Ne xt generation dynamic spectrum access cognitive2006survey

Computer Networks 50 (2006) 2127–2159
www.elsevier.com/locate/comnet

NeXt generation/dynamic spectrum access/cognitive
radio wireless networks: A survey
Ian F. Akyildiz, Won-Yeol Lee, Mehmet C. Vuran *, Shantidev Mohanty
Broadband and Wireless Networking Laboratory, School of Electrical and Computer Engineering,
Georgia Institute of Technology, Atlanta, GA 30332, United States

Received 2 January 2006; accepted 2 May 2006
Available online 17 May 2006

Responsible Editor: I.F. Akyildiz

Abstract

Today’s wireless networks are characterized by a fixed spectrum assignment policy. However, a large portion of the
assigned spectrum is used sporadically and geographical variations in the utilization of assigned spectrum ranges from
15% to 85% with a high variance in time. The limited available spectrum and the inefficiency in the spectrum usage neces-
sitate a new communication paradigm to exploit the existing wireless spectrum opportunistically. This new networking
paradigm is referred to as NeXt Generation (xG) Networks as well as Dynamic Spectrum Access (DSA) and cognitive
radio networks. The term xG networks is used throughout the paper. The novel functionalities and current research chal-
lenges of the xG networks are explained in detail. More specifically, a brief overview of the cognitive radio technology is
provided and the xG network architecture is introduced. Moreover, the xG network functions such as spectrum manage-
ment, spectrum mobility and spectrum sharing are explained in detail. The influence of these functions on the performance
of the upper layer protocols such as routing and transport are investigated and open research issues in these areas are also
outlined. Finally, the cross-layer design challenges in xG networks are discussed.
Ó 2006 Elsevier B.V. All rights reserved.

Keywords: Next generation networks; Dynamic spectrum access networks; Cognitive radio networks; Spectrum sensing; Spectrum
management; Spectrum mobility; Spectrum sharing

1. Introduction is regulated by governmental agencies and is
assigned to license holders or services on a long term
Today’s wireless networks are regulated by a basis for large geographical regions. In addition, a
fixed spectrum assignment policy, i.e. the spectrum large portion of the assigned spectrum is used spo-
radically as illustrated in Fig. 1, where the signal
*
strength distribution over a large portion of the
Corresponding author. Tel.: +1 404 894 5141; fax: +1 404 894 wireless spectrum is shown. The spectrum usage is
7883.
E-mail addresses: ian@ece.gatech.edu (I.F. Akyildiz), wylee@
concentrated on certain portions of the spectrum
ece.gatech.edu (W.-Y. Lee), mcvuran@ece.gatech.edu (M.C. while a significant amount of the spectrum remains
Vuran), shanti@ece.gatech.edu (S. Mohanty). unutilized. According to Federal Communications

1389-1286/$ - see front matter Ó 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.comnet.2006.05.001

2128 I.F. Akyildiz et al. / Computer Networks 50 (2006) 2127–2159

an opportunistic manner. Dynamic spectrum access
techniques allow the cognitive radio to operate in
the best available channel. More specifically, the cog-
nitive radio technology will enable the users to (1)
determine which portions of the spectrum is avail-
able and detect the presence of licensed users when
a user operates in a licensed band (spectrum sens-
ing), (2) select the best available channel (spectrum
management), (3) coordinate access to this channel
with other users (spectrum sharing), and (4) vacate
the channel when a licensed user is detected (spec-
trum mobility).
Fig. 1. Spectrum utilization. Once a cognitive radio supports the capability to
select the best available channel, the next challenge
is to make the network protocols adaptive to the
Commission (FCC) [20], temporal and geographical available spectrum. Hence, new functionalities are
variations in the utilization of the assigned spectrum required in an xG network to support this adaptivity.
range from 15% to 85%. Although the fixed spec- In summary, the main functions for cognitive radios
trum assignment policy generally served well in the in xG networks can be summarized as follows:
past, there is a dramatic increase in the access to
the limited spectrum for mobile services in the • Spectrum sensing: Detecting unused spectrum
recent years. This increase is straining the effective- and sharing the spectrum without harmful inter-
ness of the traditional spectrum policies. ference with other users.
The limited available spectrum and the inefficiency • Spectrum management: Capturing the best avail-
in the spectrum usage necessitate a new communi- able spectrum to meet user communication
cation paradigm to exploit the existing wireless spec- requirements.
trum opportunistically [3]. Dynamic spectrum access • Spectrum mobility: Maintaining seamless com-
is proposed to solve these current spectrum ineffi- munication requirements during the transition
ciency problems. DARPAs approach on Dynamic to better spectrum.
Spectrum Access network, the so-called NeXt Gener- • Spectrum sharing: Providing the fair spectrum
ation (xG) program aims to implement the policy scheduling method among coexisting xG users.
based intelligent radios known as cognitive radios
[67,68]. These functionalities of xG networks enable spec-
NeXt Generation (xG) communication net- trum-aware communication protocols. However,
works, also known as Dynamic Spectrum Access the dynamic use of the spectrum causes adverse
Networks (DSANs) as well as cognitive radio net- effects on the performance of conventional commu-
works, will provide high bandwidth to mobile users nication protocols, which were developed consider-
via heterogeneous wireless architectures and ing a fixed frequency band for communication. So
dynamic spectrum access techniques. The inefficient far, networking in xG networks is an unexplored
usage of the existing spectrum can be improved topic. In this paper, we also capture the intrinsic
through opportunistic access to the licensed bands challenges for networking in xG networks and lay
without interfering with the existing users. xG net- out guidelines for further research in this area. More
works, however, impose several research challenges specifically, we overview the recent proposals for
due to the broad range of available spectrum as well spectrum sharing and routing in xG networks as
as diverse Quality-of-Service (QoS) requirements of well as the challenges for transport protocols. More-
applications. These heterogeneities must be cap- over, the effect of cross-layer design is addressed for
tured and handled dynamically as mobile terminals communication in xG networks.
roam between wireless architectures and along the The xG network communication components
available spectrum pool. and their interactions are illustrated in Fig. 2. It is
The key enabling technology of xG networks is evident from the significant number of interactions
the cognitive radio. Cognitive radio techniques pro- that the xG network functionalities necessitate a
vide the capability to use or share the spectrum in cross-layer design approach. More specifically, spec-

I.F. Akyildiz et al. / Computer Networks 50 (2006) 2127–2159 2129

Fig. 2. xG network communication functionalities.

trum sensing and spectrum sharing cooperate with A ‘‘Cognitive Radio’’ is a radio that can change its
each other to enhance spectrum efficiency. In spec- transmitter parameters based on interaction with
trum management and spectrum mobility functions, the environment in which it operates.
application, transport, routing, medium access and
From this definition, two main characteristics of
physical layer functionalities are carried out in a
the cognitive radio can be defined [27,58]:
cooperative way, considering the dynamic nature
of the underlying spectrum.
• Cognitive capability: Cognitive capability refers
This paper presents a definition, functions and
to the ability of the radio technology to capture
current research challenges of the xG networks. In
or sense the information from its radio environ-
Section 2, we provide a brief overview of the cognitive
ment. This capability cannot simply be realized
radio technology. The xG network architectures on
by monitoring the power in some frequency band
licensed band and on unlicensed band are presented
of interest but more sophisticated techniques are
in Section 3. In Section 4, we explain the existing
required in order to capture the temporal and
work and challenges in spectrum sensing. Then, we
spatial variations in the radio environment and
describe the xG network functionalities: spectrum
avoid interference to other users. Through this
management, spectrum mobility and spectrum shar-
capability, the portions of the spectrum that
ing in Sections 5, 6, and 7, respectively. In Section 8,
are unused at a specific time or location can be
we investigate how xG features influence the perfor-
identified. Consequently, the best spectrum
mance of the upper layer protocols, i.e., routing
and appropriate operating parameters can be
and transport. Finally, we explain how xG functions
selected.
can be implemented in a cross-layer approach in Sec-
• Reconfigurability: The cognitive capability pro-
tion 9 and conclude the paper in Section 10.
vides spectrum awareness whereas reconfigu-
rability enables the radio to be dynamically
2. Cognitive radio programmed according to the radio environ-
ment. More specifically, the cognitive radio can
Cognitive radio technology is the key technology be programmed to transmit and receive on a
that enables an xG network to use spectrum in a variety of frequencies and to use different trans-
dynamic manner. The term, cognitive radio, can mission access technologies supported by its
formally be defined as follows [20]: hardware design [34].


The cognitive radio concept was first introduced before. Since most of the spectrum is already
in [45,46], where the main focus was on the radio assigned, the most important challenge is to share
knowledge representation language (RKRL) and the licensed spectrum without interfering with the
how the cognitive radio can enhance the flexibility transmission of other licensed users as illustrated
of personal wireless services. The cognitive radio is in Fig. 3. The cognitive radio enables the usage of
regarded as a small part of the physical world to temporally unused spectrum, which is referred to
use and provide information from environment. as spectrum hole or white space [27]. If this band is
The ultimate objective of the cognitive radio is to further used by a licensed user, the cognitive radio
obtain the best available spectrum through cogni- moves to another spectrum hole or stays in the same
tive capability and reconfigurability as described band, altering its transmission power level or mod-
ulation scheme to avoid interference as shown in
Fig. 3.
In the following subsections, we describe the
physical architecture, cognitive functions and recon-
figurability capabilities of the cognitive radio
technology.

2.1. Physical architecture of the cognitive radio

A generic architecture of a cognitive radio trans-
ceiver is shown in Fig. 4(a) [34]. The main compo-
nents of a cognitive radio transceiver are the radio
front-end and the baseband processing unit. Each
Fig. 3. Spectrum hole concept. component can be reconfigured via a control bus

Fig. 4. Physical architecture of the cognitive radio [12,34]: (a) Cognitive radio transceiver and (b) wideband RF/analog front-end
architecture.


to adapt to the time-varying RF environment. In In this architecture, a wideband signal is received
the RF front-end, the received signal is amplified, through the RF front-end, sampled by the high
mixed and A/D converted. In the baseband process- speed analog-to-digital (A/D) converter, and mea-
ing unit, the signal is modulated/demodulated and surements are performed for the detection of the
encoded/decoded. The baseband processing unit of licensed user signal. However, there exist some limi-
a cognitive radio is essentially similar to existing tations on developing the cognitive radio front-end.
transceivers. However, the novelty of the cognitive The wideband RF antenna receives signals from var-
radio is the RF front-end. Hence, next, we focus ious transmitters operating at different power levels,
on the RF front-end of the cognitive radios. bandwidths, and locations. As a result, the RF front-
The novel characteristic of cognitive radio trans- end should have the capability to detect a weak sig-
ceiver is a wideband sensing capability of the RF nal in a large dynamic range. However, this capabil-
front-end. This function is mainly related to RF ity requires a multi-GHz speed A/D converter with
hardware technologies such as wideband antenna, high resolution, which might be infeasible [12,13].
power amplifier, and adaptive filter. RF hardware The requirement of a multi-GHz speed A/D con-
for the cognitive radio should be capable of tuning verter necessitates the dynamic range of the signal to
to any part of a large range of frequency spectrum. be reduced before A/D conversion. This reduction
Also such spectrum sensing enables real-time can be achieved by filtering strong signals. Since
measurements of spectrum information from radio strong signals can be located anywhere in the wide
environment. Generally, a wideband front-end archi- spectrum range, tunable notch filters are required
tecture for the cognitive radio has the following struc- for the reduction [12]. Another approach is to use
ture as shown in Fig. 4(b) [12]. The components of a multiple antennas such that signal filtering is per-
cognitive radio RF front-end are as follows: formed in the spatial domain rather than in the fre-
quency domain. Multiple antennas can receive
• RF filter: The RF filter selects the desired band signals selectively using beamforming techniques
by bandpass filtering the received RF signal. [13].
• Low noise amplifier (LNA): The LNA amplifies As explained previously, the key challenge of the
the desired signal while simultaneously minimiz- physical architecture of the cognitive radio is an
ing noise component. accurate detection of weak signals of licensed users
• Mixer: In the mixer, the received signal is mixed over a wide spectrum range. Hence, the implementa-
with locally generated RF frequency and con- tion of RF wideband front-end and A/D converter
verted to the baseband or the intermediate fre- are critical issues in xG networks.
quency (IF).
• Voltage-controlled oscillator (VCO): The VCO 2.2. Cognitive capability
generates a signal at a specific frequency for a
given voltage to mix with the incoming signal. The cognitive capability of a cognitive radio
This procedure converts the incoming signal to enables real time interaction with its environment
baseband or an intermediate frequency. to determine appropriate communication parame-
• Phase locked loop (PLL): The PLL ensures that ters and adapt to the dynamic radio environment.
a signal is locked on a specific frequency and can The tasks required for adaptive operation in open
also be used to generate precise frequencies with spectrum are shown in Fig. 5 [27,46,58], which is
fine resolution. referred to as the cognitive cycle. In this section,
• Channel selection filter: The channel selection fil- we provide an overview of the three main steps of
ter is used to select the desired channel and to the cognitive cycle: spectrum sensing, spectrum anal-
reject the adjacent channels. There are two types ysis, and spectrum decision. The details and the
of channel selection filters [52]. The direct conver- related work of these functions are described in Sec-
sion receiver uses a low-pass filter for the channel tions 4 and 5.
selection. On the other hand, the superheterodyne The steps of the cognitive cycle as shown in Fig. 5
receiver adopts a bandpass filter. are as follows:
• Automatic gain control (AGC): The AGC main-
tains the gain or output power level of an ampli- 1. Spectrum sensing: A cognitive radio monitors the
fier constant over a wide range of input signal available spectrum bands, captures their infor-
levels. mation, and then detects the spectrum holes.


the most suitable operating frequency can be
determined and the communication can be
dynamically performed on this appropriate oper-
ating frequency.
• Modulation: A cognitive radio should reconfigure
the modulation scheme adaptive to the user
requirements and channel conditions. For exam-
ple, in the case of delay sensitive applications, the
data rate is more important than the error rate.
Thus, the modulation scheme that enables the
higher spectral efficiency should be selected. Con-
versely, the loss-sensitive applications focus on
the error rate, which necessitate modulation
schemes with low bit error rate.
Fig. 5. Cognitive cycle.
• Transmission power: Transmission power can be
reconfigured within the power constraints. Power
2. Spectrum analysis: The characteristics of the control enables dynamic transmission power con-
spectrum holes that are detected through spec- figuration within the permissible power limit. If
trum sensing are estimated. higher power operation is not necessary, the cog-
3. Spectrum decision: A cognitive radio determines nitive radio reduces the transmitter power to a
the data rate, the transmission mode, and the lower level to allow more users to share the spec-
bandwidth of the transmission. Then, the appro- trum and to decrease the interference.
priate spectrum band is chosen according to the • Communication technology: A cognitive radio can
spectrum characteristics and user requirements. also be used to provide interoperability among
different communication systems.
Once the operating spectrum band is determined,
the communication can be performed over this spec- The transmission parameters of a cognitive radio
trum band. However, since the radio environment can be reconfigured not only at the beginning of a
changes over time and space, the cognitive radio transmission but also during the transmission.
should keep track of the changes of the radio envi- According to the spectrum characteristics, these
ronment. If the current spectrum band in use parameters can be reconfigured such that the
becomes unavailable, the spectrum mobility function cognitive radio is switched to a different spectrum
that will be explained in Section 6, is performed to band, the transmitter and receiver parameters are
provide a seamless transmission. Any environmen- reconfigured and the appropriate communication
tal change during the transmission such as primary protocol parameters and modulation schemes are
user appearance, user movement, or traffic variation used.
can trigger this adjustment.
3. The xG network architecture
2.3. Reconfigurability
Existing wireless network architectures employ
Reconfigurability is the capability of adjusting heterogeneity in terms of both spectrum policies
operating parameters for the transmission on the and communication technologies [3]. Moreover,
fly without any modifications on the hardware com- some portion of the wireless spectrum is already
ponents. This capability enables the cognitive radio licensed to different purposes while some bands
to adapt easily to the dynamic radio environment. remain unlicensed. For the development of commu-
There are several reconfigurable parameters that nication protocols, a clear description of the xG net-
can be incorporated into the cognitive radio [20] work architecture is essential. In this section, the xG
as explained below: network architecture is presented such that all pos-
sible scenarios are considered.
• Operating frequency: A cognitive radio is capable The components of the xG network architecture,
of changing the operating frequency. Based on as shown in Fig. 6, can be classified in two groups as
the information about the radio environment, the primary network and the xG network. The basic


Fig. 6. xG network architecture.

elements of the primary and the xG network are cols for the primary network access of xG
defined as follows: users, which is explained below.
• xG network: xG network (or cognitive radio net-
• Primary network: An existing network infrastruc- work, Dynamic Spectrum Access network, sec-
ture is generally referred to as the primary net- ondary network, unlicensed network) does not
work, which has an exclusive right to a certain have license to operate in a desired band. Hence,
spectrum band. Examples include the common the spectrum access is allowed only in an oppor-
cellular and TV broadcast networks. The compo- tunistic manner. xG networks can be deployed
nents of the primary network are as follows: both as an infrastructure network and an ad
– Primary user: Primary user (or licensed user) hoc network as shown in Fig. 6. The components
has a license to operate in a certain spectrum of an xG network are as follows:
band. This access can only be controlled by – xG user: xG user (or unlicensed user, cognitive
the primary base-station and should not be radio user, secondary user) has no spectrum
affected by the operations of any other unli- license. Hence, additional functionalities are
censed users. Primary users do not need any required to share the licensed spectrum band.
modification or additional functions for coex- – xG base-station: xG base-station (or unli-
istence with xG base-stations and xG users. censed base-station, secondary base-station)
– Primary base-station: Primary base-station (or is a fixed infrastructure component with xG
licensed base-station) is a fixed infrastructure capabilities. xG base-station provides single
network component which has a spectrum hop connection to xG users without spectrum
license such as base-station transceiver system access license. Through this connection, an xG
(BTS) in a cellular system. In principle, the user can access other networks.
primary base-station does not have any xG – Spectrum broker: Spectrum broker (or sched-
capability for sharing spectrum with xG users. uling server) is a central network entity that
However, the primary base-station may be plays a role in sharing the spectrum resources
requested to have both legacy and xG proto- among different xG networks. Spectrum


broker can be connected to each network and
can serve as a spectrum information manager
to enable coexistence of multiple xG networks
[10,32,70].

The reference xG network architecture is shown
in Fig. 6, which consists of different types of net-
works: a primary network, an infrastructure based
xG network, and an ad-hoc xG network. xG net-
works are operated under the mixed spectrum envi-
ronment that consists of both licensed and Fig. 7. xG network on licensed band.
unlicensed bands. Also, xG users can either commu-
nicate with each other in a multihop manner or
access the base-station. Thus, in xG networks, there Fig. 7, where the xG network coexists with the pri-
are three different access types as explained next: mary network at the same location and on the same
spectrum band.
• xG network access: xG users can access their own There are various challenges for xG networks on
xG base-station both on licensed and unlicensed licensed band due to the existence of the primary
spectrum bands. users. Although the main purpose of the xG net-
• xG ad hoc access: xG users can communicate work is to determine the best available spectrum,
with other xG users through ad hoc connection xG functions in the licensed band are mainly aimed
on both licensed and unlicensed spectrum bands. at the detection of the presence of primary users.
• Primary network access: The xG users can also The channel capacity of the spectrum holes depends
access the primary base-station through the on the interference at the nearby primary users.
licensed band. Thus, the interference avoidance with primary users
is the most important issue in this architecture.
According to the reference architecture shown in Furthermore, if primary users appear in the spec-
Fig. 6, various functionalities are required to sup- trum band occupied by xG users, xG users should
port the heterogeneity in xG networks. In Section vacate the current spectrum band and move to the
3.1, we describe the xG network functions to sup- new available spectrum immediately, called spec-
port the heterogeneity of the network environment. trum handoff.
Moreover, in Sections 3.2 and 3.3, we overview xG
network applications and existing architectures. 3.1.2. xG network on unlicensed band
Open spectrum policy that began in the industrial
3.1. xG network functions scientific and medical (ISM) band has caused an
impressive variety of important technologies and
As explained before, xG network can operate in innovative uses. However, due to the interference
both licensed and unlicensed bands. Hence, the func- among multiple heterogeneous networks, the spec-
tionalities required for xG networks vary according trum efficiency of ISM band is decreasing. Ulti-
to whether the spectrum is licensed or unlicensed. mately, the capacity of open spectrum access, and
Accordingly, in this section, we classify the xG net- the quality of service they can offer, depend on the
work operations as xG network on licensed band degree to which a radio can be designed to allocate
and xG network on unlicensed band. The xG network spectrum efficiently.
functions are explained in the following sections xG networks can be designed for operation on
according to this classification. unlicensed bands such that the efficiency is
improved in this portion of the spectrum. The xG
3.1.1. xG network on licensed band network on unlicensed band architecture is illustrated
As shown in Fig. 1, there exist temporally unused in Fig. 8. Since there are no license holders, all
spectrum holes in the licensed spectrum band. network entities have the same right to access the
Hence, xG networks can be deployed to exploit spectrum bands. Multiple xG networks coexist in
these spectrum holes through cognitive communi- the same area and communicate using the same por-
cation techniques. This architecture is depicted in tion of the spectrum. Intelligent spectrum sharing


will be deployed in dense urban areas with the pos-
sibility of significant contention [38]. For example,
the coverage area of xG networks can be increased
when a meshed wireless backbone network of infra-
structure links is established based on cognitive
access points (CAPs) and fixed cognitive relay nodes
(CRNs) [6]. The capacity of a CAP, connected via a
wired broadband access to the Internet, is distrib-
uted into a large area with the help of a fixed
CRN. xG networks have the ability to add tempo-
rary or permanent spectrum to the infrastructure
Fig. 8. xG network on unlicensed band. links used for relaying in case of high traffic load.
Emergency network: Public safety and emergency
networks are another area in which xG networks can
algorithms can improve the efficiency of spectrum be implemented [41]. In the case of natural disasters,
usage and support high QoS. which may temporarily disable or destroy existing
In this architecture, xG users focus on detecting communication infrastructure, emergency personnel
the transmissions of other xG users. Unlike the working in the disaster areas need to establish emer-
licensed band operations, the spectrum handoff is gency networks. Since emergency networks deal with
not triggered by the appearance of other primary the critical information, reliable communication
users. However, since all xG users have the same right should be guaranteed with minimum latency. In
to access the spectrum, xG users should compete with addition, emergency communication requires a sig-
each other for the same unlicensed band. Thus, nificant amount of radio spectrum for handling huge
sophisticated spectrum sharing methods among xG volume of traffic including voice, video and data. xG
users are required in this architecture. If multiple networks can enable the usage of the existing spec-
xG network operators reside in the same unlicensed trum without the need for an infrastructure and by
band, fair spectrum sharing among these networks maintaining communication priority and response
is also required. time.
Military network: One of the most interesting
3.2. xG network applications potential applications of an xG network is in a mil-
itary radio environment [47]. xG networks can
xG networks can be applied to the following enable the military radios choose arbitrary, interme-
cases: diate frequency (IF) bandwidth, modulation
Leased network: The primary network can pro- schemes, and coding schemes, adapting to the vari-
vide a leased network by allowing opportunistic able radio environment of battlefield. Also military
access to its licensed spectrum with the agreement networks have a strong need for security and pro-
with a third party without sacrificing the service tection of the communication in hostile environ-
quality of the primary user [56]. For example, the ment. xG networks could allow military personnel
primary network can lease its spectrum access right to perform spectrum handoff to find secure spec-
to a mobile virtual network operator (MVNO). trum band for themselves and their allies.
Also the primary network can provide its spectrum
access rights to a regional community for the pur- 3.3. Existing architectures
pose of broadband access.
Cognitive mesh network: Wireless mesh networks The main representative examples of the xG net-
are emerging as a cost-effective technology for pro- work architectures are described in this section.
viding broadband connectivity [4]. However, as the Spectrum pooling: In [61,62], a centralized
network density increases and the applications spectrum pooling architecture is proposed based
require higher throughput, mesh networks require on orthogonal frequency division multiplexing
higher capacity to meet the requirements of the (OFDM). This architecture consists of an xG
applications. Since the cognitive radio technology base-station and mobile xG users. OFDM has the
enables the access to larger amount of spectrum, advantage of feeding certain sub-carriers with zeros
xG networks can be used for mesh networks that resulting in no emission of radio power on the


carriers that are occupied by the licensed users. The is managed by the base-station, which instructs the
licensed user detection is performed through detec- various CPEs to perform distributed measurement
tion frames that are periodically broadcast by the activities. The IEEE 802.22 system specifies spectral
base-station. During the detection frames, the efficiencies in the range of 0.5–5 bit/s/Hz. A distinc-
mobile users perform spectrum sensing. The sensing tive feature of IEEE 802.22 WRAN as compared to
information is then gathered at the base-station. the existing IEEE 802 standards is the base-station
Mobile terminals modulate a complex symbol at coverage range, which can go up to 100 km if the
maximum power in the sub-carriers where a licensed power is not an issue. Current specified coverage
user appears. Through this operation, the base- range is 33 km at 4 W CPE effective isotropic radi-
station receives an amplified signal on all sub-carri- ated power (EIRP) [66]. Table 1 depicts the capacity
ers with new licensed users. Physical and MAC layer and coverage of IEEE 802.22 WRAN system. IEEE
issues, such as detection of spectral access, schedul- 802.22 working group was formed in 2004 and has
ing, and handoff are ongoing investigations in this finalized the specification of technical requirements.
architecture. The first draft of IEEE 802.22 standard will be
CORVUS: In [8,14], a cognitive radio approach ready around mid 2006.
for usage of virtual unlicensed spectrum (CORVUS) DIMSUMnet: The dynamic intelligent manage-
system is presented to exploit unoccupied licensed ment of spectrum for ubiquitous mobile network
bands. In CORVUS, based on the local spectrum (DIMSUMnet) [10] implements statistically multi-
sensing, the primary user detection and the spec-
trum allocation are performed in a coordinated
manner. This cooperative effort greatly increases
the system’s ability in identifying and avoiding pri-
Table 1
mary users. In CORVUS, a group of users form a IEEE 802.22 WRAN system capacity and coverage [65]
secondary user group (SUG) to coordinate their
RF channel bandwidth 6 MHz
communication. Each member of this group senses Average spectrum efficiency 3 bit/s/Hz
the spectrum pool, which is divided into sub-chan- Channel capacity 18 Mbit/s
nels. A universal control channel is used by all System capacity per subscriber 1.5 Mbit/s
groups for coordination and separate group control (forward)
channels are used by the members of a group to System capacity per subscriber 384 kbit/s
(return)
exchange sensing information and establish second- Forward/return ratio 3.9
ary user links. The performance of the physical and Over-subscription ratio 50
link layers are evaluated through the CORVUS test- Number of subscribers per 600
bed [44]. Moreover, recently, a reliable link mainte- forward channel
nance protocol is proposed within CORVUS to (FDD operation is assumed to
maximize system capacity for large
maintain the quality of secondary user communica- coverage distances, TDD would reduce
tion [64]. capacity to 75% per TV channel)
IEEE 802.22: IEEE 802.22 is the first worldwide
Minimum number of subscribers 90 subs.
standard based on the cognitive radio technology Assumed early take-up rate 3 bit/s/Hz
[16,31] and is now in the process of standardization. Potential number of subscribers 1800 subs.
This project, formally called the standard for wire- Assumed number of persons 2.5 persons
less regional area networks (WRAN), focuses on per household
Total number of persons 4500 persons
constructing fixed point-to-multipoint WRAN that
per coverage area
will utilize UHF/VHF TV bands between 54 and WRAN base station EIRP 98.3 W
862 MHz. Specific TV channels as well as guard Radius of coverage for WRAN system 30.7 km
bands will be used for communication in IEEE Minimum population density covered 1.5 person/km2
802.22. The IEEE 802.22 system specifies a fixed Assuming 1.5 Mbit/s forward and 384 kbit/s return in a 6 MHz
point-to-multipoint wireless air interface whereby channel with 3 bit/s/Hz and 50:1 over-subscription, each TV
a base-station manages its own cell and all associ- channel can provide service to up to 600 subscribers. The area
ated users, which are denoted as consumer premise that the Wireless Regional Area network (WRAN) operator
needs to cover should include enough subscribers to make it
equipments (CPEs). IEEE 802.22 base-station man- economically viable early in the process when the take-up rate is
ages a unique feature of distributed sensing. This is low. This will define the potential subscriber base and the pop-
needed to ensure proper incumbent protection and ulation density for a sustainable business case.


plexed access (SMA) to spectrum in the coordinated for temporal and spatial DSA is still a major
access band (CAB). While the CAB improves the challenge.
spectrum access efficiency and fairness, the SMA Nautilus: Nautilus project is designed to empha-
is focused on improving the spectrum utilization. size distributed coordination enabled spectrum shar-
CAB is a contiguous chunk of spectrum reserved ing, without relying on centralized control [48]. In
by regulating authorities. A spectrum broker the Nautilus project, a distributed, scalable and effi-
permanently owns the CAB and leases it according cient coordination framework for open spectrum ad
to requests. DIMSUMnet uses a centralized, regio- hoc networks is proposed, which accounts for spec-
nal network level brokering mechanism that aims trum heterogeneity while not relying on the existence
to significantly improve spectrum utilization while of a pre-defined common channel for control traffic
reducing the complexity and the agility require- [73,74]. Based on this framework, three different col-
ments of the deployed system. The base-station reg- laborative spectrum access schemes are presented. In
isters with its designated radio access network [73], a graph coloring based collaborative spectrum
manager (RANMAN), which negotiates a lease access scheme is proposed, where a topology-opti-
with a spectrum information and management mized allocation algorithm is used for the fixed
(SPIM) broker for an appropriate portion of the topology. In mobile networks, however, the network
spectrum. If the lease is successfully obtained, the topology changes due to the node mobility. Using
RANMAN configures the leased spectrum in this global optimization approach, the network
the base-station. The base-station sends the spec- needs to completely recompute spectrum assign-
trum information received from the RANMAN to ments for all users after each change, resulting in
its users for the configuration of client. The spec- high computational and communication overhead.
trum utilization of DIMSUMnet is currently mea- Thus, distributed spectrum allocation based on local
sured in existing Code Division Multiple Access bargaining is proposed in [15], where mobile users
(CDMA) and Global System for Mobile communi- negotiate spectrum assignment within local self-
cation (GSM) cellular networks, aimed at character- organized groups. For the resource constrained net-
izing feasibility of CAB and SMA [35]. Recent work works such as sensor and ad hoc network, rule-based
focuses on the spectrum pricing and allocation func- device centric spectrum management is proposed,
tions for spectrum brokers [11]. where unlicensed users access the spectrum indepen-
DRiVE/OverDRiVE project: The European dently according to both local observation and pre-
Dynamic Radio for IP Services in Vehicular Envi- determined rules. Currently, this project focuses on
ronments (DRiVE) project focuses on dynamic selecting the best channel for data transmission
spectrum allocation in heterogeneous networks by using proposed distributed coordination framework.
assuming a common coordinated channel [75]. The OCRA network: In [5], an OFDM-based cogni-
follow-up project, Spectrum Efficient Uni- and Mul- tive radio (OCRA) network is proposed. OCRA
ticast Over Dynamic Radio Networks in Vehicular network considers all possible deployment scenarios
Environments (OverDRiVE) aims at UMTS over the heterogeneous xG network environment
enhancements and coordination of existing radio and develops cross-layer operations for the OFDM
networks into a hybrid network to ensure spectrum based dynamic spectrum access. OCRA network
efficient provision of mobile multimedia services architecture and its components are shown in
[26]. Two aspects of dynamic spectrum allocation Fig. 6. For the spectrum decision and the spectrum
were investigated in DRiVE/OverDRiVE, i.e., handoff, OCRA network provides a novel concept
temporal dynamic spectrum allocation (DSA), and of an OFDM-based spectrum management over
spatial DSA [39]. In the case of the temporal the heterogeneous spectrum environment. Based
DSA, a radio access network (RAN) can use the on this physical layer (PHY) structure, a dual-mode
spectrum that is not currently being used by other spectrum sharing framework is proposed, which
RANs, at that time. On the other hand, spatial enables access to existing networks as well as coor-
DSA allows spectrum allocations to adapt to regio- dination between xG users. Furthermore, a new
nal fluctuations in traffic demands. The efficiency of routing paradigm that considers joint re-routing
these DSA schemes depends on the ability to predict and spectrum handoff is proposed. Moreover,
the traffic load. Although these projects have shown OCRA network introduces multi-spectrum trans-
significant potential for increasing spectral effi- port techniques to exploit the available but
ciency, the reconfigurable system implementation non-contiguous wireless spectrum for high quality


communication. In [5], the testbed design for evalu- from primary transmitter is locally present in a cer-
ation and integration of the OCRA network is pro- tain spectrum. Transmitter detection approach is
posed. The OCRA testbed is based on IEEE based on the detection of the weak signal from a pri-
802.11a/g technology, which exploits the OFDM mary transmitter through the local observations of
technology. Moreover, a standalone cognitive sens- xG users. Basic hypothesis model for transmitter
ing unit is developed to emulate the spectrum sens- detection can be defined as follows [25]:
ing capabilities of the cognitive radio. (
nðtÞ H 0;
xðtÞ ¼ ð1Þ
4. Spectrum sensing hsðtÞ þ nðtÞ H 1 ;

An important requirement of the xG network is where x(t) is the signal received by the xG user, s(t)
to sense the spectrum holes. As explained in Section is the transmitted signal of the primary user, n(t) is
2, a cognitive radio is designed to be aware of and the AWGN and h is the amplitude gain of the chan-
sensitive to the changes in its surrounding. The spec- nel. H0 is a null hypothesis, which states that there is
trum sensing function enables the cognitive radio to no licensed user signal in a certain spectrum band.
adapt to its environment by detecting spectrum On the other hand, H1 is an alternative hypothesis,
holes. which indicates that there exist some licensed user
The most efficient way to detect spectrum holes is signal.
to detect the primary users that are receiving data Three schemes are generally used for the trans-
within the communication range of an xG user. In mitter detection according to the hypothesis model
reality, however, it is difficult for a cognitive radio [53]. In the following subsections, we investigate
to have a direct measurement of a channel between matched filter detection, energy detection and cyclo-
a primary receiver and a transmitter. Thus, the most stationary feature detection techniques proposed
recent work focuses on primary transmitter detec- for transmitter detection in xG networks.
tion based on local observations of xG users.
Generally, the spectrum sensing techniques can 4.1.1. Matched filter detection
be classified as transmitter detection, coopera- When the information of the primary user signal
tive detection, and interference-based detection, as is known to the xG user, the optimal detector in sta-
shown in Fig. 9. In the following sections, we tionary Gaussian noise is the matched filter since it
describe these spectrum sensing methods for xG net- maximizes the received signal-to-noise ratio (SNR)
works and discuss the open research topics in this [53]. While the main advantage of the matched filter
area. is that it requires less time to achieve high process-
ing gain due to coherency, it requires a priori know-
4.1. Transmitter detection (non-cooperative ledge of the primary user signal such as the
detection) modulation type and order, the pulse shape, and
the packet format. Hence, if this information is
The cognitive radio should distinguish between not accurate, then the matched filter performs
used and unused spectrum bands. Thus, the cognitive poorly. However, since most wireless network sys-
radio should have capability to determine if a signal tems have pilot, preambles, synchronization word
or spreading codes, these can be used for the coher-
ent detection.

4.1.2. Energy detection
If the receiver cannot gather sufficient informa-
tion about the primary user signal, for example, if
the power of the random Gaussian noise is only
known to the receiver, the optimal detector is an
energy detector [53]. In order to measure the energy
of the received signal, the output signal of bandpass
filter with bandwidth W is squared and integrated
over the observation interval T. Finally, the output
Fig. 9. Classification of spectrum sensing techniques. of the integrator, Y, is compared with a threshold, k,


to decide whether a licensed user is present or not ces, or cyclic prefixes, which result in built-in period-
[17]. icity. These modulated signals are characterized as
If the energy detection can be applied in a non- cyclostationarity since their mean and autocorrela-
fading environment where h is the amplitude gain tion exhibit periodicity. These features are detected
of the channel as shown in (1), the probability of by analyzing a spectral correlation function. The
detection Pd and false alarm Pf are given as follows main advantage of the spectral correlation func-
[17]: tion is that it differentiates the noise energy from
pffiffiffiffiffi pffiffiffi modulated signal energy, which is a result of the fact
P d ¼ P fY kjH 1 g ¼ Qm 2c; k ; ð2Þ that the noise is a wide-sense stationary signal with
Cðm; k=2Þ no correlation, while modulated signals are cyclosta-
P f ¼ P fY kjH o g ¼ ; ð3Þ tionary with spectral correlation due to the embed-
CðmÞ
ded redundancy of signal periodicity. Therefore, a
where c is the SNR, u = TW is the time bandwidth cyclostationary feature detector can perform better
product, C(Æ) and C(Æ, Æ) are complete and incomplete than the energy detector in discriminating against
gamma functions and Qm( ) is the generalized Mar- noise due to its robustness to the uncertainty in noise
cum Q-function. From the above functions, while power [57]. However, it is computationally complex
a low Pd would result in missing the presence of and requires significantly long observation time.
the primary user with high probability which in turn For more efficient and reliable performance, the
increases the interference to the primary user, a high enhanced feature detection scheme combining cyclic
Pf would result in low spectrum utilization since spectral analysis with pattern recognition based on
false alarms increase the number of missed opportu- neural networks is proposed in [22]. Distinct fea-
nities. Since it is easy to implement, the recent work tures of the received signal are extracted using cyclic
on detection of the primary user has generally spectral analysis and represented by both spectral
adopted the energy detector [23,53]. coherent function and spectral correlation density
In [25], the shadowing and the multi-path fading function. The neural network, then, classifies signals
factors are considered for the energy detector. In into different modulation types.
this case, while Pf is independent of C, when the
amplitude gain of the channel, h, varies due to 4.2. Cooperative detection
the shadowing/fading, Pd gives the probability of
the detection conditioned on instantaneous SNR The assumption of the primary transmitter detec-
as follows: tion is that the locations of the primary receivers are
Z pffiffiffiffiffi pffiffiffi unknown due to the absence of signalling between
Pd ¼ Qm 2c; k fc ðxÞ dx; ð4Þ primary users and the xG users. Therefore, the cog-
x
nitive radio should rely on only weak primary trans-
where fc(x) is the probability distribution function mitter signals based on the local observation of the
of SNR under fading. xG user [72,73]. However, in most cases, an xG
However, the performance of energy detector is network is physically separated from the primary
susceptible to uncertainty in noise power. In order network so there is no interaction between them.
to solve this problem, a pilot tone from the pri- Thus, with the transmitter detection, the xG user
mary transmitter is used to help improve the cannot avoid the interference due to the lack of
accuracy of the energy detector in [53]. Another the primary receiver’s information as depicted in
shortcoming is that the energy detector cannot dif- Fig. 10(a). Moreover, the transmitter detection
ferentiate signal types but can only determine the model cannot prevent the hidden terminal problem.
presence of the signal. Thus, the energy detector is An xG transmitter can have a good line-of-sight to a
prone to the false detection triggered by the unin- receiver, but may not be able to detect the transmit-
tended signals. ter due to the shadowing as shown in Fig. 10(b).
Consequently, the sensing information from other
4.1.3. Cyclostationary feature detection users is required for more accurate detection.
An alternative detection method is the cyclosta- In the case of non-cooperative detection ex-
tionary feature detection [12,22,57]. Modulated sig- plained in Section 4.1, the xG users detect the pri-
nals are in general coupled with sine wave carriers, mary transmitter signal independently through their
pulse trains, repeating spreading, hopping sequen- local observations. Cooperative detection refers to


Fig. 10. Transmitter detection problem: (a) Receiver uncertainty and (b) shadowing uncertainty.

spectrum sensing methods where information from While cooperative approaches provide more
multiple xG users are incorporated for primary user accurate sensing performance, they cause adverse
detection. Cooperative detection can be imple- effects on resource-constrained networks due to
mented either in a centralized or in a distributed the additional operations and overhead traffic.
manner [23,71]. In the centralized method, the xG Furthermore, the primary receiver uncertainty
base-station plays a role to gather all sensing infor- problem caused by the lack of the primary receiver
mation from the xG users and detect the spectrum location knowledge is still unsolved in the coopera-
holes. On the other hand, distributed solutions tive sensing. In the following section, we explain
require exchange of observations among xG users. interference-based detection methods, which aim
Cooperative detection among unlicensed users is to address these problems.
theoretically more accurate since the uncertainty in
a single user’s detection can be minimized [25]. 4.3. Interference-based detection
Moreover, the multi-path fading and shadowing
effect are the main factors that degrade the perfor- Interference is typically regulated in a transmit-
mance of primary user detection methods [44]. ter-centric way, which means interference can be
However, cooperative detection schemes allow to controlled at the transmitter through the radiated
mitigate the multi-path fading and shadowing power, the out-of-band emissions and location of
effects, which improves the detection probability in individual transmitters. However, interference actu-
a heavily shadowed environment [25]. ally takes place at the receivers, as shown in
In [55], the limitation of non-cooperative spec- Fig. 10(a) and (b).
trum sensing approaches is investigated. Generally, Therefore recently, a new model for measuring
the data transmission and sensing function are co- interference, referred to as interference temperature
located in a single xG user device. However, this shown in Fig. 11 has been introduced by the FCC
architecture can result in suboptimal spectrum deci- [21]. The model shows the signal of a radio station
sion due to possible conflicts between data transmis- designed to operate in a range at which the received
sion and sensing. In order to solve this problem, in power approaches the level of the noise floor. As
[55], two distinct networks are deployed separately, additional interfering signals appear, the noise floor
i.e., the sensor network for cooperative spectrum increases at various points within the service area, as
sensing and the operational network for data trans- indicated by the peaks above the original noise
mission. The sensor network is deployed in the floor. Unlike the traditional transmitter-centric
desired target area and senses the spectrum. A cen- approach, the interference temperature model
tral controller processes the spectrum information manages interference at the receiver through the
collected from sensors and makes the spectrum interference temperature limit, which is represented
occupancy map for the operational network. The by the amount of new interference that the receiver
operational network uses this information to deter- could tolerate. In other words, the interference tem-
mine the available spectrum. perature model accounts for the cumulative RF


Licensed Signal Minimum Service
Range with
Interference Cap
New Opportunities
for Spectrum Access Service Range at
Power at Interference Original
Original Noise Floor
Receiver Temperature Limit

Original Noise Floor

Fig. 11. Interference temperature model [21].

energy from multiple transmissions and sets a max- tively measuring the interference temperature.
imum cap on their aggregate level. As long as xG An xG user is naturally aware of its transmit
users do not exceed this limit by their transmissions, power level and its precise location with the help
they can use this spectrum band. of a positioning system. With this ability, however,
However, there exist some limitations in measur- its transmission could cause significant interfer-
ing the interference temperature. In [9], the interference at a neighboring receiver on the same fre-
ence is defined as the expected fraction of primary quency. However, currently, there exists no
users with service disrupted by the xG operations. practical way for a cognitive radio to measure or
This method considers factors such as the type of estimate the interference temperature at nearby
unlicensed signal modulation, antennas, ability to primary receivers. Since primary receivers are usu-
detect active licensed channels, power control, and ally passive devices, an xG user cannot be aware of
activity levels of the licensed and unlicensed users. the precise locations of primary receivers. Further-
However, this model describes the interference dis- more, if xG users cannot measure the effect of their
rupted by a single xG user and does not consider transmission on all possible receivers, a useful
the effect of multiple xG users. In addition, if xG interference temperature measurement may not
users are unaware of the location of the nearby pri- be feasible.
mary users, the actual interference cannot be • Spectrum sensing in multi-user networks: Usually,
measured using this method. xG networks reside in a multi-user environment
In [63], a direct receiver detection method is pre- which consists of multiple xG users and primary
sented, where the local oscillator (LO) leakage power users. Furthermore, the xG networks can also
emitted by the RF front-end of the primary receiver be co-located with other xG networks competing
is exploited for the detection of primary receivers. In for the same spectrum band. However, current
order to detect the LO leakage power, low-cost interference models [9,21] do not consider the
sensor nodes can be mounted close to the primary effect of multiple xG users. Multi-user environ-
receivers. The sensor nodes detect the leakage LO ment makes it more difficult to sense the primary
power to determine the channel used by the primary users and to estimate the actual interference.
receiver and this information is used by the unli- Hence, spectrum sensing functions should be
censed users to determine the operation spectrum. developed considering the possibility of multi-
user/network environment. In order to solve the
4.4. Spectrum sensing challenges multi-user problem, the cooperative detection
schemes can be considered, which exploit the spa-
There exist several open research challenges that tial diversity inherent in a multi-user network.
need to be investigated for the development of the • Detection capability: One of the main require-
spectrum sensing function. ments of xG networks is the detection of the
primary users in a very short time [24,53].
• Interference temperature measurement: The diffi- OFDM-based xG networks are known to be
culty of this receiver detection model lies in effec- excellent fit for the physical architecture of xG


networks [5,57,62]. Since multi-carrier sensing In order to describe the dynamic nature of xG
can be exploited in OFDM-based xG networks, networks, each spectrum hole should be character-
the overall sensing time can be reduced. Once a ized considering not only the time-varying radio
primary user is detected in a single carrier, sens- environment and but also the primary user activity
ing in other carriers is not necessary. In [57], a and the spectrum band information such as operat-
power-based sensing algorithm in OFDM net- ing frequency and bandwidth. Hence, it is essential
works is proposed for detecting the presence of to define parameters such as interference level,
a primary user. It is shown that the overall detec- channel error rate, path-loss, link layer delay, and
tion time is reduced by collecting information holding time that can represent the quality of a par-
from each carrier. However, this necessitates ticular spectrum band as follows:
the use of a large number of carriers, which
increases the design complexity. Hence, novel • Interference: Some spectrum bands are more
spectrum sensing algorithms need to be devel- crowded compared to others. Hence, the spectrum
oped such that the number of samples needed band in use determines the interference character-
to detect the primary user is minimized within a istics of the channel. From the amount of the
given detection error probability. interference at the primary receiver, the permissi-
ble power of an xG user can be derived, which is
used for the estimation of the channel capacity.
5. Spectrum management • Path loss: The path loss increases as the operating
frequency increases. Therefore, if the transmis-
In xG networks, the unused spectrum bands will sion power of an xG user remains the same, then
be spread over wide frequency range including both its transmission range decreases at higher fre-
unlicensed and licensed bands. These unused spec- quencies. Similarly, if transmission power is
trum bands detected through spectrum sensing increased to compensate for the increased path
show different characteristics according to not only loss, then this results in higher interference for
the time varying radio environment but also the other users.
spectrum band information such as the operating • Wireless link errors: Depending on the modula-
frequency and the bandwidth. tion scheme and the interference level of the spec-
Since xG networks should decide on the best trum band, the error rate of the channel changes.
spectrum band to meet the QoS requirements over • Link layer delay: To address different path loss,
all available spectrum bands, new spectrum man- wireless link error, and interference, different
agement functions are required for xG networks, types of link layer protocols are required at dif-
considering the dynamic spectrum characteristics. ferent spectrum bands. This results in different
We classify these functions as spectrum sensing, link layer packet transmission delay.
spectrum analysis, and spectrum decision. While • Holding time: The activities of primary users can
spectrum sensing, which is discussed in Section 4, affect the channel quality in xG networks. Hold-
is primarily a PHY layer issue, spectrum analysis ing time refers to the expected time duration that
and spectrum decision are closely related to the the xG user can occupy a licensed band before
upper layers. In this section, spectrum analysis and getting interrupted. Obviously, the longer the
spectrum decision are investigated. holding time, the better the quality would be.
Since frequent spectrum handoff can decrease
5.1. Spectrum analysis the holding time, previous statistical patterns of
handoff should be considered while designing
In xG networks, the available spectrum holes xG networks with large expected holding time.
show different characteristics which vary over time.
Since the xG users are equipped with the cognitive Channel capacity, which can be derived from the
radio based physical layer, it is important to under- parameters explained above, is the most important
stand the characteristics of different spectrum factor for spectrum characterization. Usually SNR
bands. Spectrum analysis enables the characteriza- at the receiver has been used for the capacity estima-
tion of different spectrum bands, which can be tion. However, since SNR considers only local
exploited to get the spectrum band appropriate to observations of xG users, it is not enough to avoid
the user requirements. interference at the primary users. Thus, spectrum


characterization is focused on the capacity estima- be selected for the current transmission considering
tion based on the interference at the licensed receiv- the QoS requirements and the spectrum characteris-
ers. The interference temperature model [21] given tics. Thus, the spectrum management function must
in Section 4.3 can be exploited for this approach. be aware of user QoS requirements.
The interference temperature limit indicates an Based on the user requirements, the data rate,
upper bound or cap on the potential RF energy that acceptable error rate, delay bound, the transmission
could be introduced into the band. Consequently, mode, and the bandwidth of the transmission can be
using the amount of permissible interference, the determined. Then, according to the decision rule, the
maximum permissible transmission power of an set of appropriate spectrum bands can be chosen. In
xG user can be determined. [73], five spectrum decision rules are presented,
In [63], a spectrum capacity estimation method which are focused on fairness and communication
has been proposed that considers the bandwidth cost. However, this method assumes that all chan-
and the permissible transmission power. Accord- nels have similar throughput capacity. In [36], an
ingly, the spectrum capacity, C, can be estimated opportunistic frequency channel skipping protocol
as follows: is proposed for the search of better quality channel,
where this channel decision is based on SNR. In
S
C ¼ B log 1 þ ; ð5Þ order to consider the primary user activity, the num-
N þI
ber of spectrum handoff, which happens in a certain
where B is the bandwidth, S is the received signal spectrum band, is used for spectrum decision [37].
power from the xG user, N is the xG receiver noise Spectrum decision constitutes rather important but
power, and I is the interference power received at yet unexplored issues in xG networks, which are pre-
the xG receiver due to the primary transmitter. sented in the following subsection.
Estimating spectrum capacity has also been
investigated in the context of OFDM-based cogni- 5.3. Spectrum management challenges
tive radio systems in [57]. Accordingly, the spectrum
capacity of the OFDM-based xG networks is There exist several open research issues that need
defined as follows [57]: to be investigated for the development of spectrum
Z decision function.
1 Gðf ÞS 0
C¼ log2 1 þ df ; ð6Þ
X 2 N0
• Decision model: Signal to noise ratio (SNR) is not
where X is the collection of unused spectrum seg- sufficient to characterize the spectrum band in xG
ments, G(f) is the channel power gain at frequency networks. Besides the SNR, many spectrum char-
f, S0 and N0 are the signal and noise power per unit acterization parameters would affect the quality,
frequency, respectively. as investigated in Section 5.1. Thus, how to
The recent work on spectrum analysis, as dis- combine these spectrum characterization para-
cussed above, only focuses on spectrum capacity meters for the spectrum decision model is still
estimation. However, besides the capacity, other an open issue. Moreover, in OFDM based xG
factors such as delay, link error rate, and holding networks, multiple spectrum bands can be simul-
time also have significant influence on the quality taneously used for the transmission. For these
of services. Moreover, the capacity is closely related reasons, a decision framework for the multiple
to both interference level and path loss. However, a spectrum bands should be provided.
complete analysis and modeling of spectrum in xG • Multiple spectrum band decision: In xG networks,
networks is yet to be developed. In order to decide multiple spectrum bands can be simultaneously
on the appropriate spectrum for different types of used for the transmission. Moreover, the xG net-
applications, it is desirable and an open research works do not require the selected multiple bands
issue to identify the spectrum bands combining all to be contiguous. Thus, an xG user can send
characterization parameters described above. packets over non-contiguous spectrum bands.
This multi-spectrum transmission shows less
5.2. Spectrum decision quality degradation during the spectrum handoff
compared to the conventional transmission on
Once all available spectrum bands are character- single spectrum band [5]. For example, if a pri-
ized, appropriate operating spectrum band should mary user appears in a particular spectrum band,


the xG user has to vacate this band. However, purposes. To realize the ‘‘Get the Best Available
since the rest of spectrum bands will maintain Channel’’ concept, an xG radio has to capture the
the communication, abrupt service quality degra- best available spectrum. Spectrum mobility is
dation can be mitigated. In addition, transmis- defined as the process when an xG user changes
sion in multiple spectrum bands allows lower its frequency of operation. In the following sections,
power to be used in each spectrum band. As a we describe the spectrum handoff concept in xG net-
result, less interference with primary users is works and discuss open research issues in this new
achieved, compared to the transmission on single area.
spectrum band [5]. For these reasons, a spectrum
management framework should support multiple 6.1. Spectrum handoff
spectrum decision capabilities. For example, how
to determine the number of spectrum bands and In xG networks, spectrum mobility arises when
how to select the set of appropriate bands are still current channel conditions become worse or a pri-
open research issues in xG networks. mary user appears. Spectrum mobility gives rise to
• Cooperation with reconfiguration: The cognitive a new type of handoff in xG networks that we refer
radio technology enables the transmission to as spectrum handoff. The protocols for different
parameters of a radio to be reconfigured for opti- layers of the network stack must adapt to the chan-
mal operation in a certain spectrum band. For nel parameters of the operating frequency. More-
example, if SNR is fixed, the bit error rate over, they should be transparent to the spectrum
(BER) can be adjusted to maintain the channel handoff and the associated latency.
capacity by exploiting adaptive modulation tech- As pointed out in earlier sections, a cognitive
niques, e.g., cdma2000 1x EVDO [1,18]. Hence, a radio can adapt to the frequency of operation.
cooperative framework that considers both spec- Therefore, each time an xG user changes its fre-
trum decision and reconfiguration is required. quency of operation, the network protocols are
• Spectrum decision over heterogenous spectrum going to shift from one mode of operation to
bands: Currently, certain spectrum bands are another. The purpose of spectrum mobility manage-
already assigned to different purposes while some ment in xG networks is to make sure that such tran-
bands remain unlicensed. Thus, the spectrum sitions are made smoothly and as soon as possible
used by xG networks will most likely be a combi- such that the applications running on an xG user
nation of exclusively accessed spectrum and unli- perceive minimum performance degradation during
censed spectrum. In case of licensed bands, the xG a spectrum handoff. It is essential for the mobility
users need to consider the activities of primary management protocols to learn in advance about
users in spectrum analysis and decision in order the duration of a spectrum handoff. This information
not to influence the primary user transmission. should be provided by the sensing algorithm. Once
Conversely, in unlicensed bands, since all the xG the mobility management protocols learn about this
users have the same spectrum access rights, latency, their job is to make sure that the ongoing
sophisticated spectrum sharing techniques are communications of an xG user undergo only mini-
necessary. In order to decide the best spectrum mum performance degradation.
band over this heterogeneous environment, xG Consequently, multi-layer mobility management
network should support spectrum decision protocols are required to accomplish the spectrum
operations on both the licensed and the unli- mobility functionalities. These protocols support
censed bands considering these different charac- mobility management adaptive to different types
teristics. of applications. For example, a TCP connection
can be put to a wait state until the spectrum handoff
is over. Moreover, since the TCP parameters will
6. Spectrum mobility change after a spectrum handoff, it is essential to
learn the new parameters and ensure that the transi-
xG networks target to use the spectrum in a tion from the old parameters to new parameters are
dynamic manner by allowing the radio terminals, carried out rapidly. For a data communication e.g.,
known as the cognitive radio, to operate in the best FTP, the mobility management protocols should
available frequency band. This enables ‘‘Get the implement mechanisms to store the packets that
Best Available Channel’’ concept for communication are transmitted during a spectrum handoff, whereas


for a real-time application there is no need to store • Spectrum mobility in space: The available bands
the packets as the stored packets, if delivered later, also change as a user moves from one place to
will be stale packets and can not be used by the cor- another. Hence, continuous allocation of spec-
responding application. trum is a major challenge. in xG networks.

6.2. Spectrum mobility challenges in xG networks 7. Spectrum sharing

The followings are the open research issues for In xG networks, one of the main challenges in
efficient spectrum mobility in xG networks. open spectrum usage is the spectrum sharing. Spec-
trum sharing can be regarded to be similar to gen-
• At a particular time, several frequency bands eric medium access control (MAC) problems in
may be available for an xG user. Algorithms existing systems. However, as we will investigate in
are required to decide the best available spectrum this section, substantially different challenges exist
based on the channel characteristics of the avail- for spectrum sharing in xG networks. The coexis-
able spectrum and the requirements of the appli- tence with licensed users and the wide range of
cations that are being used by an xG user. available spectrum are two of the main reasons for
• Once, the best available spectrum is selected, the these unique challenges. In this section, we delve
next challenge is to design new mobility and con- into the specific challenges for spectrum sharing in
nection management approaches to reduce delay xG networks, overview the existing solutions and
and loss during spectrum handoff. discuss open research areas.
• When the current operational frequency becomes In order to provide a directory for different chal-
busy (this may happen if a licensed user starts to lenges during spectrum sharing, we first enumerate
use this frequency) in the middle of a communi- the steps in spectrum sharing in xG networks. The
cation by an xG user, then applications running challenges and the solutions proposed for these
on this node have to be transferred to another steps will then be explained in detail. The spectrum
available frequency band. However, the selection sharing process consists of five major steps.
of new operational frequency may take time.
Novel algorithms are required to ensure that 1. Spectrum sensing: An xG user can only allocate a
applications do not suffer from severe perfor- portion of the spectrum if that portion is not used
mance degradation during such transitions. by an unlicensed user. In Section 4, the solutions
• Spectrum handoff may occur due to reasons and the challenges for this problem, i.e., spec-
other than the detection of the primary user. trum sensing, are described. Accordingly, when
Thus, there exist various other spectrum handoff an xG node aims to transmit packets, it first
schemes in xG networks. If an xG user moves needs to be aware of the spectrum usage around
from one place to another, spectrum handoff its vicinity.
may occur just because the available spectrum 2. Spectrum allocation: Based on the spectrum
bands change. Thus the desired spectrum handoff availability, the node can then allocate a channel.
scheme should integrate inter-cell handoff. Apart This allocation not only depends on spectrum
from this, spectrum handoff between different availability, but it is also determined based on
networks, referred to as vertical handoff is also internal (and possibly external) policies. Hence,
likely to occur in xG networks. Under such a the design of a spectrum allocation policy to
diverse environment, it is essential that spectrum improve the performance of a node is an impor-
handoff scheme takes all the above mentioned tant research topic.
possibilities into consideration. 3. Spectrum access: In this step, another major
• Spectrum mobility in time domain: xG networks problem of spectrum sharing comes into picture.
adapt to the wireless spectrum based on available Since there may be multiple xG nodes trying to
bands on the spectrum. Since these available access the spectrum, this access should also be
channels change over time, enabling QoS in this coordinated in order to prevent multiple users
environment is challenging. The physical radio colliding in overlapping portions of the spectrum.
should ‘‘move’’ through the spectrum to meet 4. Transmitter-receiver handshake: Once a portion of
the QoS requirements. the spectrum is determined for communication,

Ne xt generation dynamic spectrum access cognitive2006survey

Ne xt generation dynamic spectrum access cognitive2006survey

Recommended

Recommended

More Related Content

What's hot

What's hot (18)

Viewers also liked

Viewers also liked (20)

Similar to Ne xt generation dynamic spectrum access cognitive2006survey

Similar to Ne xt generation dynamic spectrum access cognitive2006survey (20)

Ne xt generation dynamic spectrum access cognitive2006survey