In.
INTEGRATED
AD
HOC
NETWORKS
A.
scenario
are discussed.
1)Hotspot scenario
Scenarios to integrate ad hoc networks with Internet
In
this
Section, hotspot scenario and GPRSAJMTS
Fig.
4
shows
a
hotspot
scenario
using three
kinds
of
gateways, WA-GW (Wireless Ad hoc Gateway), MA-GW
(Mobile Ad hoc Gateway) and MH-GW (Mobile Hotspot
Ad hoc Gateway) to integrate a small size ad hoc network
with an IP-based broadband access network. The seamless
integration of ad hoc networks with the access network is
realized by ad hoc gateways. There
are
several reasons that
an
ad
hoc MT
(A-MT)
in the ad hoc network may not
be
able to access the
AP
directly. Firstly, the air interface
used
in
the ad hoc network
may
be different from that
used
in
the access network. Secondly, an A-MT
in
the ad hoc
network may be out of the coverage of
any
AP.
Thirdly, an
A-MT in the ad hoc network may have not a valid IP
address to access the global Internet. Hence, A-GWs aE
necessary to enable A-MTs in the
ad
hoc network to
acce~
the global Intemet. As A-MTs in the
ad
hoc network may
/
Global
Internet
W
Figure
4.
Hotspot
scenario
move during the communication, a dynamic routing
protocol is necessary to route
IP
packets from an A-MT to
the A-GW. Several Draft
proposals
in
LETF,
such
AODV
and DSR may be used as the routing protocol in the ad hoc
networks. But these protocols
are
developed for
a
pure
ad
hoc network.
An
adaptation is necessuy to integrate
ad
hoc
routing protocols with the mobile IP routing protocols, that
will be used in the infrastructure based
routing
domain. A-
GWs have the ability to route
IP
packets in different
routing domains. The access network consists of a gateway
(GW), switches and access points
(APs).
The
GW
is
connected to the global Internet. Depending on the size of
the coverage area
one
or more switches are connected to
the gateway. A number of
APs
are
connected to each
switch. The network is scalable and
can
achieve
a
radio
coverage of any size. More GWs may be used
to
form
a
very large coverage, such
as
a whole city. The GWs can be
organized
in
a hierarchical way to realize
a
hierarchical
micro-mobility support.
In
the access
network
Layer 2 (L2)
packet switching is integrated with the Layer
3
(L3)
IP
routing to realize the fast and efficient mobility support.
The mobility of a mobile terminal
(MT)
within
the
same
access network is dealt with
L2
routing tables at the
switches
and
the gateway of the access network.
In
this
case the IP layer will not be aware of the MT movement.
If
the MT changes the access network, at IP layer a mobility
protocol is
used
to deal with
the
MT
movement. Through
an hierarchical update of the L2 routing table, the access
network with efficient micro-mobility support realizes
seamless mobility
of
IP services.
Fig.
5
shows GPRSMS scenario where a member of ad
hoc group can access to Internet anywhere
and
any
time
using GRPSMS through a MA-GW or
MH-GW.
In
comparison with the Hotspots scenario the
data
rate is
limited
in
GPRSAJMTS
scenario.
But
this
scenario
is
important for the mobile hotspot cases, such
as
in
buses
and trains, where W-LAN-based access to the Internet is
not available during their moving time. A handoff between
the hotspot scenario and GPRSAMTS scenario is
si@icant for the mobile hotspot applications. That means
2)GPRS/UMTS
scenario
1355

public
Figure
5.
GPRSNMTS
scenario
that
at bus or railway station the Internet access of ad hoc
nodes is realized over the W-LAN-based access network,
while
GPRS/UMTS
is used during the moving time.
B.
Protocols
and Busines
models
Nodes in an integrated ad hoc network use three kinds of
gateways to access infrastmcture-based wireless networks.
(1)
WA-GWs
are
usually installed by a network operator
for the public use and is not moved during communication.
(2)
MA-GWs
are mobile
and
owned by a private end-user.
So
a
MA-GW
is instable, uncontrolled and may be moving
during communications.
(3) MH-GWs
are
mobile
concerning to access networks, but static to the ad hoc
networks.
MH-GWs
are
stable and owned by a network
operator. In addition to extend the Internet multimedia
services to ad hoc networks,
WA-GW
and
MH-GW
can
also be used to provide local contents to ad hoc networks.
Operators of
A-GWs
may
charge ad hoc users by providing
Internet services, local contents and
AAA
services to form
an
ad
hoc network. Operators could also encoilrage
ad
hoc
users
to
work
as
relay nodes
or
even
A-GWs
by granting
bonus to them.
The basic operation of the ad hoc integration can be
described
as
follows:
An
A-MT
finds an
AGW
before it
can access the global Internet.
An
A-GW
broadcasts its
existence periodically. If the
A-MT
is one hop away from
the
A-GW,
it can find the
A-GW
through receiving the
broadcast message.
If
the distance between the
A-GW
and
A-MT
is
more
than
one
hop, the
A-MT
sends a control
message
A-GW
Search
that includes IP"Home Address
and
MAC
address of the
A-MT.
After that the
A-MT
may
receive one or more replies from an
A-GW
or
A-GWs.
The
MT
selects the
A-GW
with fewest hops
as
its
A-GW
to
access the Internet. The destination address
of
the
ad hoc
packet including
this
A-GW
Search
is
a special address
indicating that
an
A-GW
is searched. After
an
A-GW
is
found, the
MT
tries to obtain
an
IP
CoA
(Care of Address)
with the help
of
the
A-GW.
Ip
packets generated at the
A-
MT
are encapsulated in the
ad
hoc packets
[15]
to be
routed to the
A-GW.
The ad hoc routing header is formed
according to the ad hoc routing protocol like
DSR
or
AODV.
The
A-GW
removes the ad hoc routing header
and
forms
a new packet that is muted to
the
access network,
Figure
6.
A
cellular
ad
hoc
network
where the
IP
packet is recovered and routed to the global
Internet. On
the
other direction
IP
packets coming from the
global Internet can
be
routed to
the
A-GW
where ad hoc
packets
are
formed and routed to the
A-MT
using the
ad
hoc routing header.
Iv. CELLULAR
AD
HOC
NETWORKS
Cellular ad hoc networks
are
self-organizing multihop
networks with multiple access points that are connected
with a broadband core network. The difference of a cellular
ad hoc network from an isolated ad hoc network is that
most of the traffic in the cellular ad hoc network is to/from
access points.
A.
After the standardization of the
3G
(3rd
generation)
systems
lMT-2000/UMTS,
researchers are beginning to
develop concepts and technologies for wireless beyond
3G
systems. Higher transmission rate and lower system cost
are
regarded
as
the key requirements of the beyond
3G
systems.
A
new air interface concept to form a
W-CHAMl3
network, that is suited for the wireless multimedia
communications beyond
3G
is presented in
[ll].
Its
main
features are self-organizing, multihop transmission and
QoS
guarantee.
As
major applications of beyond
3G
systems will require a transmission rate
10
times higher
than
that of
3G,
the beyond
3G
systems must use frequency
bands over
3GHz
in the view of availability and feasibility
of frequency spectrum to support high rate sexvices. The
respective fRquencies have very limited ability
to'
penetrate obstructions
and
have very irregular propagation
characteristics
so
that frequency planning is very difficult
there. In addition, higher transmission rates and higher
frequency bands result in a very limited communication
range
and
an
irregular
radio coverage. Two approaches
might be used to achieve a reasonable radio coverage. One
is to
increase
the number of base stations and the output
poker.
This
method might increase the system cost
significantly. The other one uses cellular ad hoc networks
Cellular ad hoc network
m
beyond
3G
Jrstems
1356

with multihop transmissions to extend the radio coverage
and realize a cost-effective solution.
B.
To study the traffic performance of cellular ad hoc
OJ
-
networks, an access scenario with 5
APs
is studied. Each
8
node in the network communicates with the nearest
AP
9
0.6
-
using up- and download
ABR
tdic connections. The
are
varied to model the various t&ic loads. Fig. 6 shows
the cellular ad hoc network with 5
Aps,
each connecting to
the fixed core network.
A
9 x 9 grid network is used. The
communication range is varied from 1.42d to 2.9d
to
model
the different network connectivity. d is the distance
between two neighbors. With a small communication
range, multihop connections must be established using the
Minhop algorithm if the communication partner is not the
12
I
I I
I
I
Traflc performance
of
cellular ad hoc networks
1-
Wic load is read from the Ethernet trace file. Data rates
FI
0.4
-
02
-
WCHAMB-
I
IEgaKq.Ll.----.
02
0.4
Od
1
11
T&l&
Figure 7. Throughput with a cellular
ad
hoc network
direct neighbor. The OFDM scheme 16
QAM
1/2 is used
in the simulation giving a transmission rate of 24 Mbit/s
[4,51.
In contrast to the isolated ad hoc network, shown in Fig. 1,
traffic performance can be improved with a reduced
transmission range in a cellular ad hoc network. Fig.
7
shows that with a communication range of 1.42d, a
maximum throughput more
than
1.0 can be achieved
because of the spatial reuse of the frequency channels.
With a transmission range of 2.9d, all nodes can reach the
AP
with
an
one-hop connection. The maximum throughput
is only 0.65 in
this
case. The reason is that with a
transmission range of 1.42d, the bottleneck at the
AP
is
sirrnificantlv reduced
as
the interference
of
other nodes
1
0.
I
e
v
g
WcHMlB-
I I I
,
IEEBw?11.----.
0.2
0.4
0.8
'1
13
T&ld
Figure 8. Mean delay with a cellular ad hoc network
d&reases. ?he benefit of the frequency spatial reuse can be
gained in cellular ad hoc networks. The mean delay
performance with a transmission range of 1.42d is
also
much better
than
with a transmission range of 2.9d with a
moderate to high traffic load, see Fig. 8. Only with a very
'
low trafftc load, the mean delay with a transmission range
of 1.42d is a little worse than that with a transmission
range of 2.9d due to multihop transmissions. In comparison
with W-CHAMB, the maximum throughput with IEEE
802.11a is much lower in
all
cases. With
R
=
1.42d, the
maximum throughput with W-CHAMB is 1.05, about
70%
more than that with IEEE 802.1 la
(0.65).
The mean delay
with IEEE 802.11a is lower than that with W-CHAMB at
the low traffic load. The reason is the same as explained in
Section
II.A.3.
V.
CONCLUSIONS
Large scale isolated ad hoc networks
are
more or less of
academic values with little commercial meanings due to
the lack of network capacity to cany the ever increasing
amount
of multimedia information. Small size ad hoc
networks will be ubiquitous from the home environments
to
business meeting places. To meet the need of a member
of ad hoc group to access the global internet or to be
reached over the global Intemet, various ad hoc gateways
can
be
used to integrate
ad
hoc networks with the global
Intemet. In contrast to the large scale isolated ad hoc
networks, a reduced transmission power in cellular ad hoc
networks can improve the system performance as the
bottleneck effect in the
AP
is reduced because of
the
reduced interference. The radio coverage of
APs
can be
enlarged using
multihop
transmissions
without
the
need
of
increasing transmission power
so
that the system cost can
keep very low. The organization remains simple by
limitation of the transmission hops. Frequcy planning
that
is very Micult in the frequency range above
3G
Hz can be
avoided by self-organization of
APs
and mobile nodes.
Cellular ad hoc networks seem to be a promising solution
for broadband wireless access networks in beyond
3G
systems.
REFERENCES
[
11
M.
Leiner, D.L. Nielson, F.A. Tobagi (Guest Editor),
Special
Issue
on
Packet Radio Networks,
Proceedings
of
the
IEEE,
vol. 75, no.
1.
Jan.
1987.
[2]
ETSI
BRAN, HiperLAN
-
Type
2,
Data Link Control (DE) Layer,
Part 1: Basic Transport Functions, Standard
TS
101 761-1, Sophia
Antipolis, France, Dec. 1999.
[3]
ETSI
BRAN, HiperLAN
-
Type
2,
Data Link Control (DLC) Layer,
Part 4: Extension
for
Home Environment, Standard
TS
101 761-4,
Soph
[4]
ETSI
BRAN, HiperLAN
-
Type
2,
Physical (PHY) Layer Standard
TS
101 475, Sophia Antipolis, France, Dec. 1999.
[5]
IEEE,
High Speed Physical Layer
in
the
5
GHz band,
Draft
Supplement to Standard
IEEE
802.1
1,
New
York,
Jan. 1999.
[6]
IEEE,
Wireless LAN Medium Access Control (MAC) and Physical
Layer (PHY) Specifications, Standard,
IEEE
802.11, New York,
Nov. 1997.
1357

[7l
B.
Walke, Mobile
Radio
Networks, New
York,
USA
Wiley
&
Sons
Lid.,
2.
ed.
2001
[8]
B.
Xu,
Self-organising Wireless Broadband Multihop Networks with
QoS
Guarantee, PHD Thesis, Aachen Univetsity
of
Technology,
[9] B.
Xu,
B. Walke, W-CHAMB: “A Wireless Channel-Oriented Ad
hoc Multihop Broadband Network
-
Comparion with IEEE 802.1
I”,
in
Roc. European Wireless’99, Munich, Germany, October 1999,
pp.
79-84,
available at
www.comnets.rwth-aachen.de.
[IO]
B.
Xu,
B. Walke, “Design
Issues
of
Self-Organidng Broadband
Wireless Networks”, Computer Networks, Vo1.37, Sept. 2001,
pp
73-8
1.
[Ill
B.
Xu,
B.
Walke,
“A New
Air
Interface Concept
for
Wireless
Multimedia Communications beyound the 3rd Generation”,
ISBN 3-86073-982-4.
Wireless Personal Communications, vol. 23, no.
1,
Oct.
2002, pp
[I21
D.
B.
Johnson,
D.
A. Maltz, Yih-Chun
Hu,
J.
G. Jetcheva,
The
Dynamic Source Routing Protocol for Mobile Ad
Hoc
Networks
@SR)
(work in progress), Internet
Dr&
draft-ietf-manet-dsr-06.txt,
21 November 2001.
[I31
C.
Perkins, E. Belding-Royer,
S.
Das,
Ad hoc On-DemandDistunce
Vector
(AODV
Routing
(work in
progress),
Internet
Draft,
draft-
ietf-manet-aodv-lOlxt, 19 January 2002.
[I41 Bellcore, Ethernet Tracefile pAug.T Available
via
Au.tclcor~lia.com’uubiworld!wel/l~
trafik.
[15]
B.
Xu,
S.
Hischke,
An
IP-Based broadband
Access
Network
Architecture supporting Ad hoc Networking Integration, in
Proc.
of
Eurescom Summit 2002, Heidelberg, Germany,
Oct.
2002, pp 417-
423.
121-135
1358