Dynamic Spectrum Access for

Cognitive Radio
by
Maniraj Angdambay

Seminar Coordinator
Dr. Andrea Munari
Institute for Networked Systems (iNETS)

A report submitted in partial fulfillment of the requirements for seminar on

Current Trends of Wireless Communications

February 2015
Abstract
Current licensing policy based on fixed spectrum allocation has created artificial spectrum
scarcity but measurement campaigns on various spatial and temporal regions show presence of
large proportion of unused or underutilized spectrum (spectrum holes/white spaces). Dynamic
Spectrum Access (DSA) and its implementation via Cognitive Radio (CR) could be the solution
to improve the efficiency of spectrum usage. For implementation of DSA in CR, changes need to
be incorporated on RF front-end in order to accommodate wide range sensing capability. Primary
User (PU) has higher priority or rights on the usage of specific part of spectrum while a
secondary User (SU) has a lower priority on the usage of spectrum access and needs to exploit
spectrum used by the primary user. In this regard, efficient detection mechanism for the presence
of PU is major challenge which needs to guarantee safety of primary user as well as spectrum
usage efficiency in secondary user. Free for all opportunistic spectrum access where any device
can transmit whenever needed would be an ideal scenario but it brings huge complexity in
hardware requirements and protocols implementation. In this regard, Coordinated DSA via use
of Spectrum broker could be an attractive solution in near future. Finally, there is a need for an
efficient and truthful auctioning policy such as VERITAS for the realization of Dynamic
spectrum access.

i
Table of Contents
Abstract ............................................................................................................................................ i
List of Figures ................................................................................................................................ iii
1 Introduction ............................................................................................................................. 1
1.1 Spectrum Access Barrier .................................................................................................. 2
1.2 Dynamic Spectrum Access............................................................................................... 3
2 Cognitive Radio ....................................................................................................................... 5
2.1 Cognitive radio architecture ............................................................................................. 5
2.2 Function of CR ................................................................................................................. 6
2.3 CR components ................................................................................................................ 7
2.4 Spectrum Sensing and identification ................................................................................ 7
2.4.1 Non cooperative sensing ........................................................................................... 8
2.4.2 Cooperative sensing .................................................................................................. 9
2.4.3 Interference based sensing ........................................................................................ 9
3 Coordinated Dynamic Spectrum Access ............................................................................... 10
3.1 Overview and approach .................................................................................................. 11
3.1.1 Coordinated Access Band (CAB) ........................................................................... 11
3.1.2 Statistical Multiplexing of Spectrum Access .......................................................... 11
3.1.3 DIMSUMnet architecture ....................................................................................... 12
4 Auctioning policy for DSA.................................................................................................... 14
4.1 VERITAS Algorithm ..................................................................................................... 14
4.2 VERITASs Truthfulness ............................................................................................... 15
4.3 VERITASs Efficiency .................................................................................................. 16
5 Conclusion ............................................................................................................................. 18
Bibliography ................................................................................................................................. 19

ii
List of Figures
Figure 1.1 Radio frequency allocation in US [3]. ...................................................................... 1
Figure 2.1 Protocol stack in Cognitive radio [11]. ..................................................................... 5
Figure 2.2 Functions of CR depicted in a cognitive cycle [11]. ................................................. 6
Figure 2.3 Components of Cognitive Radio [11]. ...................................................................... 7
Figure 2.4 Detection method with sensing accuracies and implementation complexities [19]. 8
Figure 2.5 Hidden Node problem [11]. ...................................................................................... 9
Figure 3.1 Options on DSA implementation [28]. ................................................................... 10
Figure 3.2 Model of Coordinated DSA [28]. ........................................................................... 10
Figure 3.3 Coordinated Access Band (CAB) [28]. .................................................................. 11
Figure 3.4 Spectrum utilization at location (x,y) in multi-provider network [28]. .................. 12
Figure 3.5 DIMSUMnet architecture [28]. ............................................................................... 12
Figure 4.1 Spectrum utilization, revenue and Bidder satisfaction in Non truthful auction [29].
15
Figure 4.2 Spectrum utilization, revenue and Bidder satisfaction in VERITAS auction [29]. 15
Figure 4.3 Spectrum utilization of VERITAS vs. simple truthful design (8 channels auction to
10-500 bidders) [29]. .................................................................................................................... 16
Figure 4.4 Spectrum utilization of VERITAS vs. simple truthful design (1-30 channels auction
to 300 bidders) [29]....................................................................................................................... 16

iii
1 Introduction
Radio spectrum refers to the range of frequency spectrum from 3 kHz-300 GHz which can be
used for wireless communications. RF spectrum is a valuable natural resource. Proper spectrum
management is necessary in order to maximize social benefit through efficient use as well as
maintaining quality of service by limiting the interference [1].

Traditional fixed spectrum allocation techniques are based on the administrative approach also
called command and control model which is currently employed by most regulators around the
globe. In this traditional model, regulators are the centralized authority who decides on the range
of services and technological standards that are permitted on a particular range of frequencies.
The entire spectrum is divided into chunks of blocks, each of which is allocated for certain
services [2]. In the US scenario, Federal Communications Commission (FCC) is the regulatory
body. Current fixed spectrum management policy dates back to initial era of wireless
communication where signal interference was a major issue rather than spectrum availability. In
the 80, the only major concern was radio and television broadcasting and large part of the
spectrum was unallocated. But along with the innovation in wireless technologies and high
demand for wireless broadband services, most part of the spectrum has already been populated
via different services and wireless standards [3].

Figure 1.1 Radio frequency allocation in US [3].

1
From Figure 1.1, we observe that most part of the spectrum has already been assigned leaving
very little space for the accommodation of new services. Similar statistics are observed regarding
frequency allocation in all part of the globe. But from the measurement campaign undertaken by
FCCs Spectrum Policy Task Force, it was observed that most part of the overpopulated
spectrum remain unused or underutilized both in temporal and spatial domain [4].Similar
measurement campaign in US and Europe brought into notice the implication of the current static
frequency planning. Large parts of the allocated radio spectrum remain unused or underutilized.
It was also observed that bands allocated for military, government and public safeties were rarely
in operation [5].
It is the current policy that has resulted in underutilization and this in turn has resulted in the
artificial scarcity of the spectrum. This has also restricted technological innovation as we can see
the range of services and technologies emerging in ISM band in comparison to the restricted
bands. Thus, current licensing policy is inefficient which has resulted in bandwidth as scarcest
resources for wireless communication [6].

1.1 Spectrum Access Barrier

From the finding of various measurement campaigns [4] [5], it is seen that physical scarcity of
the spectrum is not the big problem. But it is the various spectrum access barriers set by the
legacy regulatory bodies that is creating inefficiency [6].

Following five access barriers sum up the limitation in spectrum access due to static spectrum
allocation scheme:

I. Flex Use Barrier: Current regulations prevent the license holder the flexibility to employ
the technology that suits its commercial requirements by dictating the allowed services
and standard governed by license. Only certain standard or services are allowed to
operate in allocated spectrum. As an example, TV broadcast use 6MHz spectrum for
transmission which have proved to be wasteful because better quality video can be
transmitted using less bandwidth. But the current licensing policy prevents the license
holder to use the spare license for other commercial services [6]. Regulatory bodies in
India and Europe have allocated GSM operator more spectrum (50MHz) compared to the
CDMA thus awarding a wasteful technology with more spectrum [7]. There is need for
the current licensing model to be technology neutral in order to improve spectrum usage
efficiency.
II. Service Silo Barrier: Current policy also dictates mapping of specific services to fixed
bands. Service providers are prevented from offering services other than specified in the
agreement creating spectrum holes. Service silo restriction can be observed in TV bands
and several public safety bands which show below average utilization of spectrum. In
absence of these barriers, these bands could be exploited for cellular services due to its
close proximity with cellular bands [6].

2
III. License Scope Barrier: Current licensing scheme award license for large area and long
period of time. Population on the other hand is not uniformly distributed and mostly
concentrated in cities. License scope barrier prevent other services to exploit these
spectrum holes in rural regions. Change is required to reduce the licensing scope both in
spatial and temporal domain [6].
IV. License granularity Barrier: As license is awarded in large chunks, it prevents the entry
of small players to acquire the spectrum both in spatial and temporal scale. It also
prevents the cellular operator to acquire small amounts of spectrum to meet the traffic
requirements. Currently, there is no policy that enable acquiring spectrum on small
chunks [6].
V. Secondary Usage Barrier: Only the devices owned by the license holder have the
exclusive right to spectrum. Thus opportunistic low power usage of spectrum in spectrum
hole rich area is not allowed by the current regulation [6].

1.2 Dynamic Spectrum Access

In order to break these spectrum access barriers created by the legacy spectrum access policy and
improve the overall utilization of the scare spectrum resource, we require a shift from static
spectrum allocation to the notion of Dynamic Spectrum Access (DSA). This notion has gained
support after the FCCs Spectrum Policy Task Force (SPTF) report which recommends a need to
shift from traditional command-and-control spectrum management to a more dynamic
approach taking into account the rapid innovation in wireless technologies [4]. SPTF recommend
for maximum flexibility of spectrum usage, promote efficient spectrum usage and account for all
dimension of spectrum use.

DSA can be achieved under three models:

I. Exclusive Use Model: This model maintains the structural policy of current static
allocation scheme but provides flexibility via two approaches: spectrum property rights
[8] and dynamic spectrum allocation [9].Spectrum property rights model allow the
license holder to trade their spectrum and also provide flexibility in service and
technology. The latter approach provides spectrum efficiency by exploiting the spectrum
holes in spatial and temporal domain. It creates flexibility in spectrum usage as well as
provides comparable advantage in spectrum efficiency compared to legacy regulation but
fails to utilize the spectrum holes due to burst nature of traffic [10].
II. Open Sharing Model: Also called common use, this model aims towards free for all
opportunistic access of the spectrum [11]. Support for the open sharing model has be
generated from the success in the ISM band which implement free for all opportunistic
access [12].This model enables dynamic spectrum access among both homogenous and
heterogeneous networks.

3
III. Shared Use model: In this model, unlicensed secondary users can opportunistically access
the licensed band when vacant thus avoiding interference to Primary User. Spectrum overlay
and spectrum underlay models are proposed for providing opportunistic access to secondary
user [11]. In spectrum underlay approach, constraints are set on the transmission power of
secondary user in order to minimize interference to primary user. As primary user safety is
given higher priority by setting power constraints on secondary user, spectrum holes are not
optimally exploited by this model. On the other hand, spectrum overlay approach doesnt
introduce power restriction condition to secondary user. Secondary User transmission is
restricted on temporal and frequency domain. On the allocated time and designated spectrum
area of the licensed band, it is the task of the secondary user to sense and exploit the
spectrum holes.

4
2 Cognitive Radio
Dynamic spectrum access is the way forward to break the spectrum access barriers set by the
traditional fixed spectrum access policies. DSA implementation requires innovation in the
current hardware technologies which are limited towards operating on narrow frequencies ranges
and fixed radio standards. But advances in wireless communications in the field of high speed
digital signal processing, wideband A/D, Multiple inputs multiple output (MIMO) system and
software defined radios make realization of smart devices capable of meeting the objective of
Dynamic spectrum [6]. Through wide band RF components and A/D, transceivers are capable of
operating on wide frequency range. Development in high speed digital signal processing
hardware will facilitate in the integration of higher modulation schemes, MIMO processing as
well as flexibility in reconfiguration of RF parameters. The need for a shift towards DSA and the
advance in the digital communication system has lead towards the idea of Cognitive radio (CR).

The term Cognitive radio was first coined by Mitola in late 90 [13] [14] [15], introducing the
concept of an intelligent radio that is self aware of its operating environment and can
autonomously make decisions from the information about its environment. According to FCC
[16], A Cognitive Radio is a radio that can change its transmitter parameters based on
interaction with the environment in which it operates. ITUs Radio Communication Study
Group has similar definition for CR [17]: A radio or system that senses, and is aware of, its
operational environment and can dynamically and autonomously adjust its radio operating
parameters accordingly.

2.1 Cognitive radio architecture

Figure 2.1 Protocol stack in Cognitive radio [11].

CR needs to ensure highly efficient and reliable communication on one hand while maximizing
the utilization of the frequency spectrum on the other. Figure 2.1 shows protocol stack in typical
CR. Software defined radio (SDR) is used in RF front end in order to provide flexibility in
reconfiguration of the transmission parameters. On higher layers of the protocol stack, smart
adaptive protocols should be implemented which are capable of detecting variations in the CR
environment. Interlink among SDR front end, adaptive protocols and higher layer application is

5
facilitated via a cognitive radio module. Thus CR module should be equipped with a smart
algorithm that is capable of processing the physical layer information as well as control the
different layer protocol parameters based on operating environment and application demand [11].

2.2 Function of CR

Figure 2.2 Functions of CR depicted in a cognitive cycle [11].

In order to support dynamic spectrum access, CR should have the following characteristics.
These major functions are can be summed up through a cognitive cycle as shown in Figure 2.2.

Spectrum Sensing: CR should have smart sensing capability for detecting any minute
environmental variations [18]. The main aim of spectrum sensing is to determine the
status of the spectrum and availability of primary user in the frequency band. Sensing can
be done either in centralized fashion or distributed. In the former scenario, a central unit
(e.g. Base station) does the sensing and information obtained is shared among other
nodes in system. Since all the sensing functions are performed at the central unit,
centralized spectrum sensing reduces complexity in the user terminals. But centralized
sensing faces hidden node problem as central controller may not be able to detect
secondary user at the edge of cell. Hidden node problem occurs when a node is visible
from a wireless access point (AP), but not from other nodes communicating with that AP.
In distributed sharing, each node are equipped with sensing responsibilities and the
information can be used individually (i.e. non-cooperative sensing) or shared among
peers (i.e. cooperative sensing).Sharing information among peers increase the accuracy of
the spectrum sensing at the expense of communication overhead [11].
Learning and adaptability: CR should be able to analyze the information obtained from
spectrum sensing and make necessary changes in the operating parameter based on the
scenario [18].
Spectrum mobility and agility: CR should be able to change the operating parameter on
the fly. Full flexibility can be achieved through software defined radio implementation
and wideband antenna technologies [18].For example, when primary user comes live in
the band operated by an unlicensed user, secondary user should be able to change to idle
spectrum band also called spectrum handoff [11].

6
2.3 CR components
A cognitive radio should be equipped with following major component block as depicted in
Figure 2.3 in order to meet its operating characteristics.

Figure 2.3 Components of Cognitive Radio [11].

I. Transceiver: An SDR based transceiver forms a major component of cognitive radio

hardware with capability of signal transmission and reception. Transmission parameter is
dynamically changeable based on higher layer protocol while receiver is equipped with
spectrum sensing capability [11].
II. Spectrum analyzer: Signal information received is analyzed via spectrum analyzer for
the presence of spectrum holes. It is the task of the spectrum analyzer to guarantee safety
of the primary user [11].
III. Learning and knowledge extraction: This module is responsible for maintaining and
upgrading the spectrum usage information obtained from spectrum analyzer in the device
database. This information is used for optimization and adaptation of transmission
parameters. In this regard, machine learning algorithms can be applied for learning and
knowledge deduction [11].
IV. Decision making: From the spectrum usage statistics, decision making process for
accessing the spectrum is carried on. Various decision making strategies can be
implemented based on the objective of the devices. For device with a single objective,
optimization theory can be implemented while system with multiple objectives can be
modeled by game theory [11].

2.4 Spectrum Sensing and identification

Spectrum sensing is the most important attribute of cognitive radio. Spectrum sensing is the
ability of the CR device to gather information about the current spectrum usage information and
provide decision on the presence of primary user [19]. Spectrum sensing can be broadly
classified under three categories: non-cooperative sensing, cooperative sensing and interference
based sensing [11].

7
2.4.1 Non cooperative sensing
In this approach, unlicensed user solely detects the primary user from measurements and local
observations. Performance of the sensing techniques is given in terms of probability of miss
detection and probability of false alarm. Miss detection occurs when a busy channel is detected
as idle which could lead to a collision with the PU and transmission for both primary and
secondary user is wasted. A false alarm occurs when an idle channel is detected busy. This is lost
opportunity for transmission to secondary user. [11] [20]. Following detection methods can be
implemented in non-cooperative sensing:

I. Matched Filter detection: Match filter is an optimal detector when apriori knowledge
about the information of the signal from primary user is known such as modulation
format or pilots [21]. Presence of primary user is detected by comparing the input signal
with these known signal format [11]. In presence of signal information, matched filtering
require less computation time for decision compared to other detection methods. But for
match filtering, we need to demodulate the received signal. This requires perfect
knowledge about the signal characteristics bringing complexity in implementation and
large power consumption [19].
II. Energy detection: Energy detection is the most common sensing mechanism due to its
low implementation complexity [19].This detection technique is optimal when no signal
information of primary user is known. In this approach, output from band pass filter is
squared and integrated over the observation interval. Output obtained from the integrator
is compared with certain threshold for the decision process [11]. Challenges associated
with energy detection include setting suitable detection threshold, lack of differentiation
between signal and noise and degrading performance under poor SNR condition [22].
III. Cyclostationary feature detection: This method exploit the periodic properties of the
signal also called cyclostationarity to detect the presence of licensed user. Period pattern
of signal help to distinguish between noises which are uncorrelated and provide a better
sensing result. But this approach requires long observation time and complex
implementation [11]. One way to implement cyclostationary detection is using pattern
recognition scheme based on neural network [23].

Figure 2.4 Detection method with sensing accuracies and implementation complexities [19].

8
Tradeoff among various detection methods is shown in Figure 2.4.One or more of the spectrum
sensing methods can be implemented for improving efficiency. Energy detection can be used
initially for quick scanning of the spectrum to eliminate bands with high density of primary
users. Match filter and cyclostationary detection can be applied on the remaining candidate bands
for accurate detection [22].

2.4.2 Cooperative sensing

Figure 2.5 Hidden Node problem [11].

However sensitive is the detection mechanism in cognitive radio devices, there is always a
possibility of miss detection and false alarm due to geographical condition and fading scenario.
As show in Figure 2.5, unlicensed users are out of the range of licensed user and cant detect the
signal transmitted by the transmitter and receiver of licensed devices. This scenario where the
nodes are out of range of other nodes or a collection of nodes is called hidden node problem. As
a result of hidden node problem, unlicensed user transmits and cause interference with the
receiver of the licensed user. Hidden node problem in non cooperative sensing can be solved via
cooperative sensing. The probability of detection can be improved by exchanging spectrum
sensing information among peer unlicensed users [11].This approach help remove the hidden
node problem where one or more of the secondary user many be in the shadowing area. Better
detection probability comes at the expense of greater signaling, computation overhead and cost
of the sensor network. One approach of implementation is the deployment of a sensor network
for sensing and operational network for spectrum access. All spectrum usage information is
collected via sensor network which are then distributed to operational network for optimum
spectrum access. [24].

2.4.3 Interference based sensing

FCC has proposed the concept of interference based sensing. In this approach, interference level
and noise are measured at the receiver of the primary user. Secondary users read this information
and adjust their operating parameters to operate under the interference temperature limit.
Secondary user can obtain information on interference level by observing the feedback from
primary user.

9
3 Coordinated Dynamic Spectrum Access
Most of the research on DSA has been targeted towards open sharing and shared user model of
uncoordinated access of spectrum [25] [26] such as DARPA XG [27]. But uncoordinated
spectrum access approach brings huge complexity in the protocol implementation, sensing range
and agility requirement at secondary user devices. Due to protocol complexities and hardware
limitations, initial implementation of Dynamic spectrum access based on open sharing and
shared user model can only be extended among homogenous network [25].

Figure 3.1 Options on DSA implementation [28].

Figure 3.1 shows various options on DSA implementation. On the left lies current static
spectrum allocation schemes and on far right is the free for all uncoordinated spectrum access.
As we move from left to right, complexity in protocol, sensing range, agility requirement and
heterogeneity of network increases. Looking at these challenges, a transition from current static
spectrum allocation towards coordinated dynamic spectrum access would be an attractive
solution in the near future [28].

Figure 3.2 Model of Coordinated DSA [28].

Figure 3.2 provides a model of Coordinated DSA for infrastructure based heterogeneous
networks where access to spectrum is regulated by a new entity called spectrum broker who
owns the right to lease the spectrum. Unlike fixed spectrum allocation, service providers do not
apriori own the spectrum and request time bound lease from the spectrum brokers. Compared to

10
free for all model, only the providers who are granted with the lease can operate. This approach
will enable new services to enter the market and potentially reduce cost. Overall, it also helps in
eliminating the spectrum usage inefficiency created by current fixed spectrum policy by enabling
bandwidth on demand [28].

3.1 Overview and approach

The concept of Coordinated DSA is realized through the introduction of new Coordinated
Access Band (CAB), Statistical Multiplexing of Spectrum access in CAB and DIMSUMnet
network architecture [28].

3.1.1 Coordinated Access Band (CAB)

Figure 3.3 Coordinated Access Band (CAB) [28].

CAB band are allocated by the regulatory bodies on various part of the spectrum for the purpose
of dynamic spectrum access. The ownership of the CAB band is given to the spectrum broker
who grants time bound lease to the requesters. Spectrum broker has the right to set the spectrum
lease condition such as region of operation, transmission power and nature of use (exclusive or
shared) [28].

Lease duration in CAB band is for short term and on temporary basis. It is different from current
licensed band where ownership is for long term, cover large spatial area and on permanent basis.
In addition, leasing process is automated which makes the licensing procedure fast and suitable
for bandwidth on demand service. In comparison to the current unlicensed band, CAB band
introduce the concept of leasing and the spectrum is also priced enabling revenue generation
[28].

Allocation of CAB band adjacent to cellular, PCS, unused TV bands and underutilized public
safety band as shown in Figure 3.3 enables dynamic spectrum access with minimum changes in
the hardware requirements on client and network. Current cellular providers will continue
operating in their existing network and only request additional CAB spectrum when experiencing
traffic overload. This help cost minimization in cell splitting and installation of new base
stations. TV operators can also take the time bound lease of CAB bands allocated in LMDS and
MMDS bands to enhance their broadband and backhaul services [28].

3.1.2 Statistical Multiplexing of Spectrum Access

While the concept of CAB band improves the efficiency in Spectrum access, multiplexing of the
CAB spectrum is implemented to help improve further spectrum utilization of CAB band..

11
 Figure 3.4 Spectrum utilization at location (x,y) in multi-provider network [28].

The spectrum usage at location (x, y) of service providers A and C with two base stations and
service provider B with one base station in operation is shown in Figure 3.4. For time period (t1,
t2) spectral utilization of provider B is at its peak while other two service providers are rarely
populated. Although per-provider spectrum usage at (x, y) is time varying, cell site analysis
allocates peak spectral bandwidth of P units per site considering the worst case scenario. Instead
of allocating peak bandwidth of P units per site if spectrum utilization requirement can be
aggregated across multiple service providers then the aggregate demand will be less than total of
peak demand. Taking advantage of the uncorrelated bandwidth requirements of various providers
in space and time, the total aggregate spectrum requirement by all major service providers can be
determined for certain area. Thus, keeping the aggregate spectrum use at particular location
constant will result in the most efficient use of spectrum. This aggregation approach of spectrum
is called statistically multiplexed access to spectrum. Since the aggregate demand is less than the
sum of the peak demands, it helps in reduction of the required spectrum and prevents spectrum
waste [28]

3.1.3 DIMSUMnet architecture

Figure 3.5 DIMSUMnet architecture [28].

12
DIMSUMnet stands for Dynamic Intelligent Management of Spectrum for Ubiquitous Mobile
access network which implements statistically multiplexed access to CAB spectrum. As shown
in Figure 3.5, DIMSUMnet architecture constitute of the following major components [28]:

I. A Spectrum Information and Management (SPIM) broker

II. A radio access network(RAN) consisting of new type of base stations
III. RAN manager(RANMAN)
IV. Intelligent mobile node (MN)
V. SPEctrum lease (SPEL) protocol
VI. Spectrum Information Channel(SPIC) protocol

DIMSUMnet architecture has two modes of operations. In simpler scenarios, end users do not
participate in the spectrum leasing process. Here, BS in RAN registers with the designated
RANMAN when active. RANMAN is responsible for negotiating desired amount of spectrum
lease with the spectrum broker. After the lease is successful then BS is configured to operate in
the allocated band by the RANMAN. Finally BS configures the MNs under its area to offer
services. SPEL and SPIC protocols are implemented for information exchange in the
DIMSUMnet architecture. The three party (the SPIM broker, the RANMAN and the BS) secure
interactions during the leasing process are carried out via SPEL protocol. SPIC protocol is
implemented for information exchange between BS and MNs over the SPI channels in the CAB
band. After BS broadcasts spectrum information which is received from RANMAN to its MNs,
clients use the information to request for application services. Two ways information exchange
between BS and MNs form the core of SPIC protocol [28].

In advance scenario, MNs are also capable for requesting spectrum. MNs request BS to acquire
spectrum via SPIC protocol. This mode of operation is handy when end user is in need for extra
bandwidth e.g. large file transfer. BS combines all such request from MNs and forward to
spectrum broker through RANMAN [28].

In the leap forward towards dynamic spectrum access, coordinated real time dynamic spectrum
access would be a more feasible transition compared to free for all opportunistic access. Concept
of CAB band and statistically multiplexed access to spectrum will improve spectrum utilization
without much change in the hardware requirements. In contrast to current fixed network
architecture, addition of SPIM broker and RANMAN makes it possible for Network operators to
request for additional spectrum in real time. Hence, realization of coordinated DSA is possible
due to introduction of CAB band, spectrum leasing of the CAB spectrum via regional spectrum
broker and new DISMUMnet architecture. [28].

13
4 Auctioning policy for DSA
As stated earlier, market driven DSA implementation drastically improve the overall spectrum
utilization [4] [6] [28].For the implementation of DSA, there is a need for suitable spectrum
auctioning policy that can guarantee both fairness in spectrum access as well as optimize
spectrum usage. What is a truthful auction policy? A truthful auction policy should guarantee
that when a spectrum bidder bids its true valuation for spectrum then its utility wont be less
compared to when it lies. In this regard, if the auctioning policy charges its winner independent
of their actual bid then it will remove the overhead in making strategies to win the bid and hence
prevent market manipulation [29].

Classical auctioning policies such as sealed- bid second- price (also called VCG auction) [30]
[31] are truthful for general auctions (e.g. paintings, electricity).In a secondary price sealed bid
auction, each bidder submits a sealed bid to the seller. The highest bidder wins the bid but
charged with the price quoted by the second-highest bidder. In regard to spectrum auctions,
spectrum is reusable and there are interdependencies among bidders who are requesting same
chunks of spectrum for different location and time. Traditional auctioning policies such as
sealed-bid second price auction are designed for general auction which do not consider
interdependencies of the bidders (i.e. only one winner for particular auction).Hence, they fail to
maintain truthfulness when applied to spectrum auction [29].

VERITAS is a truthful auctioning model that is suitable for Dynamic spectrum access. It is a
sealed bid auction where bidders submit bids privately and are charged with the price less than or
equal to their actual bids. Truthfulness is achieved by VERITAS through efficient spectrum
allocation and pricing mechanism making it suitable for online short term auction. For bidders, it
provides flexibility in requesting channels by exact numbers or in a range. For auctioneer,
VERITAS allows flexibility to configure the order of allocation and amount of spectrum offered
to maximize their revenue. For untruthful auction, revenue increases with the number of winning
bidders as they charge winner with their actual bid. VERITAS maintains truthfulness because
revenue initially increases with increase in winning bidders but decrease after certain threshold
[29].

4.1 VERITAS Algorithm

Notation:
di No. of Channel requested by bidder i
D = {d1, d2,..., dn} Set of channels demanded by n bidders
bi Price quoted by bidder i per channel
B={ b1,b2 ,b3 bn } Set of price quoted by n bidder
N(i) Neighbor of bidder i which cant share same spectrum

Algorithm [29]:
1: B=sorted B
2: while B! = 0 do
3: i=TOP (B);

14
4: if Distinct (N (i)) + di <= k then
5: Assign (i, di)
6: end if
7: B = B \ {bi}
8: end while
VERITAS algorithm sorts the bids set B and allocate bidders from highest to the lowest one
based on their bid price. During allocation, it checks whether there is enough channel to satisfy
bidders request i.e Distinct (N (i)) + di <= k. If true, bidder i is allocated with its requested
channel di. Winner i is charged with the price asked by its critical neighbor multiplied by the total
number of channel requested. A critical neighbor is a bidder who cant share the same channel
[29].

4.2 VERITASs Truthfulness

Figure 4.1 Spectrum utilization, revenue and Bidder satisfaction in Non truthful auction [29].

Figure 4.2 Spectrum utilization, revenue and Bidder satisfaction in VERITAS auction [29].

Following three performance metrics are considered for graphical representation:

Revenue: Total sum charged to all winners.
Spectrum Utilization: It shows the total sum of allocated channels to all winning bidders.

15
 Bidder satisfaction: It shows the percentage of winning bidders. If each bidder is
requesting only a single channel, then satisfaction is given by ratio of spectrum utilization
to total number of bidders. When more than one channels is requested then satisfaction is
given by percentage of bidders who are allocated with one or more channels.

From Figure 4.1 and Figure 4.2, we see that both VERITAS and revenue maximizing auction
improves spectrum utilization as the number of bidders is increased. But when observing the
revenue statistics, we find that revenue generated from non truthful auction increase with the
number of bidder. This is because revenue maximizing auction charge winner with the actual bid
price. On the other hand, revenue of VERITAS decreases after certain threshold. VERITAS
charge winner based on the price of its critical neighbor. Pool of loser decrease as the number of
auctioned channel is increased. In this regard, VERITAS provides flexibility for auctioneer to
select suitable amount of channel for auction to maximize revenue [29].

4.3 VERITASs Efficiency

Figure 4.3 Spectrum utilization of VERITAS vs. simple truthful design (8 channels auction to 10-500
bidders) [29].

Figure 4.4 Spectrum utilization of VERITAS vs. simple truthful design (1-30 channels auction to 300
bidders) [29].

16
As seen from Figure 4.3 and Figure 4.4, for same number of auctioned channels and bidders,
VERITAS achieved higher spectrum utilization compared to simple truthful auction. It is
because simple truthful auction discards frequency reuse of spectrum to achieve which results in
the waste of the spectrum. VERITAS on the other hand exploits the reusability nature of
spectrum to increase spectrum utilization [29].
Best greedy algorithms have the best known polynomial time for spectrum allocation but they
are not truthful .From the experimental results in Figure 4.1 and 4.2 we have seen that VERITAS
achieves truthfulness with comparable spectrum allocation efficiency like the best greedy
algorithm(non truthful) on market. Again comparing with the simple truthful allocation scheme,
VERITAS is highly efficient as shown in Figure 4.3 and Figure 4.4.From the economic
perspective, the major obstacle is the overhead taken by both auctioneer and bidders to avoid
market manipulation. In this regard, VERITAS removes the obstacles due to market
manipulation through its truthful allocation schemes. Hence, VERITAS can function as the new
eBay like online market place for spectrum auctions, allowing bidders to obtain spectrum on
demand and auctioneer processing the request immediately. VERITAS with its truthful and
efficient allocation promotes a win-win situation for both the auctioneers as well as bidders in
spectrum market [29].

17
5 Conclusion
Current spectrum allocation policy is the root cause for spectrum underutilization that has lead to
physical scarcity. Findings from various measurement campaign [4] [5], show presence of large
number of spectrum holes and white space but spectrum access barriers set by the legacy
regulatory bodies prevent from its utilization.

Dynamic Spectrum access is a solution to break these spectrum access barriers set by legacy
spectrum allocation policies. This notion of DSA has also gained support after the FCCs
Spectrum Policy Task Force (SPTF) report which recommends a need to shift from traditional
command-and-control spectrum management to a more dynamic approach taking into account
the rapid innovation in wireless technologies [4]. The need for shift towards the Dynamic
spectrum access paradigm and the advance in the digital communication system has lead towards
the idea of Cognitive radio (CR). Spectrum sensing is the most important attribute of cognitive
radio classified under three categories: non-cooperative sensing, cooperative sensing and
interference based sensing [11].

Most of the research on DSA has been targeted towards open sharing and shared user model of
uncoordinated access of spectrum [25] [26] such as DARPA XG [27]. But protocol complexities
and hardware constraints limit its initial implementation only among homogenous network [25].
In this scenario, transition from current static spectrum allocation towards pragmatic coordinated
dynamic spectrum access would be an attractive solution in the near future [28]. Coordinated
DSA is realized through the introduction of new Coordinated Access Band (CAB), Statistical
Multiplexing of Spectrum access in CAB and DIMSUMnet network architecture [28].

Suitable auctioning policy for DSA is needed that can guarantee both fairness in spectrum access
as well as optimize the spectrum usage. Truthful and efficient auctioning policy specially
designed for dynamic spectrum access like VERITAS could be the solution. VERITAS
maintains truthfulness by charging bidders with the price of its critical neighbor. This helps
eliminate bad practices of market manipulation. VERITAS also address the issue of efficient
spectrum utilization by considering the spectrum reuse factor in its algorithm [29].

18
Bibliography
[1] M. Cave et al., Essentials of Modern Spectrum Management, 1st ed. New York, USA: Cambridge
University Press, Aug. 2007.
[2] Linda E. Doyle, "A Cognitive Radio world," in Essentials of Cognitive Radio, William Webb, Ed.
New York, USA: The CambridgeWireless Essentials Series, 2009, ch. 1, pp. 1-31.
[3] National Telecommunications & Information Administration. (2011, August) ntia.doc.gov. [Online].
http://www.ntia.doc.gov/files/ntia/publications/spectrum_wall_chart_aug2011.pdf
[4] FCC Spectrum Policy Task Force, "Report of the spectrum efficiency working group," FCC, ET
Docket No. 02- 135 , Nov 2002.
[5] M. McHenry, "Dupont Circle Spectrum Utilization During Peak Hours," The Shared Spectrum
Company, 2002.
[6] Milind M. Buddhikot, "Understanding Dynamic Spectrum Access: Models," in 2nd IEEE Int. Symp.
on New Frontiers in Dynamic Spectrum Access Networks, 2007. DySPAN 2007., Dublin, April 2007,
pp. 649-663.
[7] M. Dongegan. (2007, march) Spectrum up for Grabs in Europe. [Online].
http://www.lightreading.com/spectrum-up-for-grabs-in-europe/d/d-id/639361
[8] D.N Hatfield and P.J. Weiser, "Property rights in spectrum: taking the next step," in 1st IEEE Int.
Symp. on New Frontiers in Dynamic Spectrum Access Networks, 2005. DySPAN 2005., Baltimore,
MD, USA, Nov. 2005, pp. 43-55.
[9] Lin Xu et al., "DRiVE-ing to the Internet: Dynamic Radio for IP services in Vehicular
Environments," in Proc. 25th Annu. IEEE Conf. on Local Computer Networks, 2000. LCN 2000.,
Tampa, FL, Nov 2000, pp. 281-289.
[10] Qing Zhao and B. M. Sadler, "A Survey of Dynamic Spectrum Access," Signal Processing
Magazine, IEEE, vol. 24, no. 3, pp. 79-89, May 2007.
[11] Ekram Hossain et al., "Introduction to cognitive radio," in Dynamic Spectrum Access and
Management in Cognitive Radio Networks. New York, USA: Cambridge University Press, 2009, ch.
2, pp. 39-73.
[12] W. Lehr and J. Crowcroft, "Managing shared access to a spectrum commons," in 1st IEEE Int. Symp.
on New Frontiers in Dynamic Spectrum Access Networks, 2005. DySPAN 2005., Baltimore, MD,
USA, Nov 2005, pp. 420-444.
[13] J. Mitola and G.Q. Maguire, "Cognitive radio: making software radios more personal," Personal
Communications, IEEE, vol. 6, no. 4, pp. 13-18, Aug 1999.
[14] J. Mitola, "Cognitive radio for flexible mobile multimedia communications," in IEEE Int. Workshop
on Mobile Multimedia Communications, 1999. (MoMuC '99), San Diego, CA, Nov. 1999, pp. 3-10.
[15] J. Mitola, "Cognitive Radio: An Integrated Agent Architecture for Software Defined Radio,"
Computer Communication System Laboratory, Department of Teleinformatics, Royal Institute of
Technology (KTH), Stockholm, Sweden, Ph.D. dissertation May 2000.
[16] Federal Communications Commission, "Notice of proposed rule making and order: Facilitating
opportunities for flexible, efficient, and reliable spectrum use employing cognitive radio
technologies," FCC, Washington, D.C, FCC 03-322, Dec 2003.

19
[17] Radiocommunication Study Groups, "Cognitive Radio Systems [(CRS) applications] in the land
mobile service," ITU, Geneva, Working document towards a preliminary draft new Report ITU-R
[LMS.CRS2], 2004.
[18] Alexander M. Wyglinski et al., "When radio meet software," in Cognitive Radio Communications
and Networks Principles and Practice.: ELSEVIER Inc., 2010, ch. 1, pp. 1-10.
[19] T. Yucke and H.Arslan, "A Survey of Spectrum Sensing Algorithms for Cognitive Radio
Applications," Communications Surveys & Tutorials, IEEE, vol. 11, no. 1, pp. 116-130, Mar. 2009.
[20] Alexander M. Wyglinski et al., "Spectrum sensing and identification," in Cognitive Radio
Communications and Networks.: ELSEVIER Inc., 2010, ch. 4, pp. 85-95.
[21] J.Proakis, Digital Communications, 5th ed.: McGraw-Hill, Nov 2007.
[22] H. Tang, "Some physical layer issues of wide-band cognitive radio systems," in 1st IEEE Int. Symp.
on New Frontiers in Dynamic Spectrum Access Networks. DySPAN 2005. , Baltimore, MD, USA,
Nov. 2005, pp. 151-159.
[23] Fehske A. et al., "A new approach to signal classification using spectral correlation and neural
networks," in 1st IEEE Int. Symp. on New Frontiers in Dynamic Spectrum Access Networks, 2005.
DySPAN 2005., Baltimore, MD, USA, 2005, pp. 144-150.
[24] N.S Shankar et al., "Spectrum agile radios: utilization and sensing architecture," in 1st IEEE Int.
Symp. on New Frontiers in Dynamic Spectrum Access Networks, 2005. DySPAN 2005., Baltimore,
MD, USA, Nov. 2005, pp. 160-169.
[25] W.D. Horne, "Adaptive Spectrum Access: Using the Full Spectrum Space," in Proc. Int. Symp.
Advanced Radio Technologies, Mar. 2003.
[26] W. Lehr et al., "Software Radio: Implications for Wireless Services, Industry Structure, and Public
Policy," Massachusetts Institute of Technology , 2003.
[27] DARPA neXt Generation (XG) Program. [Online].
http://www.darpa.mil/ato/programs/XG/index.htm.
[28] M.M. Buddhikot et al., "DIMSUMNet: New Directions in Wireless Networking Using Coordinated
Dynamic Spectrum Access," in 6th IEEE Int.l Symp. on a World of Wireless Mobile and Multimedia
Networks, 2005. WoWMoM 2005., Jun. 2005, pp. 78-85.
[29] Xia Zhou et al., "eBay in the Sky:Strategy-Proof Wireless Spectrum Auctions," in Proc. of the 14th
ACM int. conf. on Mobile computing and networking, New York, USA, 2008, pp. 2-13.
[30] W. Vickery, "Counterspeculation, auctions and competitive sealed tenders," Journal of Finance, vol.
16, no. 1, pp. 8-37, Mar. 1961.
[31] Theodore Groves, "Incentive in terms," Econometrica, vol. 41, no. 4, pp. 617-631, July 1973.
[32] Kothari et al., "Approximately strategyproof and tractable multi-unit auctions," in Proc. of the 4th
ACM conference on Electronic Commerce, San Diego, California, USA, Jun. 2003, pp. 166-175.
[33] Daniel Lehmann and L. Shoham, "Truth revelation in approximately efficient combinatorial
auctions," Journal of the ACM, vol. 49, no. 5, pp. 1-26, Oct. 2002.

20