Hindawi Publishing Corporation

Journal of Electrical and Computer Engineering

Volume 2012, Article ID 245895, 12 pages
doi:10.1155/2012/245895

Research Article
Power Allocation in the TV White Space under Constraint on
Secondary System Self-Interference

Byungjin Cho, Konstantinos Koufos, Kalle Ruttik, and Riku Jantti

Department of Communications and Networking, School of Electrical Engineering, Aalto University, 02150 Espoo, Finland

Correspondence should be addressed to Byungjin Cho, [email protected]

Received 10 February 2012; Revised 30 April 2012; Accepted 26 June 2012

Academic Editor: Paul Dan Cristea

Copyright 2012 Byungjin Cho et al. This is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

The Electronic Communication Committee (ECC) in Europe proposed a location-based transmission power allocation rule
for secondary devices operating in the TV white space (TVWS). The further the secondary device is located from the TV cell
border the higher transmission power level it can utilize. The Federal Communication Committee (FCC) in the US proposed
a fixed transmission power allocation rule for all secondary transmitters. Both rules do not consider the secondary systems
self-interference while setting the transmission power levels. In this paper, we propose a power allocation scheme for a cellular
secondary system. Unlike the ECC and the FCC proposals we do the power allocation by considering the self-interference. We
define the power allocation scheme as an optimization problem. The sum cell border data rate of the secondary network is selected
to be the optimization objective. We observe that the optimal transmission power levels become approximately constant over the
secondary deployment area. The FCC rule captures the general trend for cellular deployment in the TVWS, since it suggests the
use of constant power. However, the transmission power should not be set equal to 4 W but according to the allowable generated
interference at the borders of the TV and secondary cells.

1. Introduction margin can be treated as an available resource and the

problem of allocating the transmission power to the cellular
Recently, cellular networks have transitioned from providing network can be viewed as a resource-sharing problem. Each
mobile telephony with limited data to supporting diverse secondary user is allowed to take a bite out of the available
types of applications with high capacity requirements. The resource; that is, it is allowed to generate some amount of
increasing capacity demands for the next-generation cellular interference at the TV receivers provided that the aggregate
systems cannot be accommodated within the currently interference of secondary transmissions does not exceed the
allocated spectrum resources. To some extent, the cellular interference margin.
spectrum deficit can be overcome by enabling cellular access The standardization bodies in the USA and Europe have
to the unused portions of spectrum in the TV bands, also so far proposed dierent approaches for transmission power
known as TV white spaces (TVWSs). The operation of allocation to secondary spectrum users. The Electronic
cellular networks in the TVWS has been already recognized Communication Committee (ECC) in Europe [3] proposes
as a scenario with clear business and economic impact [1]. a location-based transmission power allocation rule: the
The main requirement for secondary operation in the further the secondary user is located from the TV cell
TVWS is to maintain the QoS at the TV receivers. The QoS border the higher transmission power it can utilize. For
can be maintained if the interference level at the TV receivers a small number (up to four) of simultaneous secondary
is controlled. In the absence of secondary transmissions transmissions, the ECC suggests to reduce the transmission
the TV receivers experience only the TV systems self- power of each user by a fraction equal to the number of
interference. The dierence between the TV self-interference active users. This means that each user generates the same
level and the maximum interference level not violating amount of interference at the TV cell border. Essentially,
their QoS is called interference margin [2]. The interference the available interference margin is shared equally among
2 Journal of Electrical and Computer Engineering

the active users. For a higher number of simultaneous current FCC proposal may violate the interference margin
secondary transmissions the ECC rule controls the aggregate as demonstrated in [5]. The constant power allocation rule
interference by means of an additional safety margin. The has low complexity and it can be useful for determining
Federal Communications Committee (FCC) in the USA [4] countrywide power allocation for cellular networks in the
proposes a fixed transmission power level allocation equal TVWS. Thanks to the low complexity, this rule can be used
to 4 W. The interference generated at the TV cell border for determining the minimum possible protection distance
can be controlled by setting protection distance. Given the that allows the TV and the cellular network to coexist without
same transmission power level and the dierent distances to generating harmful interference to each other.
the TV cell border, the secondary users take unequal shares The outline is as follows. Section 2 presents the system
of the available interference margin. One can deduce that model and introduces the interference margin for the TV
the current proposals by ECC and FCC have a fundamental and the cellular systems. Section 3 formulates the problem of
dierence in splitting the available resource under multiuser downlink transmission power allocation as an optimization
transmission scenario. problem. Section 4 illustrates that a constant power alloca-
The existing aggregate interference control methods tion rule is approximately optimal for secondary cells of the
adopted by the two standards can lead to unacceptable same size. The rule is also utilized to plan a secondary cellular
interference increase at the TV receivers as demonstrated in network in a country-wide level. Section 5 concludes this
[5]. Aggregate interference control algorithms that guarantee paper.
the protection of TV receivers in the presence of secondary
operation have been proposed in the academic research 2. System Model
community. Most of the proposed algorithms make one of
the following assumptions: (i) the generated interference Figure 1 depicts the coverage area of a TV transmitter and
is modelled only through its mean value and channel a part of the cellular network deployed outside of the TV
uncertainties due to the fading are not considered [6, 7]; protection contour. The cellular network operates in the
(ii) the generated interference is controlled only at a single same frequency spectrum as the TV transmitter.
point and not along the TV coverage cell border [79]; (iii) In TV network planning the location probability
the transmission power allocation in the secondary devices describes the percentage of locations within a square area of
is uniform [810]. In [11] the aggregate interference control 100 100 m2 , also known as pixel, where the TV reception
algorithm does not suer from the above simplifications. is satisfactory [3]. For evaluating the location probability in
The algorithm proposes to divide the secondary deployment the presence of secondary transmissions we allocate a set of
area into multiple regions. The emitted spatial power density pixels at the border of the TV coverage area. If the location
is allocated per region such that the sum power density probability at these pixels, hereafter referred to as the TV test
is maximized and the aggregate interference along the points, is maintained above a target threshold the operation
TV cell border is controlled. Unfortunately, the capacity of any TV receiver inside the TV coverage area is deemed
requirements of the cellular systems are overlooked in [11]. satisfactory. Similarly, we assume that a minimum target data
The proposed algorithm will associate regions that are rate is available at the cellular end users if it can be achieved
deployed far from the TV coverage areas with a high power at the cellular cell border. For that purpose, another set of
density at the cost of regions located close to the TV cell test points is allocated at the borders of secondary cells (see
borders which will remain practically silent. Figure 1).
In this paper, we propose a scheme for setting the For successful TV operation a target SINR, t , must be
transmission power level in a cellular secondary network maintained with specific outage probability OTV due to the
without violating the protection criteria of the TV receivers. slow fading. The outage probability OTV is complementary
While planning the coverage of a cellular network the system to the location probability, q = 1 OTV , widely used in the
designer must guarantee a minimum signal to interference definition of TV coverage contour [3]. The SINR, j , at the
and noise ratio (SINR) at the cell edge. Unlike the ECC jth TV test point is
and the FCC proposals we take into consideration the
secondary system self-interference constraints while setting Sj
j = , (1)
the transmission power levels. Our scheme is formulated as ITV, j + ISU, j + PN
an optimization problem. The sum cell border data rate is
selected to be the optimization objective. Our scheme can where S j is the wanted TV signal level, ITV, j , ISU, j denote
be viewed as a method to divide the available interference the aggregate interference due to the interfering TV and
margin among the cellular base stations. secondary transmissions, respectively, and PN denotes the
We observe that the optimal transmission power noise power level at the TV receivers. The condition for
becomes approximately constant over the secondary deploy- acceptable TV operation can be read as
ment area. The FCC rule appears to capture the general  
trend for cellular secondary deployment in the TVWS, OTV Pr j t , j. (2)
since it suggests the use of constant power. However, the
transmission power level should not be arbitrarily set equal Similarly, while planning the coverage of a cellular
to 4 W but according to the interference margin available network, a minimum data rate should be guaranteed at
at the borders of the TV and secondary cells. Note that the the cell edge. The impact of fast fading to the achievable
Journal of Electrical and Computer Engineering 3

rp

RTV

BS 1

BS 2

BS 0MS

DVB BS 3

Figure 1: System illustration for single TV transmitter case. The cellular network operates co-channel to the TV transmitter and it is deployed
outside of the TV protection area. The aggregate interference has to be controlled at the TV test points and the cellular cell borders.

data rate is ignored which is a valid assumption for a where Pm is the transmission power level for the mth
low mobility scenario. In the presence of slow fading the interfering TV broadcaster, m is the TV broadcaster where
minimum data rate is achieved if a target SINR t with the jth TV test point belongs, and G is the channel model
given outage probability OSU is satisfied. The condition for used to design the TV system.
successful operation of a cellular end user can be read as The generated interference at the ith cellular test point
  consists also of two parts: cellular self-interference, ISU,i , and
OSU Pr i t , i. (3) TV interference ITV,i :
The SINR, i , at the ith secondary test point is   
ISU,i = pk g rk,i ,
si
i = , (4) k=
/ k
ITV,i + ISU,i + pN  (7)
 
where si is the wanted signal level and pN denotes the noise ITV,i = Pm G rm,i ,
m
power at the secondary receivers.
Data trac in cellular systems has an asymmetric where k is the cell where the ith test point belongs. For
behaviour with more trac generated in the downlink than modelling the TV and cellular transmissions
in the uplink [12]. Due to the unbalanced trac behaviour,
the downlink transmissions will aect the TV services more G(r) = ATV (r) 10XTV /10 ,
critically [13]. In this context, only the downlink interference (8)
is modelled. It is assumed that the cellular base stations are g(r) = ASU (r) 10XSU /10 ,
located at the centre of secondary cells. The aggregate cellular
interference at the jth TV test point is where the XTV and XSU are zero mean Gaussian random
   variables modelling the slow fading with standard deviations
ISU, j = pk g rk, j , (5) SU and TV , respectively, and ATV (r) and ASU (r) describe
k
the distance-based pathloss due to the TV and cellular
where pk is the transmission power of the kth cellular base transmissions, respectively, and as a function of the propa-
station, g stands for the channel model used to describe the gation distance r and other parameters such as operational
secondary transmissions, and rk, j is the distance between the frequency, antenna height, and environment.
kth base station and the jth TV test point. Similarly, the Since the locations of the secondary transmitters and the
generated TV self-interference is TV receivers are known, the parameters ATV (r) and ASU (r)
   are deterministic. The only randomness in the propagation
ITV, j = Pm G rm, j , (6) pathloss is introduced through the parameters XTV and
m=
/ m XSU .
4 Journal of Electrical and Computer Engineering

2.1. Interference Margin. In the absence of secondary trans- of the aggregate TV and secondary self-interference and the
missions condition (2) is satisfied while planning the TV mSU,i (pk ) is the mean useful signal level in dB at the ith test
network. The dierence between the TV systems self- point of the kth cellular base station. The MI,i depends on
interference and the maximum generated interference not the transmission power levels of the interfering cellular base
violating (2) is the interference margin [2]. The interference stations through the parameter I,i and on the transmission
margin is in general dierent for dierent TV test points power level of the base station generating the useful signal
because dierent locations in the TV coverage area experi- through the function mSU,i ().
ence dierent SINR. Unlike the TV test points, the generated interference
The interference margin at the TV coverage cell border at the cellular cell borders is in the same order with the
has been calculated in [2] by using the Wilkinson method useful signal level. Nevertheless, Figure 2(b) shows that the
to approximate the distribution of the aggregate secondary approximation by setting I,i = 0 in (10) is valid unless
interference. The accuracy of the Wilkinson approximation the standard deviation SU takes a high value or the reuse
for log-normal fading [14] has been already investigated distance becomes small. After setting I,i = 0 in (10) the
(SU)
in [2, 15]. The approximation is good particularly in the interference margin I,i depends only on the transmission
upper tail of the interference distribution which necessitates power pk . In fact, it is a linear function of pk . Hereafter,
a good match at the lower tail of the SINR distribution. In
our system model this is of particular importance because
  
it determines whether the protection of TV receivers is (SU)   MI,i t , pk
satisfactory or not. I,i t , pk = exp

In general, the interference margin depends on the
locations of secondary transmitters. However, one has to  
2   
notice that the secondary generated interference is usually an exp TV2 Pm ATV rm,i pN ,
order of magnitude less than the TV signal level. This fact 2 m
provides the approximation tightness for the lower bound (11)
of the interference margin illustrated in [11]. The lower
bound of interference margin at the jth TV test point,
(TV)
I, j , is independent of the secondary users locations and where MI,i (t , pk ) = Q1 (1 OSU )SU ln(t ) + mSU,i (pk ).
because of that it makes the interference control process
easier:
3. Problem Formulation
   
(TV) MI, j (t ) 2 We are looking for the power allocation maximizing the sum
I, j (t ) = exp exp TV2
2 cell border data rate of the cellular system while not violating
   (9) the protection criteria of TV and cellular systems. We assume
Pm ATV rm, j PN , one-by-one scheduling in each cell. In order to evaluate the
m=
/ m sum rate optimization function we compute for each cell
the average cell border data rate over the test points of that
where MI, j (t ) = Q1 (1 OTV )TV ln(t ) + MTV, j . The cell. Then, we sum the calculated values over all the cells.
MTV, j is the mean useful TV signal level in dB at the jth Our optimization function has the form of (12a). Note that
TV test point, m are the indices of the TV broadcasters it is straightforward to extend the optimization function to
generating interfering signals at the jth TV test point, = include also test points inside the cells. For reducing the
10/ ln(10) is a scaling constant, and Q1 is the inverse of the amount of computations in the country-wide case study,
Gaussian Q function. we have considered only points located at the cellular cell
By using a similar approach as in [2] the interference borders.
(TV) (SU)
margin at the ith test point of the cellular system is The interference margins I, j , I,i are equal to the
maximum mean secondary interference level that does not
    violate the TV and cellular protection criteria, respectively.
  MI,i t , p 2 In order to maintain the mean generated interference under
(SU)
I,i t , p = exp + I,i2
2 the margins (9) and (11) the cellular base stations must set
  appropriately their transmission power levels. In the presence
2    of slow fading the mean interference level is equal to the
exp TV2 Pm ATV rm,i pN ,
2 m distance-based path loss calculated in (5) scaled with the
(10) mean of the slow fading exp(SU 2
/2 2 ). Solving I,
TV
j (t )

exp(SU /2 2 ) k pk ASU (rk, j ) for the unknown power levels

2

pk allows to express the TV system constraint (2) in the

where MI,i (t , p) = Q1 (1 OSU ) SU 2
+ I,i
2
ln(t ) + form of (12b). By following the same approach for the
mSU,i (pk ). The p stands for the vector of transmission power cellular test points constraint (12c) is obtained. Finally, the
levels for the cellular base stations. The kth element of p has power allocation scheme can be formulated as the following
been denoted by pk . The I,i in dB is the standard deviation constrained optimization problem:
Journal of Electrical and Computer Engineering 5

W    
Maximize : log2 1 + k,i p , (12a)
p Np k i
 
   2 (TV)
Subject to : pk ASU rk, j exp SU2 I, j (t ), j, (12b)
k
2
 
   2 (SU)  
pk ASU rk ,i exp SU2 I,i t , p k , i, k, (12c)
k =
2
/k

where W is the transmission bandwidth of the secondary function (12a) can be written as a dierence of two
system, N p is the number of test points per secondary cell, terms:
and k,i is the SINR at the ith test point of the kth secondary
cell, W    
log2 Ck,i + pk ASU rk,i
Np k i k
 
k,i p (14)
W    
  log2 Ck,i + pk ASU rk ,i ,
pk ASU rk,i Np k i
=
 
  . k = k /
k =
/k pk ASU rk ,i + m Pm ATV rm,i + pN

(13) where Ck,i = pN + m Pm ATV (rm,i ) are constants.

Both terms are increasing concave functions in p.
Note that the optimization problem (12a)(12c) may not Therefore the optimization problem (12a)(12c) can be for-
have a feasible solution. If the SINR target t is high, it may mulated as a dierence convex (DC) programming problem
not be satisfied no matter how high transmission power the under linear constraints. One can use, for instance, the
branch and bound method [19] to obtain the global optimal
cellular base station utilizes. Also, the SINR target at the TV
solution. However, the monotonic optimization techniques
test points t imposes an upper limit on how much power
are practical only for small-scale optimization problems. In
the cellular base station can utilize.
this paper, we will consider a countrywide power allocation
scenario with a large number of secondary cells. Due to these
3.1. Comments on Optimal Algorithm. The data rates of reasons we resort to the constrained nonlinear minimization
secondary cells are coupled due to mutual interference. solver with interior point algorithm in the optimization
Optimal power allocation is known to be dicult to achieve toolbox of Matlab [20].
due to this complicated coupling among the SINR of
dierent links [16]. A single term in (12a) is a quasiconcave 3.2. Feasibility Check. Feasibility means there exists a set of
function of the transmission power levels p. In contrast to positive transmit power levels p such that the cellular cover-
the property of concave functions, quasi-concave functions age constraints (12c) are met while the interference caused
are not closed under addition, and the optimization problem to the TV test points does not violate the TV protection
(12a)(12c) is not convex. constraint (12b). Checking for feasibility in optimization
Possible methods for solving this nonconvex optimiza- problem (12a)(12c) is degenerated to identifying whether
tion problem are approximation, relaxation, and trans- the system of linear inequalities (12b)(12c) is consistent.
formation. For instance, in the high SINR regime the This is straightforward to check by using, for instance,
individual rate can be approximated by W log2 (k,i ) [17]. the simplex method. In Figure 3 we plot the minimum
Under this approximation, the optimization problem (12a) protection distance for dierent secondary SINR target
(12c) can be transformed into a convex problem in the levels. By increasing the distance separation between the two
form of geometric programming by proper change of systems the spatial spectrum reuse decreases but the data rate
variables. However, the high SINR assumption is not valid of the cellular system increases.
in general where fading environment is considered or
nearby links heavily interfere with each other. The authors 3.3. Constant Power Allocation. It is computationally dicult
in [18] solved the problem by using a generalized linear to solve (12a)(12c) for a large number of cellular base
fractional program. However, this algorithm is limited to stations. In Section 4 it will be illustrated that the optimal
only small-scale problems. In this paper we consider the solution to the optimization problem (12a)(12c) results in
cell border data rate in our optimization function, and approximately uniform power allocation over the secondary
thus, we cannot use the high SINR approximation. By using deployment area. This can be justified as follows: the
the quotient property of the logarithm, the optimization transmission power level of base stations located close to the
6 Journal of Electrical and Computer Engineering

100 25

Secondary transmission power level (W)

10 1 20

10 2 15
CDF

10 3 10

10 4 5

0
40 30 20 10 0 10 20 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
SINR (dB) Secondary cell radius (km)

R = 0.5 km, simulation R = 2 km, calculation Actual margin with SU = 5 dB Lower bound with SU = 4 dB
R = 0.5 km, calculation R = 5 km, simulation Lower bound with SU = 5 dB Actual margin with SU = 3 dB
R = 2 km, simulation R = 5 km, calculation Actual margin with SU = 4 dB Lower bound with SU = 3 dB
(a) (b)

Figure 2: (a) Distribution of the SINR at the cellular cell borders. The simulations are compared to the calculations. In our calculations
the aggregate interference is modelled by the log-normal distribution. Dierent secondary cell radiuses are tested. The standard deviation is
taken equal to SU = 5 dB. (b) Transmission power level required to achieve SINR target t = 3.5 dB at the cellular cell border with outage
probability OSU = 10% for dierent cell radiuses and reuse distance K = 3. The results obtained by using the actual interference margin
(10) are compared to the results obtained by using the lower bound of the interference margin (11). The rest of the parameter setting can be
found in Section 4.1.

50
45
Minimum protection distance (km)

40
35
30
25
20
15
10
5
0
5 0 5 10
Target SINR (dB) at cellular cell borders

K= 3
K= 4
K= 7

Figure 3: Minimum protection distance satisfying TV and secondary system cell coverage constraints with respect to dierent cellular SINR
target t . Dierent reuse distances K are tested. The cell radius is taken equal to 1 km. The rest of the parameter settings are the same used in
Section 4.

TV cell border is limited due to the generated interference TV test points is limited due to the interference they generate
at the TV test points. The base stations located close to the at nearby located secondary cells. Even though secondary
TV coverage areas should use just enough power to satisfy base stations located far from the TV coverage border can
their own coverage constraints (3). On the other hand, the utilize high transmission power the gains in data rate will be
transmission power level of base stations located far from the marginal due to the secondary self-interference.
Journal of Electrical and Computer Engineering 7

The optimal common power level that maximizes the cellular coverage constraints can be obtained by solving the
cellular cell border data rate under the TV protection and following concave problem:

W    
Maximize : log2 1 + k,i pSU , (15a)
pSU Np k i
  (TV)
2 I, j (t )
Subject to : pSU exp SU2
 , j, (15b)
2 k ASU rk, j
   
2 I (SU) t , pSU
,i
pSU exp SU2
 , i, k. (15c)
2 / k ASU rk ,i
k =

Thanks to its low complexity, the uniform power allo- modulation level and 2/3 code rate in the Rician channel
cation rule gives an opportunity to get quickly insight on [23, page 279]. The selected target SINR value t for the
the impact of various parameters on the cellular data rate cellular network with reuse distance K = 3 gives minimum
and the TV protection criteria. For instance, the system distance separation between the two systems approximately
designer can identify the minimum possible protection equal to 9 km which is at the same order with the protection
distance that allows the TV and the cellular network to distance proposed by FCC for cochannel primary/secondary
coexist without violating their own protection constraints. operation [4].
Also, the low complexity makes the uniform power allocation First, we carry out the simulation in a single TV cell
rule attractive for cellular network planning in the TVWS scenario and ignore the self-interference in the TV network,
over a country-wide level. ITV, j = 0 in (1). The TV cell radius is taken equal to 140 km
as an illustrative case study. For 350 kW TV transmission
4. Numerical Illustrations power level, the SNR at the TV cell border in the absence
of secondary transmissions becomes lower than 21.1 dB with
In this section we study the problem of allocating the 10% outage probability. Since the target SINR at the same
transmission power level in a cellular network deployed outage level is equal to 17.1 dB, there is a 4 dB margin
outside the protection area of a TV transmitter. When the cell that creates opportunity for secondary transmissions. The
size is fixed, it is illustrated that all the cellular base stations operating frequency is set to 482 MHz same as the TV
transmit approximately at the same transmission power channel 22 in Finland.
level. When the cell size can vary based on the population The optimization constraint (12b) has to be satisfied at
density it is illustrated that cells of the same size tend to use hundred test points uniformly allocated along the TV cell
approximately equal transmission power levels. We use this border. Outside of the TV protection area the secondary
approximation to study a country-wide cellular deployment system is deployed. The coverage area of the secondary
in the TVWS. system is limited at 40 km far from the TV protection area
border. This value has been motivated from [24] where it is
4.1. Parameter Settings. Dierent models are used to esti- shown that secondary interferers located further than 30 km
mate the field attenuation in the propagation path for the TV from the protection area border do not contribute much to
transmitters and the cellular base stations. The propagation the aggregate interference level. Also, constraint (12c) should
prediction for DVB-T signal over land path is obtained be satisfied at twelve points located at the border of each
by using the Recommendation ITU-R P.1546 [21]. The secondary cell.
modified HATA model [22] for suburban areas is used to
describe the path loss in cellular system. The receiver heights 4.2. Uniform Cell Size. In Figure 4 the transmission power
for TV and secondary reception are equal to 10 m and 1.5 m, levels for secondary base stations have been calculated by
respectively. The antenna height for cellular base stations is solving the optimization problem (12a)(12c). The protec-
taken equal to 10 m. The standard deviations for the log- tion distance must be larger than 9 km to give feasibility. It
normal fading distributions are TV = SU = 5 dB. The is selected equal to 11 km. For better illustration the spatial
outage probability is set OTV = OSU = 10% equal to the power density emitted from the secondary deployment area
complementary of the target location probability motivated is depicted. The spatial power density is computed by
from [3]. The thermal noise power is PN = pN = 106 dBm dividing the transmission power level pk with the secondary
for signal bandwidth W = 8 MHz. The SINR targets for cell size scaled by the reuse distance. One can observe that
the TV and the secondary receivers are taken equal to t = satisfying only the TV constraints results in high dierence in
17.1 dB and t = 3.5 dB, respectively. The selected SINR transmission power levels between the base stations located
target t for satisfactory TV operation is the minimum close and far from the TV cell border (Figure 4(a)). If
required value for quasi-error-free reception with 64 QAM the cellular coverage constraint is also taken into account
8 Journal of Electrical and Computer Engineering

0.01 0.01 0.01

5.49 5.49 5.49

Power density (W/km2 )

Power density (W/km2 )

Power density (W/km2 )
30 4.94 30 4.94 30 4.94
20 4.39 20 4.39 20 4.39
3.84 3.84 3.84
10 3.29 10 3.29 10 3.29
(km)

(km)

(km)
0 2.74 0 2.74 0 2.74
TV 2.2 TV 2.2 TV 2.2
10 10 10
1.65 1.65 1.65
20 1.1 20 1.1 20 1.1
30 0.55 30 0.55 30 0.55
0 0 0
140 150 160 170 180 140 150 160 170 180 140 150 160 170 180
(km) (km) (km)
(a) (b) (c)

Figure 4: Spatial power density emitted from the secondary deployment area obtained by solving the optimization problem (12a)(12c)
when taking into account (a) only TV protection constraints (12b) with protection distance 0 km, (b) both TV and secondary constraints
(12b), (12c) with protection distance 11 km, and (c) only TV protection constraints (12b) with protection distance 11 km.

1 1
0.9 0.9
0.8 0.8
0.7 0.7
0.6 0.6
CDF

CDF

0.5 0.5
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0 0
5.8 5.9 6 6.1 6.2 6.3 6.4 5.8 5.9 6 6.1 6.2 6.3 6.4
Data rate (Mbps) Data rate (Mbps)

r p = 11 km r p = 20 km r p = 14.4 km r p = 23 km
r p = 14 km Suboptimal r p = 17 km Suboptimal
r p = 17 km Optimal r p = 20 km Optimal
(a) (b)

Figure 5: Distribution of average cell data rate for dierent protection distances. The cellular transmission power levels are obtained by
solving the optimization problem (12a)(12c) and also by solving the optimization problem (15a)(15c). The cell radius is taken equal to
(a) 1 km (b) 2 km.

(Figure 4(b)), two observations can be made. Firstly, a In Figure 5, the distribution of the average cell data rate
protection distance between the TV and the cellular system is depicted for fixed cell radius, equal to 1 km in Figure 5(a)
is required to guarantee that the two systems do not generate and 2 km in Figure 5(b). For computing the average data rate
harmful interference to each other. Secondly, the transmis- in a cell, we generate a dense grid of points inside a cell
sion power levels allocated to secondary cells close and far and calculate the average data rate over all the grid points.
from the TV cell border are about the same. One may argue We compare the distribution resulting from optimizing
that the uniform transmission power levels can be attributed the transmission power per base station (12a)(12c) to
solely to the protection distance. However, it is the cellular the distribution resulting from using constant transmission
constraint that prohibits the cells located far from the TV cell power allocation (15a)(15c). The data rate distribution
border to utilize a high transmission power level; compare curves almost overlap. The cellular base stations far from the
Figures 4(b) and 4(c). Note that the higher transmission TV cell border are able to utilize high transmission power
power levels at the outer region of the secondary deployment levels without violating the protection limits of TV receivers.
area are attributed to the border condition, that is, the limited However, the gains in data rate are marginal due to the
self-interference these cells experience. secondary self-interference. Because of that, the constant
Journal of Electrical and Computer Engineering 9

30

20

10

(km)
0
TV
10

20

30

140 145 150 155 160 165 170 175 180

(km)

Figure 6: System illustration for dierent cell size coexistence case.

power allocation rule results in near-optimal data rate Figure 7 with dashed curves. One can notice that the data
values. rate distribution curves almost overlap. Assuming that cells
One can also notice that the uniform power allocation of the same size utilize same transmission power level results
rule becomes more accurate for larger protection distance. in near-optimal solutions. Next, we use this approximation
When the secondary network is deployed far from the TV to study power allocation in the TVWS at a country-wide
cell border, it can reach the interference-limited mode. In level.
interference-limited mode and provided that all cells have
the same radius, the data rate is maximized when all cells 4.4. Finland Case Study. For the Finland case study the TV
utilize the same maximum transmission power level. In our transmission power levels are not arbitrarily set as for the
computations the maximum allowable transmission power is single TV cell study. The actual transmission power levels
set equal to 100 W. have been taken from Finnish Communications Regulatory
Note also that for 2 km cell radius the optimization Authority (FICORA). The DVB-T coverage is calculated
problems (12a)(12c) and (15a)(15c) do not give any according to [25] by using the Recommendation ITU-R
feasible solution for protection distance equal to 11 km. The P.1546 [21] for land paths. The mean useful TV signal field
cellular base stations deployed close to the TV cell border strength for 500 MHz is equal to 52.5 dBuV/m [25] which
cannot utilize high enough transmission power for meeting corresponds to 78.6 dBm useful signal power. Also, the TV
their own target SINR without violating the TV constraints. self-interference terms are taken into consideration while
Nevertheless, when constant power allocation is employed, solving the optimization problems.
it is computationally easy to find the minimum protection In order to simplify the cellular deployment we cover the
distance resulting in feasible solutions. country with square cells. We consider three dierent cell
types, urban, suburban, and rural. The distance R from the
4.3. Nonuniform Cell Size. Next, we consider a cellular centre of the square to its vertices for the dierent cell types is
layout with non-uniform cell size (Figure 6). The secondary taken equal to 0.5 km, 2 km, and 5 km, respectively. The cell
deployment area is divided into two parts. The upper part type is selected such that the population inside a cell does not
is covered with hexagonal cells of radius 1 km, while the exceed the 10 000. In Figure 4 the cellular layout in Finland is
lower part is covered with cells of radius equal to 2 km. depicted. Densely populated cities in the south are covered
The rest of the parameter settings remains the same as with small urban cells while sparsely populated areas in the
in the previous case study. Firstly, the transmission power north are covered with large rural cells. The cellular base
level is optimized per cell. The distribution of data rate stations inside the protection contour of TV transmitters
values for the cells of 1 km and 2 km cell radius is depicted using channel 22 do not transmit. Our scheme does not
in Figures 7(a) and 7(b), respectively, with solid curves. consider so far adjacent channel TV protection. Similar to the
Secondly, it is assumed that cells of the same size utilize FCC rules, it is assumed that the cellular transmitters inside
the same transmission power level. Therefore, only two the protection area of the first adjacent channels do not
transmission power levels are identified while solving the transmit. The protection distances used in our simulation
optimization problem (12a)(12c). For instance, for 14 km are taken from the FCC rules [4], 14.4 km for the co-channel
protection distance the small cells utilize transmission power and 0.74 km for the adjacent channel. For frequency reuse
equal to 0.275 W while the large cells utilize transmission distance K = 3 it is not possible to satisfy simultaneously
power level equal to 1.340 W. The distribution of data rate the TV and the cellular constraints. Because of that, reuse
using same power for cells of the same type is depicted in distance K = 4 is utilized.
10 Journal of Electrical and Computer Engineering

1 1
0.9 0.9
0.8 0.8
0.7 0.7
0.6 0.6
CDF

CDF
0.5 0.5
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0 0
5.8 5.9 6 6.1 6.2 6.3 6.4 6.5 6.6 5.8 5.9 6 6.1 6.2 6.3 6.4 6.5 6.6
Data rate (Mbps) Data rate (Mbps)

r p = 14 km Suboptimal r p = 14 km Suboptimal
r p = 20 km Optimal r p = 20 km Optimal
r p = 25 km r p = 25 km
(a) (b)

Figure 7: Distribution of average cell data rate by optimizing the transmission power for each base station and also by assuming equal
transmission power for cells of the same size. (a) Upper part of secondary deployment area with cell radius 1 km. (b) Lower part of secondary
deployment area with cell radius 2 km.

15.816
5 km
14.059

Average data rate (Mbps)

12.301
10.544
2 km
8.787
7.03
0.5 km 5.272
3.515
1.758
Blackspace
0

(a) (b)

Figure 8: (a) Cellular layout in Finland based on the population density. The black space corresponds to the area where the secondary
transmissions are not allowed. (b) Color-coded map of average data rate for a cellular network operating in TV channel 22 in Finland.

We assume that the transmission power level is common The power density can be calculated as the ratio of
for cells of the same type. In that case, the solution of the allocated transmission power divided by the cellular cell
optimization problem (12a)(12c) results in transmission size scaled by the reuse distance. The power density values
power levels equal to 88 mW for urban cells, 1.54 W and allocated to urban, suburban, and rural cells are equal to
2
10.2 W for suburban and rural cells, respectively. Figure 8(b) {44, 48, 51} mW/km , respectively. One can deduce that the
depicts the color-coded map of cell average data rate in power density emitted from the deployment area of the
Finland by using the above-mentioned transmission power cellular system is almost uniform over all the country. This
values. The cell average data rate is described by the average remark agrees with the recently proposed uniform power
of the data rate values at 36 points uniformly located along density allocation rule [26].
the cell border and inside the cell. One can see that the Figure 9 validates that the protection criteria for the
average data rate for an urban cell in the TVWS for the TV and the cellular system are satisfied while solving
calculated transmission power levels is on average 10 Mbps. the optimization problem (12a)(12c). The distribution
Journal of Electrical and Computer Engineering 11

1 1
0.9 0.9
0.8 0.8
0.7 0.7
0.6 0.6
CDF

CDF
0.5 0.5
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0 0
20 10 0 10 20 30 40 50 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
SINR (dB) Outage probability

TV cell borders TV cell borders

Cellular cell borders Cellular cell borders
(a) (b)

Figure 9: (a) SINR distribution at the TV test points and the cellular cell borders. (b) Distribution of the outage probability at the TV test
points and the cellular cell borders.

0.5 1
0.45 0.9
0.4 0.8
0.35 0.7
0.3 0.6
CDF

CDF

0.25 0.5
0.2 0.4
0.15 0.3

0.1 0.2

0.05 0.1

0 0
10 12 14 16 18 20 22 24 26 28 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
SINR at TV edge (dB) Outage probability

TV and secondary system TV and secondary system

Only TV system Only TV system
(a) (b)

Figure 10: (a) SINR distribution at the TV test points. (b) Distribution of the outage probability at the TV test points.

of the SINR at the TV test points and the cellular cell Figure 10(a) shows the SINR degradation for the
borders is simulated and depicted in Figure 9(a). The outage country-wide scenario due to the secondary transmissions.
probability is approximately equal to 4% at the TV and In Figure 10(b) one can see that the mean location prob-
the cellular cell borders which is lower compared with ability is reduced from 98.2% to 96.3%. Overall, by using
their outage probability target, 10%. Obviously, only few the power allocation algorithm proposed in the paper,
of the constraints (12b)-(12c) are tight. This observation is the secondary transmissions would reduce the location
confirmed in Figure 9(b) depicting the distribution of the probability of the TV system but will not bring it under its
outage probability at the test points. One can observe that target limit. In this sense, the coverage area of the TV system
80% of the TV test points experience an outage probability is not compromised because the target location probability
of less than 5%. at the target SINR is not violated.
12 Journal of Electrical and Computer Engineering

5. Conclusions and Computing (MobiCom 09), Beijing, China, September

2009.
In this paper we proposed a method to set the transmission [9] N. S. Shankar and C. Cordeiro, Analysis of aggregated
power level in a cellular network operating in the TVWS interference at DTV receivers in TV bands, in Proceedings of
without violating the protection criteria of TV receivers. the 3rd International Conference on Cognitive Radio Oriented
Unlike the existing ECC and FCC rules we consider the cellu- Wireless Networks and Communications (CrownCom 08), May
lar self-interference constraint while setting the transmission 2008.
power levels. The paper shows that allocating same power to [10] E. Larsson and M. Skoglund, Cognitive radio in a frequency-
cells of the same size results in approximately maximum sum planned environment: some basic limits, IEEE Transactions
on Wireless Communications, vol. 7, no. 12, pp. 48004806,
cell border data rate. However, the common transmission
2008.
power level should not be set equal to 4 W as proposed by [11] K. Koufos, K. Ruttik, and R. J. Jantti, Controlling the interfer-
FCC rules but according to the interference margin available ence from multiple secondary systems at the TV cell border,
at the borders of the TV and secondary cells. In this paper we in Proceedings of the Annual IEEE International Symposium on
did not consider the impact of secondary interference to the Personal, Indoor and Mobile Radio Communications (PIMRC
adjacent channel TV receivers. We assumed that the FCC rule 11), Toronto, Canada, September 2011.
not allowing secondary transmissions inside the protection [12] W. Lu, Broadband Wireless Mobile: 3G and Beyond, Wiley,
area of the first adjacent channels is sucient. In our future Hoboken, NJ, USA, 2002.
work we plan to extend the power allocation rule including [13] A. Sadek and S. Shellhammer, Aggregate interference at a
also adjacent channel protection from aggregate secondary DTV receiver for a hexagonal cell structure, IEEE 802. 22-
interference. 07/0411r1, 2007.
[14] G. L. Stuber, Principles of Mobile Communications, Publishers,
Kluwer Academic, 1996.
Acknowledgments [15] P. Cardieri and T. S. Rappaport, Statistical analysis of co-
channel interference in wireless communications systems,
This work was partially supported by the TEKES-funded Wireless Communications and Mobile Computing, vol. 1, no. 1,
Project IMANET+ and by the EU FP7 Project INFSO-ICT- pp. 111121, 2001.
248303 QUASAR. [16] Z. Q. Luo and S. Zhang, Dynamic spectrum management:
complexity and duality, IEEE Journal on Selected Topics in
Signal Processing, vol. 2, no. 1, pp. 5773, 2008.
References [17] C. W. Sung, Log-convexity property of the feasible SIR region
in power-controlled cellular systems, IEEE Communications
[1] J. Zander, J. Kronander, A. Achtzehn, M. Nekovee, K. W. Letters, vol. 6, no. 6, pp. 248249, 2002.
Sung, and S.-L. Kim, QUASAR scenarios for white space [18] L. P. Qian, Y. J. Zhang, and J. Huang, MAPEL: achieving
assessments and exploitation, in URSI EMC Conference, global optimality for a non-convex wireless power control
Worclaw, Poland, September 2010. problem, IEEE Transactions on Wireless Communications, vol.
[2] K. Ruttik, K. Koufos, and R. Jantti, Modeling of the secondary 8, no. 3, pp. 15531563, 2009.
systems generated interference and studying of its impact on [19] R. Horst and N. V. Thoai, Dc programming: overview,
the secondary system design, Radioengineering, vol. 19, no. 4, Journal of Optimization Theory and Applications, vol. 103, no.
pp. 488493, 2010. 1, pp. 143, 1999.
[3] Technical and operational requirements for the possible [20] P. Venkataraman, Applied Optimization with MATLAB Pro-
operation of cognitive radio systems in the white spaces of the gramming, John Wiley & Sons, 2009.
frequency band 470-790 MHz, ECC Report 159 (under public [21] Method for point-to-area predictions for terrestrial services
consultation), 2010, http://www.ero.dk/. in the frequency range 30 MHz to 3000 MHz, ITU-R P. 1546-
[4] In the Matter of Unlicensed Operation in the TV Broadcast 4, 2009.
Bands: second Memorandum Opinion and Order, Federal [22] ERC Report 68, Monte Carlo Radio Simulation Methodol-
Communications Commission, FCC 10-174, 2010, http:// ogy, Naples, Italy, 2000.
hraunfoss.fcc.gov/edocs public/attachmatch/FCC-10-174A1 [23] U. Reimers, DVB-the Family of International Standards for
.pdf. Digital Video Broadcasting, Springer, Berlin, Germany, 2nd
[5] R. Jantti, J. Kerttula, K. Koufos, and K. Ruttik, Aggregate edition, 2005.
interference with FCC and ECC white space usage rules: case [24] K. Ruttik, K. Koufos, and R. Jantti, Computation of aggregate
study in Finland, in Proceedings of the IEEE International interference from multiple secondary transmitters, IEEE
Symposium on Dynamic Spectrum Access Networks (DySPAN Communications Letters, vol. 15, no. 4, pp. 437439, 2011.
11), pp. 599602, May 2011. [25] Final Acts of the Regional Radiocommunication Conference
[6] M. Vu, N. Devroye, and V. Tarokh, On the primary exclusive for planning of the digital terrestrial broadcasting service in
region of cognitive networks, IEEE Transactions on Wireless parts of Regions 1 and 3, in the frequency bands 174-230 MHz
Communications, vol. 8, no. 7, pp. 33803385, 2009. and 470-862 MHz (RRC-06), ITU-R, 2006.
[7] N. Hoven and A. Sahai, Power scaling for cognitive radio, in [26] K. Harrison and A. Sahai, Potential collapse of whitespaces
WirelessCom Symposium on Emerging Networks, Technologies and the prospect for a universal power rule, in Proceedings of
and Standards, June 2005. the IEEE International Symposium on Dynamic Spectrum Access
[8] S. A. R. Zaidi, D. C. Mclernon, M. Ghogho, and B. Zafar, On Networks (DySPAN 11), pp. 316327, May 2011.
outage and interference in 802. 22 cognitive radio networks
under deterministic network geometry, in Proceedings of the
ACM International Symposium on Mobile Ad Hoc Networking
 International Journal of

Rotating
Machinery

International Journal of
The Scientific
Engineering Distributed
Journal of
Journal of

Hindawi Publishing Corporation

World Journal
Hindawi Publishing Corporation Hindawi Publishing Corporation
Sensors
Hindawi Publishing Corporation
Sensor Networks
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014

Journal of

Control Science
and Engineering

Advances in
Civil Engineering
Hindawi Publishing Corporation Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014

Submit your manuscripts at

http://www.hindawi.com

Journal of
Journal of Electrical and Computer
Robotics
Hindawi Publishing Corporation
Engineering
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014

VLSI Design
Advances in
OptoElectronics
International Journal of

International Journal of
Modelling &
Simulation
Aerospace
Hindawi Publishing Corporation Volume 2014
Navigation and
Observation
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
in Engineering
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Engineering
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2010
Hindawi Publishing Corporation
http://www.hindawi.com
http://www.hindawi.com Volume 2014

International Journal of
International Journal of Antennas and Active and Passive Advances in
Chemical Engineering Propagation Electronic Components Shock and Vibration Acoustics and Vibration
Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014