![Illustration of the IEEE 802.22 network model used in the evaluation; [7, Fig. 1], [6, Fig. 1.1].](https://www.researchgate.net/profile/Pal-Gronsund/publication/258509804/figure/fig1/AS:669973202616335@1536745467619/Illustration-of-the-IEEE-80222-network-model-used-in-the-evaluation-7-Fig-1-6-1_Q320.jpg)
Content uploaded by Pal Gronsund
Author content
All content in this area was uploaded by Pal Gronsund on Feb 02, 2015
Content may be subject to copyright.
Sensing of Wireless Microphones in IEEE 802.22:
A System Level Performance Evaluation
P˚
al Grønsund∗†, Przemysław Pawełczak‡,JihoonPark
§,andDanijelaCabric
¶
∗Telenor, Snarøyveien 30, 1331 Fornebu, Norway
†University of Oslo, Gaustadallen 23, 0373 Oslo, Norway
‡Delft University of Technology, Mekelweg 4, 2628 CD, Delft, The Netherlands
§Samsung Electronics Co., Maetan 3-dong, Yeongtong-gu, Suwon-si, Gyeonggi-do, South Korea, 443-742
¶University of California, Los Angeles, 56-125B EngineeringIVBuilding,LosAngeles,CA90095-1594,USA
Email:∗pal.gronsund@telenor.com, ‡p.pawelczak@tudelft.nl,§jpark0@samsung.com,¶danijela@ee.ucla.edu
Abstract—We present results on the system level performance
of the IEEE 802.22 standard with sensing functionality, using
a highly detailed implementation of the IEEE 802.22 protocol
stack in the NS-2 simulator. Our attention is focused on the
effect of spatio-temporal wireless microphone (WM) activity on
the performance of the IEEE 802.22 network with spectrum
sensing considered. In general we find that the frequency of
WM appearance and activity duration should be quite high in
all channels not used by TV broadcasters to reduce IEEE 802.22
throughput, for example about 50% WM occupancy in each of
total of four channels. Impact on WM performance is found to
be low in general using the two-stage spectrum sensing strategy
with frequent sensing stages.
I. INTRODUCTION
Anumberofstandardizationandregulatorybodieswork
towards a better exploitation of the TV white spaces [1].
The IEEE 802.22 [2] standard published in 2011 is the first,
most sophisticated and one of the most discussed standards
for operation in TV white spaces [3]. In short, the system
specified in the IEEE 802.22 standard utilizes TV white spaces
to provide a fixed broadband service in the rural areas.
The IEEE 802.22 devices use geo-location and communi-
cate with a database to obtain information about available fre-
quencies and allowed transmit power levels at their locations.
In addition the IEEE 802.22 system can use sensing techniques
to detect sudden appearance of primary users of the TV bands
such as TV transmitters and Wireless Microphones (WMs).
The TV broadcasters update the database about their frequency
usage and transmit power levels at all locations. Other low
power devices operating in these bands such as WMs might
update the database, but might also appear suddenly at random
locations without notification. Detection and protection of
these WMs are considered as one of the greatest challenges
for the IEEE 802.22 system. By using sensing technologies,
the IEEE 802.22 system should be able to detect WMs and
then cease transmission and switch to a vacant TV channel.
Understanding the complete system level performance of
IEEE 802.22 is complex and difficult to achieve with analytical
models. Therefore, we have developed the most complete
Parts of this work has been supported by the Dutch Technology Foundation
STW under contract 12491.
implementation of the IEEE 802.22 stack in the NS-2 sim-
ulator. This simulation software was used in our other related
works to: (a) evaluate delay at the application layer of IEEE
802.22 [4]; (b) evaluate different QoS profiles (e.g. video
and voice over IP) in IEEE 802.22 [5], (c) evaluate IEEE
802.22 network as a function of Customer Premises Equipment
(CPE) to Base Station (BS) distance (for different power and
modulation levels of IEEE 802.22 CPE and varying WM
power) [6]; and (d) four channel selection techniques IEEE
802.22 based on WM signal characteristics [7].
In this paper we present further detailed system level
performance evaluation of IEEE 802.22. The new contribution
is the detailed study of the impact of temporal activity of WM
on the IEEE 802.22 network.
II. SYSTEM MODEL
For the ease of reading we re-introduce the system model
of our NS-2 IEEE 802.22 implementation, which is also found
and introduced in [4], [5], [6], [7].
A. IEEE 802.22 Network Model
We cons i d e r a n I E E E 8 0 2.22 system limited to o n e B S a n d
asetofCPEsfixedatcertainlocationsinaccordancewiththe
IEEE 802.22 standard. An example network setup is illustrated
in Fig. 1 with four CPEs and four WM Tx-Rx pairs. The
small oval illustrates the coverage area for WM 1a and the big
oval the coverage area for the IEEE 802.22 BS. The simulator
supports all channels in the UHF band, but a subset of channels
are selected for operation in the studied simulation scenarios,
exactly four channels using the frequencies denoted F1–F4. A
personal computer (PC) connected via Ethernet to the BS will
establish links to the CPEs and run traffic models.
In this study we assume that TV transmissions are known
from the database and focus only on WMs as the primary
users. WM activity will be detected only by sensing techniques
with no support from beacon protocol like IEEE 802.22.1 [8].
Finally, the IEEE 802.22 self coexistence protocol is not used
since a single cell is considered.
CPE 3
CPE 2
CPE 1
CPE 4
F1
F3
F2
F1-F4 F4
ddwm
cpe
PC
802.22
BS
WM1a
WM4a
WM3aWM2a
Ethernet
WM4b
WM1b
WM2b WM3b
Fig. 1. Illustration of the IEEE 802.22 network model used in the evaluation;
[7, Fig. 1], [6, Fig. 1.1].
B. Traffic Model
1) IEEE 802.22 Traffic Model: The traffic model variable
bit rate (VBR) is used to measure the maximum IEEE 802.22
system throughput. VBR will run over UDP where maximum
rates will be simulated by constantly transmitting UDP packets
to each CPE of size 1500 Bytes every 0.33 ms, which amounts
to a bit rate higher than the channel capacity. VBR will use
the best effort (BE) QoS traffic profile in IEEE 802.22 which
provides fairness between the CPEs using the BE profile.
2) Wireless Microphone Traffic Model: When the WM
becomes active, its traffic pattern is characterized by a 100%
duty cycle (irrespective whether someone is speaking to the
microphone or not) until the WM disappears. It should be
noted that the NS-2 simulator is packet transmission-based,
hence we assume that WM virtually transmits data packets
consecutively during the whole IEEE 802.22 OFDMA frame
in order to simulate the 100% duty cycle traffic pattern.
We are not aware of any statistics on WM traffic distribu-
tion, hence we assume that all WMs generate new connections
according to the negative exponential distribution for the
average inter-arrival time 1/λwand average occupancy time
1/µwwhich is common in wireless communications [9].
III. SIMULATION MODEL
Again, for the ease of exposition we re-introduce the system
model of our NS-2 IEEE 802.22 implementation, which is also
found in [4], [5], [6], [7]. In this study we adapt an extensive
implementation of IEEE 802.16e in NS-2 [10] developed
for the WiMAX Forum [11] to implement the IEEE 802.22
standard. The features that are different from IEEE 802.16 are
implemented and conform to the best extent to the functional
requirements as specified in the IEEE 802.22 standard. Note
that the main parameters of our IEEE 802.22 NS-2 simulator,
in relation to IEEE 802.22 standard, are presented in [5, Table
II], [4, Table IV], [6, Table 8.1], and [7, Table I].
A. OFDMA and Channelization Structure
Both the IEEE 802.22 downlink and uplink subframes have
totally 60 subchannels, where each subchannel consists of 28
subcarriers out of which 24 data and 4 pilot subcarriers. The
IEEE 802.22 network can operate on any vacant TV channel
not used by the primary user. Channel bonding of scattered
available channels is not considered in this study, therefore
only one available UHF channel will be used by the IEEE
802.22 system at any time. A WM occupies only one channel.
In our IEEE 802.22 NS-2 implementation only the 6 MHz
profile is used. For the subcarrier allocation strategy, partially
used subcarrier (PUSC) allocation [12] is used. Guard bands
are considered at both ends of the channel bandwidths with a
total of 368 guard and null subcarriers.
The transmit/receive transition gap (TTG) is set to 210 µs,
which supports a 30 km distance between BS and CPE. A
dynamic TTG is needed for greater distances, however this
is not implemented since the simulation scenarios considered
involve only small distances. The receive/transmit transition
gap (RTG) is set to 81.8 µs. There are totally 26 symbols,
each of 373.33 µsduration.TheDL:ULratioissetto2:1.
B. OFDMA Traffic Scheduling
The MAC layer of IEEE 802.22 uses linear scheduling
to allocate OFDMA slots to traffic from the upper layers
in both the downlink and uplink subframes, as opposed to
arectangularschedulinginIEEE802.16.AnOFDMAslot
can be characterized as a {subchannel, symbol}-tuple in the
frequency and time domain. Vertical striping is used for both
DL and UL subframes in the simulator, which means that
OFDMA slots are allocated in frequency first and then in time,
i.e. the first symbol is filled with data before the next symbol.
C. Propagation Model
The propagation model used in the simulator is the COST
Hata [13] path loss model, configured for suburban scenarios.
Further, the COST Hata model is combined with a Clarke-
Gans [14] implementation of Rayleigh fading, which has been
extended to support 2048 subcarriers for IEEE 802.22. The
Veh i cu l a r A IT U p o w er d e lay s p re a d m od e l [15 ] i s us e d w hi c h
is suited for the considered IEEE 802.22 scenario with large
cell in suburban areas and a tall BS antenna. Both the BS and
CPEs are configured with 36 dBm transmit power. Antenna
gains for both the BS and CPE are set to 0 dB. Dynamic
transmit power adaptation is not implemented.
Interference modeling is done at the subcarrier level by
capturing packets from all IEEE 802.22 nodes and WMs.
When the received signal to interference plus noise ratio
(SINR) on each subcarrier is calculated for each packet, a
decision is made to further process or drop the packet. This
is done by first finding the exponential effective SIR mapping
(EESM) [16] to get the effective SINR and then extracting
the block error rate (BLER) from the SINR, modulation and
coding rate and block size. Based on the BLER value a
decision is made whether to drop the packet or not1.
D. Error Protection
Channel errors are considered in the simulations. For error
protection the IEEE 802.22 standard uses ARQ. For error
correction coding the IEEE 802.22 uses convolutional turbo
codes (CTC) and block turbo code (BTC), although only ARQ
1Please refer to [10] for a detailed description of of the OFDMA physical
layer implementation and interference modeling. Note that the propagation
model, operating frequency range and system profiles are reimplemented to
fit the UHF bands, IEEE 802.22 and suburban scenarios.
is implemented in NS-2. However, if BLER found as described
in Section III-C is above a threshold set to 4% (recommended
by the WiMAX Forum IEEE 802.16e implementation [11])
the simulator emulates that the erroneous bits are corrected.
E. Wireless Microphone Implementation
A typical analog WM in the TV bands uses a narrow
bandwidth of 200 kHz, which amount to 68 active subcarriers
in the NS-2 OFDMA simulator. Since an analogue WM most
of the time focuses transmit power on a narrow part of the
200 kHz bandwidth, we assume that the WM on average uses
half the bandwidth with 34 subcarriers when implementing
the WM in the NS-2 simulator. In this case, if the WM
transmits with 200 mW, the transmit power per subcarrier for
the WM is set to 0.2 W/34=0.0059 W in the NS-2 simulator.
The WM implementation will transmit consecutively in both
directions to simulate a realistic WM with 100% duty cycle.
The modulation and coding rate is QPSK 1/2.
F. Wireless Microphone Detection Process
The main functions involved in primary user detection in
the IEEE 802.22 standard are the spectrum manager at the BS
and the spectrum automaton at the CPE which controls the
spectrum sensing and the geo-location function. All functions
are implemented in NS-2, except for geo-location (due to
the assumption of TV broadcast absence in the considered
channel) and dynamic transmit power control. Since the latter
is not implemented, the IEEE 802.22 will switch to a new
channel if a WM is detected on that channel. If no channels
are available IEEE 802.22 will cease transmission.
G. Spectrum Manager
The spectrum manager is implemented in the BS and is
responsible for deciding which channel to use. It specifies the
set of channel lists, i.e. the backup, candidate and protected
channel lists. In the IEEE 802.22 standard a channel will get
status as backup channel when sensed as unused every six
seconds over a period of 30 seconds. An algorithm can be
applied to optimize which of the backup channels will be
selected as the operating one, which in our NS-2 simulator
will be the first available channel in the list.
In the CPE, the spectrum automaton is a lightweight version
of the spectrum manager in the BS. The spectrum automaton
is controlled by the BS and is mostly responsible for reporting
information to the spectrum manager. In rare cases, the spec-
trum automaton itself is responsible for sensing at initial CPE
power on, when it looses contact with the BS and during an
idle time when there are no tasks pending.
1) Spectrum Sensing Implementation: The spectrum sens-
ing function can first be classified into in-band sensing, that
senses the operating channel, and out-of-band sensing, that
senses activity on channels that potentially can be used by
the IEEE 802.22 system. For in-band sensing, the two-stage
spectrum sensing approach, as specified by the standard, is
implemented in the simulator. At the coarse sensing stage (first
stage) a simple energy detection is used for frequent and short
sensing periods, tc=1ms in the simulator. If coarse sensing
detects a WM signal, it switches to the fine sensing stage
(second stage) that uses a more detailed WM detection process
for a longer period, ts=30ms in the simulator (spanning
three OFDMA frames). A simple energy detector is also used
for fine sensing in the simulator. Coarse sensing is carried out
during allocated time periods at the end of the uplink OFDMA
subframe, and occurs at every second OFDMA frame. If a WM
signal is detected by fine sensing then the operating channel
is switched to one of the backup channels.
Probability of detection pdand probability of false alarm
pfare set in the simulation separately for coarse and fine
sensing. In the simulation we were able to manipulate pdand
pfin order to simulate performance with different sensing
constraints. The effect of pdand pfon the IEEE 802.22
will mostly be considered for coarse sensing which is more
unreliable than fine sensing. Fine sensing that uses more
advanced sensing techniques is considered to be more accurate
and pdand pfare therefore set to one and zero, respectively.
Further, cooperative sensing with the OR rule is implemented
for all sensing stages, which is mandatory in the US as
specified in the IEEE 802.22 standard [2, Sec. 8.6.1.3].
Finally, in IEEE 802.22, and our NS-2 implementation, out-
of-band sensing is performed during quiet periods allocatedfor
the fine sensing periods, and all relevant channels are sensed
during one fine sensing period.
IV. SYSTEM PERFORMANCE METRICS
Just like in [4], [5], [6], [7] different metrics are used to eval-
uate the performance of the IEEE 802.22 standard depending
on the studied scenario. The metrics used to evaluate IEEE
802.22 performance are:
1) Throughput: measured at the transport layer for the
actual application used. This will not reflect the physical
layer throughput which includes management frames for
the MAC layer and the general network layer overhead.
2) Packet Loss: measured at the network layer as the
percentage of packets transmitted but not received.
3) SINR: Signal to Interference plus Noise Ratio at the CPE
resulting from Gaussian noise and interference from
WMs.
Furthermore, to assess the impact on the WM performance we
use the following metrics:
1) WM C/I: is the carrier to interference ratio measured
at the WM, which is considered as the metric that
best describes the performance of the analog WM.
Noise is assumed to be Gaussian and therefore the only
interference will be from the IEEE 802.22 BS in the
downlink and CPEs in the uplink.
All metrics are measured locally at the actual device, i.e. at
the receiving CPE for throughput, packet loss and SINR, and
at the WM for C/I. The metrics will mostly be presented as
one average value for all CPEs and WMs.
TABL E I
PARAMETER VALUES FOR THE SIMULATION SCENARIOS;COMPARE
WITH [5, TABLE III]
Parameter BS CPE WM
Nodes 144pairs
Location Fixed Random Random
Height (m) 30 5 1.5
Transmit Power (W) 440.2
Antenna Gain (dBi) 000
Modulation/FEC 16-QAM 1/2 16-QAM 1/2 QPSK 1/2
Prob. of false alarm (pf)0.01 0.01 —
Prob. of detection (pd)0.99 0.99 —
Channel bandwidth (MHz) 660.2
Channels used (MHz) 575–593 575–593 575–593
V. P ERFORMANCE EVALUATION:IMPACT OF THE
TEMPORAL ACTIVITY OF WMSONTHEIEEE 802.22
NETWORK
The specific values for the simulation scenarios are summa-
rized in Table I. In the IEEE 802.22 network as illustrated in
Fig. 1, totally 4 CPEs are connected to 1 BS which applies
to a sparsely populated area. In this area there are totally
4channelsof6MHznotusedbytheTVbroadcastersand
therefore available for the IEEE 802.22 network. Furthermore,
there are 4 WM pairs that each appear randomly on one of the
4channelsseparately,hencetheIEEE802.22networkhasto
switch channel when a WM appears on the operating channel.
The distance between two WMs in a WM pairs is set to 150 m.
Note that some parameters will be changed for the simulation
scenarios, but this will be stated clearly.
Each simulation is run 15 times, each with a duration of
500 s, and the results are averaged. A warm up time of 20 s
is used to ensure that a stable point of network operation is
reached. Considering that in NS-2 all nodes will receive pack-
ets from all other nodes within detection range, irrespective of
actual frequency used, and that the nodes processes the packet
fully or partially, each single simulation takes about 45 minutes
on a modern computing cluster2.
The 4 WM pairs appear on separate channels randomly
within a radius of 1 km. Also, each CPE is located randomly
within 1 km from the BS. The short radius used in this scenario
is to assure that WMs are detected according to the pdand pf.
For the spectrum sensing, as a special case, only the BS does
two-stage spectrum sensing with tc=1ms and ts=30ms.
In reality, WMs appear very infrequently and operate for
short durations, i.e. in the order of hours. To be able to simulate
the WM activity we scale the realistic 1/λwand 1/µwto
values resulting in short simulation time. The scaling will not
impact the physical layer performance metrics, but will impact
on the higher layer metrics throughput and packet loss for very
low values of 1/λwand 1/µwwhen 1/λw<1/µw. This will
result in lower bound values for the metrics of interest. In most
simulations, however, this will not be an issue.
2) Throughput: Averag e a g grega t e t hroughp u t f or the CPEs
versus increasing WM inter-arrival times in the range 1/λw=
2IBM cluster with dual quad core compute nodes Xeon L5420 / 2.5 GHz
processors, each with 8 cores, 8 GB of memory and operating system linux
Ubuntu Hardy Heron.
1,...,10 sfortheoccupancytimes1/µw={2,4,6}sisgiven
in Fig. 2(a). A first observation is that maximum throughput
achieved in the downlink is 2.76Mbps, which is in accordance
with the standard when considering overhead from higher
layer protocols and sensing. A saturation in throughput is
observed when inter-arrival time (1/λw)increasesbecausethe
probability of finding an opportunity to transmit in all channels
increases. It is also observed that 1/µw=2ssaturates
before 1/µw=4sand1/µw=6swhichisbecausethe
lower the occupancy time (1/µw), the positive value from
1/λw−1/µwbecomes higher, hence the probability of finding
an opportunity to transmit increases. It can also be observed
that the lower the occupancy time (1/µw), the higher the
throughput, which especially is visible for the lower inter-
arrival times where 1/µw<1/λw.Thisisalsothereasonfor
the observed distance between the curves where throughput
increases when the WM occupancy times (1/µw)increases.
As an overall observation, the frequency of WM appearance
and activity duration should be quite high in all channels not
used by the TV broadcaster to reduce IEEE 802.22 throughput.
3) Packet Loss: It can be seen that average packet loss
for the CPEs given in Fig. 2(b) is higher for lower inter-
arrival times (1/λw)andespeciallyforhigheroccupancytimes
(1/µw). When 1/λw"1/µwthe probability that the IEEE
802.22 system will find an opportunity to transmit is extremely
low. Then, when the IEEE 802.22 system first transmits the
probability that a WM appears on that channel is extremely
high. Consequently, the percentage of packet loss increases
considerably due to interference from the WM.
4) SINR: It can be seen that average SINR for the CPEs
presented in Fig. 2(c) increases as WM inter-arrival time
increases. Lower SINR is seen for lower inter-arrival times
because the actual time the IEEE 802.22 system is able to
access the channel decreases. Hence, the number of packets
that experience interference from the WM in proportion to
the total number of packets received increases. This can
also be illustrated by the average percentage of packets that
obtain SINR below the modulation and coding threshold (MC-
Threshold) for the IEEE 802.22 CPEs presented in Fig. 2(d),
indicating harmful interference to the IEEE 802.22 CPE,
which increases for lower WM inter-arrival times. We note
that the number of simulations should be higher to obtain
more accurate SINR result due to the random CPE location
within the cell. However, a notable increase in SINR is seen
in Fig. 2(c) for 1/µw=2sat1/λw=4s, for 1/µw=4sat
1/λw=5sandfor1/µw=6sat1/λw=6s.
5) Impact on WMs: The percentage of time the WM expe-
rience C/I values below the required C/I level of 20 dB [17]
given in Fig. 2(e) show that the higher the inter-arrival time
(1/λw), the more interference is caused to the WMs. This is
because the probability that the IEEE 802.22 system transmits
when a WM appears is higher the more 1/λw>1/µw.
However, we find that the interference is only caused at the
instantaneous time when the WM appears, and in general it
can be observed that the percentage of time the WM drops
below the required C/I level is very low. The average WM C/I
1 2 3 4 5 6 7 8 9 10
0
0.5
1
1.5
2
2.5
3
WM inter−arrival time, 1/λ (s)
Throughput (Mbps)
1/µ=2
1/µ=4
1/µ=6
(a) Aggregate IEEE 802.22 Throughput
1 2 3 4 5 6 7 8 9 10
0
10
20
30
40
50
60
70
80
90
100
WM inter−arrival time, 1/λ (s)
Packet Loss (%)
1/µ=2
1/µ=4
1/µ=6
(b) CPE Packet Loss
1 2 3 4 5 6 7 8 9 10
22
23
24
25
26
27
28
WM inter−arrival time, 1/λ (s)
SINR (dB)
1/µ=2
1/µ=4
1/µ=6
(c) CPE SINR
1 2 3 4 5 6 7 8 9 10
0
0.2
0.4
0.6
0.8
1
1.2
1.4 x 10−4
WM inter−arrival time, 1/λ (s)
SINR < MCThreshold (%)
1/µ=2
1/µ=4
1/µ=6
(d) CPE SINR <MCThreshold (%)
1 2 3 4 5 6 7 8 9 10
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2x 10−3
WM inter−arrival time, 1/λ (s)
C/I < 20dB (%)
1/µ=2
1/µ=4
1/µ=6
(e) WM C/I <20 dB (%)
1 2 3 4 5 6 7 8 9 10
38.84
38.85
38.86
38.87
38.88
38.89
38.9
38.91
38.92
38.93
38.94
WM inter−arrival time, 1/λ (s)
C/I (dB)
1/µ=2
1/µ=4
1/µ=6
(f) WM C/I (dB)
Fig. 2. IEEE 802.22 performance for various WM inter-arrival times, 1/λw={1,...,10}s, and average occupancy times 1/µw={2,4,6}s.
presented in Fig. 2(f) also confirms this.
VI. CONCLUSIONS
In this paper we have studied the network level performance
of IEEE 802.22 with sensing functionality by means of an
extensive network level simulations implemented in NS-2.
We studi e d t h e t h r o u ghput for diff e r e n t a c tivity levels of
wireless microphones (WMs) in channels not occupied by TV
broadcasters, and found that the WM activity level should be
quite high in all channels to reduce IEEE 802.22 throughput.
For example, about 50% WM occupancy in each of total of
four channels to reduce throughput remarkably. Impact on WM
performance was found to be low in general using the two-
stage spectrum sensing strategy with frequent sensing stages.
REFERENCES
[1] Y.-C. Liang, A. T. Hoang, and H.-H. Chen, “Cognitive radioonTV
bands: A new approach to prove wireless connectivity for rural areas,”
IEEE Wireless Commun. Mag., vol. 15, no. 3, pp. 16–22, Jun. 2008.
[2] IEEE Standard for Information Technology–Telecommunications and
information exchange between systems Wireless Regional Area Net-
works (WRAN)–Specific requirements Part 22: Cognitive Wireless RAN
Medium Access Control (MAC) and Physical Layer (PHY) Specifica-
tions: Policies and Procedures for Operation in the TV Bands, IEEE
IEEE Std 802.22-2011, Jul. 2011.
[3] C. R. Stevenson, G. Chouinard, Z. Lei, W. Hu, S. J. Shellhammer, and
W. Cal d w e ll, “I E E E 8 0 2.22 : T h e fi r st cog n i t ive r a d i o wir e l e s sregional
area network standard,” IEEE Commun. Mag., vol. 47, no. 1, pp. 130–
138, Jan. 2009.
[4] J. Park, P. Pawełczak, P. Grønsund, and D. ˇ
Cabri´
c, “Analysis framework
for opportunistic spectrum OFDMA and its application to IEEE 802.22
standard,” IEEE Trans. Veh. Technol., vol. 61, no. 5, pp. 2271–2293,
Jun. 2012.
[5] P. Grønsund, P. Pawełczak, J. Park, and D. ˇ
Cabri´
c, “System level
performance of IEEE 802.22 with sensing-based detection of wireless
microphones,” 2012, submitted to IEEE Commun. Mag.
[6] P. Pawełczak, J. Park, P. Grønsund, and D. ˇ
Cabri´
c, “System level
analysis of OFDMA-based networks in TV white spaces: IEEE 802.22
case study,” in TV White Space Spectrum Technologies: Regulations,
Standards, and Applications, R. A. Saeed and S. J. Shellhammer, Eds.
CRC Press, 2011.
[7] P. Grønsund, P. E. Engelstad, P. Pawełczak, O. Grøndalen,P.H.Lehne,
and D. Cabric, “Spectrum selection in cognitive radio networks with
spectrum sensing aided long-term spectrum management,” submitted to
IFIP Networking 2013.
[8] IEEE Draft Standard for Information technology Telecommunications
and information exchange between systems Local and metropolitan area
networks Specific requirements Part 22.1: Standard to Enhance Harmful
Interference Protection for Low Power Licensed Devices Operating in
TV Broadcast Bands,IEEEStd.P802.22.1Draft8,Jul.2010.
[9] M. Wellens, J. Riihij¨
arvi, and P. M¨
ah¨
onen, “Empirical time and fre-
quency domain models of spectrum use,” Elsevier Physical Communi-
cation Journal,vol.2,no.1–2,pp.10–32,Mar.–Jun.2009.
[10] N. Golmie, R. Rouil, D. Doria, X. Guo, R. Iyengar, S. Kalyanaraman,
S. Krishnaiyer, S. Mishra, B. Sikdar, R. Jain, R. Patneyand, C. So-In,
and S. Parekh. (2009) WiMAX forum system level simulator NS-2
MAC+PHY add-on for WiMAX (IEEE 802.16). [Online]. Available:
http://code.google.com/p/ns2-wimax- awg/
[11] WiMAX forum website. [Online]. Available: http://www.wimaxforum.
org/
[12] (2006, August) Mobile WiMAX–part 1: A technical overview and
performance evaluation. [Online]. Available: http://www.wimaxforum.
org/resources/documents/marketing/whitepapers
[13] “COST 231. digital mobile radio towards future generation systems.
final report,” E. Demosso, Ed. Brussels, Belgium, EU: European
Commission, 1999. [Online]. Available: http://www.lx.it.pt/cost231
[14] R. H. Clarke, “A statistical theory of mobile-radio reception,” Bell
System Technical Journal, vol. 47, pp. 957–1000, 1968.
[15] “Guidelines for evaluations of radio transmission technologies for IMT-
2000,” ITU ITU-R, Tech. Rep. M.1225, 1997.
[16] “Effective SIR computation for OFDM system-level simulations,” 3GPP-
TSG-RAN-1, Tech. Rep. RI-03-1370, 2003.
[17] “Technical and operational requirements for the possible operation of
cognitive radio systems in the ‘white spaces’ of the frequencyband
470-790 MHz,” European Communication Committee, Tech. Rep. ECC
Report 159, Jan. 2011.
