. 13
( 87 .)


to a network. Detecting access points is relatively easy even when they do not broadcast
their so-called SSID periodically (which they do more often than not), and since most of
the access points provide access to a network with a DHCP server, attaching to foreign
networks is a relatively easy process for hackers. WEP was supposed to provide an op-
tional encryption service in the MAC layer to enable the communication between access
points and clients that share the same secret key. With WEP enabled, the MAC layers will
encode each IEEE802.11 frame before transmission with an RC4 cipher (by RSA Securi-
ty) using a 40, 64, or 128 (WEP-2) bit key and a pseudorandom 24 bit number, whereas
the other side will decode the same stream using the same key and random number. The
random number is used to increase the lifetime of the key, yet it has been shown that in a
busy network, just by listening to the channel for a while, keys can be easily decoded if
the original shared key remains the same [6, 9].
In the rest of this section, readers will be introduced to the IEEE802.11 variants (Task
Groups), starting with the most popular IEEE802.11b or Wi-Fi 2.4 GHz, and continuing
with the strongly emerging IEEE802.11a or Wi-Fi 5.2 GHz. Some insight will be provid-
ed into the soon-to-be approved IEE802.11g and the other Task Groups™ work (e.g., TGs c,
d, e, f, g, and h). IEEE802.11b (Wi-Fi 2.4 GHz). The goal of Task Group b was to increase the
maximum bit rate in the 2.4 GHz frequency range while maintaining interoperability with
the original standard. The standard was released in 1999, keeping the original MAC layer
but redefining the PHY layer to only work with DSSS, thus increasing the spectral effi-
ciency of the three channels with bit rates of up to 11 Mbps each (with fall-back rates of
5.5, 2, and 1 Mbps). It did not take long for Wi-Fi to become widely accepted throughout
the world for corporate WLANs, wireless home networks, and so-called hotspots at air-
ports and caf©s, as well as by the ad hoc networking community as an easy-to-set up basis
for ad hoc testbeds. IEEE802.11a (Wi-Fi 5.2 GHz). Although Task Groups a and b were estab-
lished at the same time and the standards were accepted at the same time, IEEE802.11a
products did not arrived on the market until late 2001 due to technological difficulties.
The goal of Task Group a was to port IEEE802.11 to the newly available U-NII at 5.2
GHz and to provide higher bit rates. Thus, the original MAC layer was kept and the
PHY was reworked to provide rates of up to 54 Mbps (with fall-back rates of 48, 36, 24,
18, 12, 9, and 6 Mbps). Since the available band at U-NII is about 300 MHz, eight
nonoverlapping bands were defined; thus, eight different IEEE802.11a-based WLAN
networks can operate in the same space without interference. This is essential to build
cellular kinds of structures, in which neighboring cells should not use the same fre-
quency (to reduce interference). With eight different bands (compared to three with
IEE802.11b), it becomes relatively easy to establish noninterfering cellular structures.
DSSS was not efficient at working with these high bit rates while satisfying frequency
regulatory specifications, so a new spectrum spreading technology called OFDM
(Orthogonal Frequency Division Multiplexing) or COFDM (Code OFDM) was accept-
ed. OFDM was specifically developed for indoor environments, addressing indoor-
specific fading effects.
With OFDM, the signal to be transmitted is modulated over several frequency carriers.
In IEEE802.11a, a 20 MHz bandwidth channel is divided into 52 subcarriers, each about
300 kHz wide; 48 of these subcarriers are used as carriers for the data, whereas the re-
maining four are employed for forward-error correction. Modulation is performed by

changing the phase and amplitude of each of the subcarriers. To provide different symbol
rates, different levels of amplitudes and phase shift keying are employed (e.g., binary
phase shift keying, 16-level shift keying, etc.).
Although the power attenuation due to distance is at least four times as much at the 5.2
GHz range than at the 2.4 GHz range, and signal energy is more likely to be absorbed by
obstacles, it has been shown by researchers at Atheros Communications [13]”a pioneer
of IEE802.11a products”that the performance of IEEE802.11a is superior to the perfor-
mance of IEEE802.11b at distances less than 70 meters, by at least a factor of two (see
Figure 2.2). Due to this fact and due to the availability of eight channels, IEEE802.11a is
likely going to have a prosperous future. Equipment manufactured by some companies
extends the standard by introducing even higher-rate modes capable of transmitting with a
108 Mbps symbol rate. IEEE802.11g. Task Group g is working on an extension to IEEE802.11b at
2.4 GHz, enabling transmission at symbol rates of 54 Mbps while retaining the fall-back
speeds of IEEE802.11b, thus ensuring interoperability. After a long and rough debate,
Task Group g has agreed to the adoption of OFDM technology (while keeping DSS for
the interoperability mode); the standard is expected to be finalized at the end of 2002.
Although IEEE802.11g-based equipment will provide the same symbol rate as
IEEE802.11a, it will still have the same three-channel restriction of the original standard
as well as it will operate in the crowded 2.4 GHz range.
All of the previously outlined IEEE802.11-based technologies can be used and de-
ployed as the PHY and MAC layers of ad hoc networks. Other IEEE802.11 Task Groups
IEEE802.11h. There were strong European concerns that 802.11a could interfere with
NATO satellites and microwave radar systems. To avoid such interference, two extensions


Symbol Rate [Mbps]





0 7.5 15 22.5 30 37.5 45 52.5 60 67.5
Range (m)

Figure 2.2. Symbol rates of IEEE802.11b versus IEEE802.11a [13].

to the PHY of 802.11a were added in 802.11h, one of them being the capability to select
the employed channel automatically based upon observations (DFS”Dynamic Frequency
Selection), the other ensuring the enforcement of strict radio power control (TPC”Trans-
mit Power Control).
IEEE802.11e. Task Group e is addressing the flaw of IEEE802.11, working in a best-
effort mode but not being able to provide with any QoS provisioning. This Task Group is
redefining both the centrally controlled channel access as well as redefining the con-
tention-based channel access of CSMA/CA, including priorities to ensure that packets
with higher priorities enjoy access benefits comparable to lower-priority packets in a Dif-
ferentiated Services manner. This later function is called the Enhanced Distributed Coor-
dination Function (EDCF).
IEEE802.11c is a wireless extension to IEEE802.1D, enabling bridging using
IEEE802.11 (irrelevant to ad hoc networking).
IEEE802.11d deals with including country-specific information into the beacon trans-
missions, so STAs are informed of what part of the spectrum is available and what radio
constraints they have to obey to (e.g., maximum transmission power).
IEEE802.11f is defining a standard interaccess-point communication protocol for
users roaming between access points (irrelevant to ad hoc networks).
IEEE802.11i addresses the flaws of WEP, improving the wireless security at the MAC
layer. Further Reading. The reader interested in more high-level details is referred
to the 802.11-Planet [3], an online resource on IEEE802.11-related information and news.
Readers looking for a more detailed description can obtain the freely available IEEE 802
standards [1, 2] as a result of a new initiative of IEEE 802 to increase interoperability of
devices. For a brief online explanation of the OFDM principles, the reader is referred to
McCormick™s tutorial [26] or to the online white papers and materials of the OFDM Fo-
rum [34].

2.2.2 HiperLAN 1 and 2
HiperLAN [16] is the well-known name of the WLAN standardization efforts of the Euro-
pean Telecommunications Standards Institute (ETSI); more precisely, it is being devel-
oped by the BRAN (Broadband Radio Access Networks) project of ETSI. HiperLAN 2
[17] is the new version of the standard, providing more bandwidth and interoperability
considerations with third-generation wireless networks (e.g., Universal Mobile Telecom-
munication System or UMTS).
HiperLAN 1 is defined to work in the 5.2 GHz U-NII band, providing symbol rates of
up to 23.5 Mbps. Unfortunately, HiperLAN was not picked up by any companies to man-
ufacture products”it quickly became obsolete. ETSI-BRAN has proposed HiperLAN 2,
hoping for better acceptance.
The PHY layer of HiperLAN 2 is nearly identical to that of IEEE802.11h (which is a
European-initiated extension to IEEE802.11a), using OFDM as the basis. The main dif-
ference between HiperLAN2 and IEEE802.11a lies in the definition of the MAC layer. As
IEEE802.11a relies on a CSMA/CA-based channel access related to Ethernet, HiperLAN
2 is based on a TDMA approach, with scheduling principles taken from Wireless ATM.
HiperLAN 2 thus is able to provide QoS provisioning and can be used for guaranteed real-

time data delivery. The MAC layer of HiperLAN 2 defines both a centralized (infrastruc-
ture) mode and an ad hoc mode, similarly to IEEE802.11.
Since no company has yet manufactured inexpensive, commercially available Hiper-
LAN products, there are no ad hoc network testbeds based on HiperLAN. The 802.11
standards seem to be more widely accepted than HiperLAN 1 or 2, despite the advertised
superiority of HiperLAN 2. Just as with HiperLAN 1, there are no products currently
available in large quantities for HiperLAN 2 hindering its deployment as the basis for ad
hoc networks. Ad hoc routing protocols (and their simulation) relying on HiperLAN have
been proposed [14, 19] but not as widely as protocols relying on IEEE802.11 standards.
Optimized Link State Routing (OLSR) [14] is specifically tailored toward HiperLAN.
The reader interested in more details is referred to the standards [16, 17] or the excellent
white papers provided at the HiperLAN2 Global forum [24].

2.2.3 Infrared WLANs
Although not mentioned yet, the commercial history of WLANs began in 1979 with the
Diffused Infrared WLAN project of IBM in Switzerland. The main disadvantage of using
photonic electromagnetic waves is that light requires line-of-sight transmission”the re-
ceiver and transmitter have to be physically visible to each other. Although fixed environ-
ments can be engineered to abide by the line-of-sight rules, mobility can render an in-
frared WLAN useless. Omnidirectionality of transmissions is not achievable since light is
absorbed by most conventional obstacles (such as furniture, the computing unit itself, or
people). Due to these major disadvantages, infrared transmission has never taken off as a
WLAN competitor (e.g., the original 802.11 defines the operation on an infrared medium
as well). It is rarely even used for short-range wireless connections, despite the fact that
many portables are equipped with an IrDA (Infrared Data Association) port.
Using infrared transmission in ad hoc networks would defeat the purpose of the ad hoc
requirements”networks have to work in all kinds of (mostly hostile) environments. Yet
there are projects (such as [12, 22]) exploiting the inexpensive infrared technology for a
limited population of ad hoc nodes in indoor environments where the diffusion of the sig-
nal can be used as a benefit to somewhat overcome the problem of obstacles.

2.2.4 UWB
Ultra Wide Band (UWB) [39] is a novel spread-spectrum technique acknowledged by the
FCC in Spring 2002. UWB can be used for communication as well as to “see through
walls,” thus its commercial usage is strongly restricted by the FCC, making it a short-to-
medium range wireless communication technology. UWB does not use conventional fre-
quency carriers but generates very short duration rectangular pulses (close to that of Dirac
pulses), thus spreading the energy of the transmission over an extremely wide spectrum.
Due to this extreme spreading of the energy, UWB does not pose a significant interfering
source at any band, and it does not require line of sight.
The first UWB chips have just appeared on the market but it will take a tremendous
amount of additional research and standardization effort until UWB-based network
adapters become commercially available. UWB has all the properties needed to be the
next most popular PHY layer for ad hoc networks. The 802.15.3 Group is also considering
UWB as the basis for a high-speed WPAN standard.

2.2.5 Using IEEE802.11 for Ad Hoc Networking
As mentioned earlier, Wi-Fi is extremely popular among ad hoc network researchers as an
off-the-shelf support for their simulation or testbedding efforts. In this subsection, some
Wi-Fi-based simulation libraries and testbeds will be outlined.
Most major network-simulation toolkits have either an integrated or a contributed
IEEE802.11 library. The three most widely used simulators for ad hoc networks”NS2
[33], OPNET [35], and GloMoSim [18]”come with their own implementation of the
MAC and PHY layers of IEEE802.11. By far the most simulation efforts of ad hoc routing
protocols are carried out assuming (and employing) IEEE802.11-based MAC and PHY
layers of one of the above simulation tools.
Due to the availability of inexpensive Wi-Fi products that can be used to establish ad
hoc networks, it would be more of a challenge to list all projects that have established an
ad hoc network testbed than to list those universities and research labs that do not have
any. Here, some of the major projects are listed, starting with possibly the most well-
known public license testbed. Uppsala University in Sweden provides everybody the op-
portunity to build their own Wi-Fi-based ad hoc testbed by providing a GNU Public Li-
cense on their Ad Hoc Protocol Evaluation (APE) Testbed [5]. APE aims to make the
establishment of ad hoc testbeds as easy as possible while providing all the functions re-
quired for customization. Project MART (Mobile Ad Hoc Routing Testbed) [30] at the
Helsinki University of Technology is establishing a college-wide Wi-Fi-based ad hoc net-
work to evaluate different proposed ad hoc routing protocols.
The MONARCH Project [32] uses a Wi-Fi-enabled ad hoc testbed to evaluate the
Dynamic Source Routing (DSR) ad hoc routing approach proposed by them. They also
provide the functionality to connect the ad hoc network to a traditional IP network using
gateways. The MOMENT Lab at the University of California, Santa Barbara, has its
own Wi-Fi-based testbed [31], running on pocket PCs, laptops, and desktops, to evalu-
ate their proposed ad hoc routing protocol: AODV (Ad Hoc On Demand Distance
Routing). A project in the R&D Group of Acticom [4] is focused on an ad hoc routing
testbed to research multimedia-aware routing protocols for ad hoc networks. The testbed
is based on the Wi-Fi 2.4 GHz technology (to be extended to Wi-Fi 5.2 GHz), using
multimedia-enabled laptops and running video conferencing applications over their ad
hoc network. The Wireless Network Testbed (WNT) [42] at the University of Surrey,
United Kingdom, focuses on the evaluation of mobility management protocols, QoS
provisioning techniques, routing, and reconfigurability with their Wi-Fi-based ad hoc
network. Trinity College in Dublin, Ireland, envisions a Wi-Fi-based ad hoc network
covering the entire city of Dublin, using their DAWN (Dublin Ad Hoc Wireless
Network) testbed [15]. DAWN is not only envisioned as a testbed but also as the ad hoc
medium for fourth-generation (4G) wireless systems, and is fully operational on the
campus. Unfortunately, as pointed out in the next paragraph, Wi-Fi was not designed to
serve multihop networks, and the community has yet to produce an inexpensive ad hoc
tailored PHY and MAC standard.
An extensive analysis of the problems related to the use of IEE802.11 in ad hoc net-
works is presented in Chapter 3. Here, we would like to point out in advance that Wi-Fi
has not been developed for ad hoc networking and, thus, it can exhibit undesired behavior
when used for ad hoc networking. Although IEEE802.11 was developed keeping an ad
hoc mode in mind, this ad hoc mode is tailored toward simple point-to-point connections;
that is, to interconnect laptops for quick file transfers without the buffering and relaying

requirement of access points. A recent article [43] in the IEEE Communication Magazine
points out the shortcomings of the IEEE802.11 MAC layer in providing for ad hoc net-
works. In [43] the authors claim that the Wi-Fi MAC does not suit ad hoc networks well
and that Wi-Fi-based ad hoc testbeds will not perform properly and may cause significant


. 13
( 87 .)