Upload
jongman
View
218
Download
2
Embed Size (px)
Citation preview
Inter-cell Transmit Power Adaptation Algorithm forCoexistence of White Space Applications
Seungil Yoon1, Jongmin Park2, Kyutae Lim1, and Jongman Kim3
1 Georgia Electronic Design Center, School of Electrical and Computer Engineering,
Georgia Institute of Technology, Atlanta, Georgia 30332, Email: {syoon7, ktlim}@ece.gatech.edu2 Qualcomm Inc., San Diego, CA 92121, Email: [email protected]
3 KORUS Research Center for Informersive Systems, School of Electrical and Computer Engineering,
Georgia Institute of Technology, Atlanta, Georgia 30332, Email: [email protected]
Abstract—White space, the unused local TV channels, areready to be used for unlicensed wireless devices. Due to itspropagation characteristics, availability of white space channelscan be limited in a densely populated area. This paper presentsan inter-cell transmit power adaptation (ITPA) algorithm thatenhances the decision procedure of a negotiation-based transmitpower control (TPC) in the selection of a shareable channel andthe decision of the new transmit power level of neighboringwhite space applications (WSAs) on the shareable channel.For the evaluation of the algorithm, we conduct theoreticalanalysis and simulation and prove that the ITPA algorithm canassist the newly-joining WSAs to find shareable channels withneighboring WSAs for the negotiation-based TPC. We expectthat the enhancement in channel assignment contributes to anincrease in the accommodation of newly-joining WSAs althoughall channels are occupied.
Index Terms—white space, inter-cell transmit power control,lower power white space applications, coexistence.
I. INTRODUCTION
In November 2008, Federal Communications Commission
(FCC) approved the rule that allows the use of white space,
unused local TV spectrum, for low power wireless devices un-
less they interfere with incumbent users [1]. Addition of white
space spectrum to mobile devices could extend the availability
of Internet access. However, because of the opportunistic
channel assignment of white space such as dynamic frequency
selection or dynamic frequency sharing according to [2-6],
the number of available white space channels can be limited
in a densely populated area. Due to this insufficiency, the
coexistence of low power unlicensed white space applications
(WSAs) becomes a major issue. Contrary to licensed wireless
systems such as cellular systems that maintain their minimum
transmit power to support the required signal-to-noise ratio
(SNR), WSAs are expected to set up factory-default or FCC
recommended transmit power level. Thus, since WSAs have
a margin to increase or decrease their transmit power on the
condition that they with increased or decreased new transmit
power can support their associated clients, transmit power
control (TPC) is an attractive coexistence approach.
Conventional TPC-based coexistence mechanisms are non-
collaborative mechanisms, where WSAs adjust their transmit
power not to interfere with their neighboring WSAs without
any negotiation of their new transmit power. Although we can
consider to adapt new schemes, introduced in [7] and [8],
WSAs cannot adjust their transmit power precisely with the
absence of collaborative negotiation. For the effective control,
WSAs prefer a collaborative TPC mechanism such as that of
3GPP self-organized network (SON), which operates under the
rules of licensed service providers. However, since low power
WSAs operate in an unlicensed mode, they have no tightly
centralized controllers as the assistant of the collaborative TPC
mechanism. Instead of tightly centralized controllers, we could
consider to use other centralized or distributed controllers,
suggested in [9 -15], but since the controllers have no authority
on channel management, all previous work do not involve the
change of channel configuration such as a current transmit
power. However, we believe that those controllers can assist
heterogeneous white space access points (WSAPs) to negotiate
their new transmit power. Based on this perspective, we sug-
gest inter-cell transmit power adaptation algorithm (ITPA) as
the decision algorithm of the negotiation-based TPC between
WSAPs.
This paper consists of following sections: Section II de-
scribes the proposed ITPA algorithm, and Section III defines
the evaluation metrics for theoretical analysis of the ITPA
algorithm. In addition, Section III presents all equations and
conditions used to evaluate the ITPA algorithm and the evalua-
tion results. Section IV presents the simulation analysis of the
ITPA algorithm-based coexistence mechanism, and Section V
summarizes our work with the future work.
II. INTER-CELL TRANSMIT POWER ADAPTATION
ALGORITHM
To estimate the power margin that neighboring WSAPs can
afford to adjust, the ITPA algorithm requires three calculation
functions: a distance between two WSAPs, the amount of an
increase in a noise floor, and the expected received signal
strength (RSS) of WSAPs at associated clients.
A. Basic Functions
To calculate a distance between two WSAPs, the algorithm
selects a proper path loss model by referring to location and
channel configuration information. We assume that WSAPs
broadcast not only location information such as an address
or a location type – indoor or outdoor – but also channel
The 8th Annual IEEE Consumer Communications and Networking Conference - Wireless Consumer Communication andNetworking
978-1-4244-8790-5/11/$26.00 ©2011 IEEE 487
configuration information such as their current transmit power,
Ptx, the worst path loss between them and their associated
clients, and their target SNR, Psnr. Let us assume that neigh-
boring WSAPs operate on the ith channel, and newly-joining
WSAPs select the ith channel as their operating channel.
Newly-joining WSAPs can estimate the path loss between
them and neighboring WSAPs using Pl = Ptx − Pmr, where
Pmr is the measured RSS of the signals of neighboring
WSAPs at the center of newly-joining WSAPs. With the
location type of neighboring WSAPs, newly-joining WSAPs
can choose a proper propagation model and a fading model and
then calculate the distance r between newly-joining WSAPs
and neighboring WSAPs using the inverse function of the
following equation: Pl = Ploss(r, f), where Ploss is the
function of path loss models, and f is the center frequency.
The signals of newly-joining WSAPs as interference gener-
ators cause an increase in the noise floor (PN ) of neighboring
WSAP as interference victims, and the increase can be calcu-
lated using the following equation [16]:
�P = �P (Per, PN , d) = 10 log(10
Per10 + 10
PN10
)− PN ,
(1)
where Per is the expected RSS of the signals of newly-joining
WSAPs at the d distance from the center of neighboring WS-
APs. Contrary to the fact that we can acquire practically mea-
sured RSS of the signals of neighboring WSAPs at the center
of newly-joining WSAPs, we can only estimate the expected
RSS of the signals of newly-joining WSAPs at the center of
neighboring WSAPs since newly-joining WSAPs do not really
transmit their signals. Instead, we can estimate the expected
RSS with the following equation: Per = Ptx − Ploss(d, f),where Ptx is the expected transmit power of newly-joining
WSAPs.
The expected RSS of the signals of a newly-joining WSAP
measured by clients associated to neighboring WSAPs should
be less than the sum of Psnr of neighboring WSAPs and
the peak-to-average power ratio (PAPR), PPR. Otherwise,
the clients cannot catch the signals of neighboring WSAPs
because of the interference caused by the signals of newly-
joining WSAPs operating on the same channel. At the clients
associated to the newly-joining APs, the expected RSS of
the signals of the neighboring WSAP at the clients should
be less than the sum of Psnr of the newly-joining WSAP
and PPR. However, since newly-joining WSAPs have no
associated clients yet, the algorithm simply regards that newly-
joining WSAPs have the same clients of neighboring WSAPs.
Thus, the expected RSS of the signals of neighboring WSAPs
at virtual clients, associated to newly-joining WSAPs, should
be less than the sum of Psnr of newly-joining WSAPs and
the PAPR. Now, we can define the desired difference, Pdiff ,
with the following equation:
Pdiff = Psnr + PPR, (2)
where we take the PAPR of 10 dB in this paper, and the
ITPA algorithm uses Pdiff to check the availability of desired
signals.
D1
R
R
nAP ngAP
D2
client
D2D3
RnLngLn
Fig. 1. Reference model of the ITPA algorithm
B. ITPA Algorithm Description
Fig. 1 illustrates the extent to which the ITPA algorithm
works. Let a ngAP denote a neighboring WSAP, and let a nAP
denote a newly-joining WSAP. With Prx that is the measured
RSS of the signals of the ngAP at the nAP and Ptx of the
ngAP, the nAP can estimate the distance, D1, between the
center of the nAP and the center of the ngAP. In addition,
with Plrx that is the lowest estimated RSS at the clients of the
ngAP and Ptx of the ngAP, the nAP can also estimate an active
service coverage, D2, from the center of the ngAP to its clients
that reported Plrx. Instead of a desired service coverage, R,
the ITPA algorithm regards D2 as an actual service coverage.
Thus, the expected RSS of the signals of the ngAP at the D2
distance from the center of the ngAP should be greater than
Pmin that also reflects interference caused by the signals of the
nAP. Since Plrx is the worst path loss between the ngAP and
its clients, the algorithm treats an actual distance calculated
with Plrx as the longest distance from the center of the ngAP.
Thus, as depicted in Fig. 1, the strong signals of the ngAP
should reach a location, Lng , where is at the D2 distance
from the center, resulting in providing better QoS to clients
than QoS measured at the client that reported the worst path
loss. However, since we have to tackle the worst case about
the location of clients, the algorithm assumes that Lng locates
on a straight line between the center of the ngAP and the
center of the nAP. Meanwhile, the strong signals of the nAP
should reach a location, Ln, where is at the distance, Rn, from
the center of the nAP. Since the nAP has no real clients, the
algorithm uses a desired service coverage, Rn, to be equal to
D2 of the ngAP. At Ln, the ITPA algorithm estimates �Png
that is the increase in the noise floor of the nAP and Pdiff
of the nAP using (2). The algorithm examines whether the
expected RSS of the signals of the nAP, Pnrx, at Ln is greater
than the sum of Pmin of the nAP and �Png , and If not, the
algorithm declares that the ITPA negotiation cannot proceed
with the ngAP. Otherwise, the algorithm examines whether
the difference of the expected RSS of the signals of the nAP,
Pnrx, and the expected RSS of the ngAP, Png
rx , at Ln is greater
than Pdiff of the nAP. If it is, the ITPA algorithm, which
regards that the nAP can support its potential clients within
a Rn service coverage, includes the channel where the ngAP
operates on in a target list of ITPA negotiation.
488
If the above operation indicates that the nAP can accept
the ITPA-based coexistence, the nAP attempts to investigate
whether the ngAP can accept the activation of the nAP with
revised channel configuration parameters such as new transmit
power of the ngAP on the same channel. At Lng , the ITPA
algorithm estimates �Pn that is the increase in the noise floor
of the ngAP and Pdiff of the ngAP. With them, the algorithm
examines whether the expected RSS of the ngAP, Pngrx , at Lng
is greater than the sum of Pmin of the ngAP and �Pn. If
not, the algorithm declares that the ITPA negotiation cannot
proceed with a given target channel, Ptx of the ngAP, and
Ptx of the nAP. Otherwise, the algorithm examines whether
the difference of the expected RSS of the ngAP, Pngrx , and
the expected RSS of the nAP, Pnrx, at Lng is greater than
Pdiff of the ngAP. If it is, the ITPA algorithm regards that the
ngAP can still serve its clients within the D2 service coverage.
Thus, the algorithm concludes that the ngAP and the nAP can
coexist on the same channel if both APs can accept ITPA-
based coexistence.
The algorithm repeats all above procedures, which examine
whether both the nAP and the ngAP can coexist, with all
possible changeable channel configuration. If the algorithm
finds an affordable configuration of all channels listed in
the spectrum map, the nAP initiates the ITPA negotiation
through the controllers described in Section I, but the detailed
procedures of the ITPA negotiation are beyond the scope in
this paper. The ngAP can break the coexistence of its operating
channel if the ngAP detects a decrease in service quality
below the minimum SNR. This case can occur because of
portable clients, and the ngAP increases its transmit power
while the nAP switches its operating channel to another unused
or reusable channels immediately. Thus, the nAP sharing the
channel of the ngAP must continue to search unused channels
to be able to switch to the new channel immediately.
III. THEORETICAL ANALYSIS
For theoretical analysis, we define two evaluation metrics
for evaluating the performance of the proposed ITPA algo-
rithm. We only consider the forward link of white space since
the majority of data transmission occurs in the forward link.
A. Network and Channel Models
We adopt indoor network environment and media traffic as
the respective reference network model and target application.
Thus, we choose the values of evaluation parameters according
to the characteristics of indoor channel environment and media
traffic service. To support the target bit error rate (BER) of
10−4 for media traffic service, this paper takes target SNRs
of 10, 20 and 30 dB for QPSK, 16/32 QAM and 256 QAM
modulations, respectively [17]. As the channel model, we
adopt Log-normal shadowing [18] and exponential Rayleigh
multi-path fading model [19] to calculate the expected mean
path loss, Ploss, and the expected RSS (ERSS), Prx, which
are expressed as
Ploss(r, f, d0) = L0 + 10η log(r/d0) +X
Prx(Ptx, Ploss) = −(Ptx − Ploss(r, f, d0)) ln(1− τ),(3)
where r is the distance to the center of the AP, and L0, which
is equal to 10 log(c/4πf)2, is the path loss to the reference
distance d0 (= 1 m). f (= 600 MHz) is the channel frequency,
c is the light speed, and Ptx is the transmit power of the
AP. η (= 3) is the log-normal path exponent, X is a zero-
mean Gaussian distribution with standard deviation, σ, and τis a random value between 0 to 1. In the theoretical analysis
and the simulation, the reverse function of (3) is used as the
function to calculate the distance between two APs. According
to the characteristics of indoor radio propagation, PN is 6 - 174
+ 10 log (6 MHz), ρ is from 0.001 to 0.004, and σ has 12 dB
[20]. Rc is the distance of sensing coverage, and this paper
assumes that the nAP can detect ngAPs within Rc (= 180 m, 6
clusters × 30 m average wall distance on IEEE 802.11n Model
F [21]).
B. Expected Success Probability of ITPA Algorithm
We define the probability of interference free (PIF) of the
ngAP at Lng and the PIF of the nAP at Ln. The PIF of
the ngAP is the probability that the expected RSS of the
signals of the ngAP at Lng is greater than not only the
required Pmin but also the expected RSS of the signals of
the nAP at Lng . In addition, the PIF of the nAP at Ln is
the probability that the expected RSS of the signals of the
nAP at Ln is greater than not only the the required Pmin
but also the expected RSS of the signals of the ngAP at
Ln. Because the expected RSS follows the Gaussian distri-
bution, we can estimate the PIF using the following equation:
PA(d, Patx, P
vtx, D) = Q((P ′min − Prx(P
vtx, Ploss))/σ), where
d is the distance to the center of an aggressor WSAP, Patx
is the current transmit power of the aggressor WSAP, Pvtx is
the current transmit power of a victim WSAP, and D is the
distance between two APs. Q(x) = (1/√2π)
∫∞x
e−t2
2 dt is
the Q-function of the normalized Gaussian distribution with
standard deviation, σ.
Thus, the PIF of the nAP is expressed as
PIFn = P (Pnrx > P ′min)× P ((Pn
rx − Pdiff ) > Pngrx ) =
PA(Rn, Pngtx , Pn
tx, D1) × P ′A(Rn, Pngtx , Pn
tx, D1), where Pnrx
is the expected RSS of the signals of the nAP at Ln, and
P ′min(= Pmin+�P (P atx, PN , D− d)) of the nAP is the new
Pmin, where D−d is a distance used to calculate the expected
RSS of the ngAP. Pngrx is the expected RSS of the signals of
the ngAPs at Ln as a noise to the nAP, and P ′A is expressed
as P ′A = 1 − Q(((PN + Pdiff ) − Prx(Ptx, Ploss))/σ).Meanwhile, the PIF of the ngAP is expressed as
PIFng = P (Pngrx > P ′min)× P ((Png
rx − Pdiff ) > Pnrx) =
PA(D2, Pntx, P
ngtx , D1) × P ′A(D2, P
ntx, P
ngtx , D1), where Png
rx
is the expected RSS of the the signals of ngAP at Lng , P ′min
of the ngAP is the sum of Pmin and �Pn, and Pnrx is the
expected RSS of the signals of the nAP at Lng as a noise
to the ngAP. A general formula of the probability of the
success of the ITPA algorithm-based coexistence is expressed
as Ps(Pntx, P
ngtx , Rn, D1, D2) = β × PIFn × PIFng , where
β is a reliability constant of SNR measurements. We assume
that β is 1, and the exact value of β is beyond the scope of
this paper. Since Ps depends on parameters, D1, D2, Pntx,
489
20 30 40 50 60 70 80 90 100 110 120 1300.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
Distance, R(m)
S ITP
A :
Ana
lysi
s R
esul
t
n = 50 n = 30 n = 10
(a) SITPA with Psnr = 10 dB.
20 30 40 50 60 70 80 90 100 110 120 1300.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
Distance, R(m)
S ITP
A :
Ana
lysi
s R
esul
t
n = 50 n = 30 n = 10
(b) SITPA with Psnr = 30 dB.
10 20 300.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
R= 30 R= 60 R= 90 R= 120
SNR (dB)
S ITPA
: A
naly
sis
Res
ult
(c) SITPA for n = 50.
Fig. 2. Analysis results of the expected success rate of the ITPA algorithm (SITPA) with Ptx = 20 dBm.
Pngtx , and Rn, with given Pn
tx, Pngtx , and Rn, the probability
of the success of the ITPA algorithm-based coexistence varies
by D1 and D2. We adopt the uniform distribution as the
probability density function (pdf) for the distribution of the
clients associated to the WSAP. The pdf for the distance D2
is expressed as fRP (D2, Rm) = (2D2/R2m), 0 ≤ D2 ≤ Rm,
where Rm is the maximum service coverage [22]. Thus, the
expected probability of the success of the ITPA negotiation
with a given D1 is expressed as
E(Ps(D1, D2)) =
∫ R
0
fRP (r,R)Ps(Pntx, P
ngtx , Rn, D1, r) dr.
(4)
The variance of D1 depends on the AP density, ρ, and the
maximum distance of sensing coverage, Rc. Using ω = ρR2c =
αn, where ω is the average number of active neighboring
APs, n is the number of white space channels, and α is a
reusable index, we can estimate ρ where the newly-joining
AP requires the ITPA algorithm with given n, Rc, and α.
We estimate ρ using P (Ne ≥ αN) = 1 − P (Ne < αN) =
1 − e−ω∑αN−1
j=0ωj
j! , where P (Ne ≥ αN) is the probability
that the detected number, Ne, of neighboring APs is greater
than αN . In other words, P (Ne ≥ αN) is the probability that
newly-joining APs require the ITPA algorithm. With α ≈ 1.3,
which we acquired through a simulation, the calculated ρvalues for αn are 1.3e-004, 4.1e-004, and 7.0e-004 for n= 10, 30, and 50 respectively. We use the calculated ρvalues to calculate the expected distance D1 with the lth
neighboring ngAP, E(Dl1). In the calculation of E(Dl
1), we
use the pdf of the lth nearest neighbor [23], which is ex-
pressed as E(Dl1) =
∫∞0
r fl(r)dr =∫∞0
r ((2πlρlr2l−1)/(l−1)!)e−πρr2dr = (1/
√ρ)(((2n)!n)/((2n n!)2), where fl(r)
is the pdf of the l th nearest ngAP within a given ra-
dius, r , and ρ is one of three values matched to αn.
With Ps and E(Dl1), the expected probability of the suc-
cess of the ITPA algorithm-based coexistence with the
lth neighbor AP is expressed as E(Ps(E(Dl1), D2)) =
(1/R2)∫ R
02r Ps(P
ntx, P
ngtx , Rn, E(Dl
1), r) dr. Thus, the ex-
pected success probability of ITPA algorithm is expressed as
SITPA = 1−αn∏l=1
(1− E(Ps(E(Dl1), D2))). (5)
As depicted in Fig. 2(a) and (b), all cases with a lower
Psnr show better results than the cases with a higher Psnr.
For example, for n = 50 and R = 60m, SITPA with
Psnr = 10 dB is about 60% while SITPA with Psnr = 30 dB
is less than 20%. With Psnr = 20 dB, as depicted in Fig. 2(c),
SITPA for all n cases except R = 30 m is less than or about
50%. The previous results are reasonable since a higher Psnr
will increase Pmin of neighboring ngAPs so that a newly-
joining AP is hard to acquire the approval of sharing their
already used channel since the signals of the newly-joining AP
become a critical interference to the ngAPs. We can observe a
similar pattern with the cases having more available channels.
However, although more channels can increase the success
of ITPA-base coexistence, with a higher Psnr, SITPA is less
than 50% as depicted in Fig. 2(b). Previous results reveal
that the ITPA algorithm-based coexistence is appropriate to
accommodate newly-joining APs requiring a lower Psnr or
a higher Psnr with a short service distance. In fact, WSAPs
providing high quality service will be reluctant to allow other
WSAPs to share their white space channels, and thus, they
are not likely cooperative about coexistence. Instead, other
WSAPs satisfying with a lower Psnr like 20 or 10 dB seem
to be cooperative in spectrum sharing. Also, newly-joining
WSAPs can have plenty of shareable channels in some areas
where we have n = 30 or 50. In this case, the ITPA-based
coexistence can enhance the success of activating the newly-
joining WSAPs after all channels are occupied.
IV. SIMULATION ANALYSIS
We implement the ITPA algorithm in MatlabTM.
A. Simulation Scenario
The simulation uniformly distributes all APs and randomly
activates one new nAP. The selected AP, nAP, performs
spectrum sensing, and the nAP calculates the expected RSS of
490
0.000 0.002 0.004 0.006 0.008 0.0100.0
0.2
0.4
0.6
0.8
1.0
the AP density (�)
(b)
The
suc
css
rate
of
chan
nel a
ssig
nmen
t Ss, n=30 Si1,n=30 Si2,n=30 Ss, n=50 Si1,n=50 Si2,n=50
(a)
Fig. 3. The success rate of channel assignment with Ptx = 20 dBm, R =90 m, and Psnr = 10 dB.
all activated ngAPs measured at the nAP. If the expected RSS
is less than or equal to PN of the nAP, the simulation declares
the success of finding its operating channel. Otherwise, the
simulation executes the ITPA algorithm with a new transmit
power, Ptx = 10 or 20 dBm.
B. The Success of Newly-joining AP Activation
We estimate the following three variables:
• Ss: the success rate of spectrum sensing.
• Si1: Ss plus the success of the ITPA with Ptx = 20 dBm.
• Si2: Si1 plus the success of the ITPA with Ptx = 10 dBm.
Let Tn denote the total number of APs. The number of
spectrum sensing attempts is equal to Tn. If the number of
the success of spectrum sensing is Ts, Ss is expressed as
Ss = Ts/Tn. If the number of the success of the ITPA
algorithm with Ptx = 20 dBm is Ts1, Si1 is expressed as
Si1 = (Ts + Ts1)/Tn. If the number of the success of the
ITPA algorithm with Ptx = 10 dBm is Ts2, Si2 is expressed
as Si2 = (Ts+Ts1+Ts2)/Tn. Fig. 3 illustrates the simulation
results with Psnr = 10 dB and R = 90m. With n = 30, Si2 is
greater than Ss by 30% as depicted in the (a) label of Fig. 3.
With n = 30, the difference decreases since Ss has a higher
success rate of channel assignment during spectrum sensing
with more available channels. With n = 50, as the AP density
increases, the difference between Ss and Si2 increases. For
instance, at ρ = 0.004, the difference for n = 50, is greater
than 40% as depicted in the (b) label of Fig. 3. Thus, while
the total success rate decreases as the AP density increases,
our approach can improve the total success rate compared to
the general approach only relying on searching unoccupied
channels.
V. CONCLUSIONS
The ITPA algorithm aims to increase channel reusability of
white space by adjusting the transmit power and the service
coverage of white space applications. Theoretical analysis
and simulation results show that the ITPA algorithm can
assist newly-joining APs to activate their service even after
spectrum sensing results report that all channels are occupied.
As a future work, we will enhance the ITPA algorithm-based
coexistence mechanism that can operate on heterogenous white
space applications.
REFERENCES
[1] Federal Communications Commission. FCC Adopts Rules for Unli-censed Use of Television White Spaces, Nov. 2008.
[2] Q. Zhao and B. M. Sadler, “A survey of dynamic spectrum access: Signalprocessing, networking, and regulatory policy,” IEEE Signal ProcessingMag., vol. 24, no. 3, pp. 79-89, May 2007.
[3] S. Pollin, “Coexistence and dynamic sharing in cognitive radionetworks,” Cognitive Wireless Communication Networks, E. Hos-sain and V. K. Bhargava, Eds. New York, NY: Springer,2007.
[4] Ir. W. Lemstra, ”Cognitive Radio defying Spectrum Management Acollection of four (short) contributions,” Competition and Regulationin Network Industries, Nov. 2008.
[5] I. F. Akyildiz, W.-Y. Lee, M. C. Vuran, and S. Mohanty, “NeXt Gen-eration/Dynamic Spectrum Access/Cognitive Radio Wireless Networks:A Survey,” Computer Networks (Elsevier) Journal, Sept. 2006.
[6] J.Lunden, “Spectrum Sensing for Cognitive Radio and Radar Systems,”Dissertation for the degree of Doctor of Science in Technology, HelsinkiUniversity of Technology. Nov. 2009.
[7] A. Akella, G. Judd, S. Seshan, and P. Steenkiste, “Self-Management inChaotic Wireless Deployment,”, Mobicom ’05, Aug. 2005.
[8] S. Shrivastava, D. Agrawal, A. Mishra,S. Banerjee, and T. Nadeem,“Understanding the Limitations of Transmit Power Control for IndoorWLANs,”, IMC ’07, Oct. 2007.
[9] IEEE 802.22 Working Group on Wireless Regional Area Networks.[Online]. Available: http://www.ieee802.org/22/
[10] C. Cordeiro and K. Challapali, “C-MAC: A Cognitive MAC Protocolfor Multi-Channel Wireless Networks,” Proceedings of IEEE DySPAN,pp. 147-157, Apr. 2007.
[11] M.M. Buddhikot, P. Kolodzy, S. Miller, K. Ryan, and J. Evans,“DIMSUMnet: new directions in wireless networking using coordinateddynamic spectrum,” WoWMoM 2005, pp. 78-85, 2005.
[12] Q. Zhao, L. Tong, and A. Swami, “Decentralized cognitive mac fordynamic spectrum access,” DySPAN ’05, pp. 224-232, 2005.
[13] I. Stanojev, O. Simeone, Y. Bar-Ness, and T. Yu, “Spectrum Leasing viaDistributed Cooperation in Cognitive Radio,” ICC ’08. pp. 3427-3431,2008.
[14] N. R. Suri and Y. Narahari, “An auction algorithm for procuring wirelesschannel in a heterogenous wireless network,” WOCN 2008. 2008.
[15] H. Zheng and C. Lili, “Device-centric spectrum management,” DySPAN’05, pp. 56-65, 2005.
[16] K. Kim, J. Park, J. Cho, K. Lim, R.C. J., K. Kim C.-H. Lee, H. Kim, and J. Laskar, “Interference Analysis and SensingThreshold of Detect and Avoid (DAA) for UWB Coexistence withWiMax,” Vehicular Technology Conference , vol. 2, pp. 1731-1735, Oct.2007.
[17] S.A. Emd, “Wireless Data Transmission Based on Adaptive Modulationand ARQ Control,” Journal of Engineering and Applied Sciences 3(4).Medwell Journals, pp. 337-343, 2008.
[18] T. Rappaport, Wireless Communications:Principles and Practice. 2nd ed.Prentice Hall, 2002.
[19] J. Mullen and H. Huang, “Impact of Multipath Fading in Wireless AdHoc Networks,” Proceedings of the 2nd ACM international workshopon Performance evaluation of wireless ad hoc, sensor, and ubiquitousnetworks, Oct. 2005.
[20] A. Goldsmith, Wireless Communications. Cambridge University Press,2005.
[21] F. Mlinarsky, “Equipment meeting the complex 600-Mbps multiple-input, multiple-output WLAN standard will require sophisticatedchannel-emulation test strategies,” Test & Measurement World. Apr.2007.
[22] C. A. St. Jean. and B. Jabbari, “Bayesian game-theoretic modellingof uplink power determination in uniform, self-organising network,”Electronics Letters vol. 40, no. 9, Apr. 2004.
[23] H. R. Thompson, “Distribution of Distance to Nth Neighbour in aPopulation of Randomly Distributed Individuals,” Ecology, vol. 37, no.2, pp. 391-394, Apr. 1956.
491