Sie sind auf Seite 1von 8

Outage Based Underlay Cognitive Radio Deployment

for a Cellular Network


Astha Sharma and Debjani Mitra
Department of Electronics Engineering
Indian School of Mines, Dhanbad
Dhanbad, India
17.asthasharma@gmail.com , debjani7@yahoo.com

Abstract— In this paper, the scope of underlay Cognitive User the underlay approach, the interference to PUs is controlled by
deployment as a function of a wide range of SIR threshold has using a predefined spectral mask on the CUs transmission
been investigated taking into account the different tolerance power such that the CUs power lies below the noise floor of the
levels of the primary network under the influence of scaled and primary receivers. The interweave approach utilizes the
controlled secondary transmitter power. The average number of temporal or geographical dynamics of the primary signal
cognitive transmitters confined beyond the primary exclusive occupancy which aids in the determination of the spectral holes
region that can coexist while maintaining a particular primary to avoid interference with PUs. CUs are allowed to transmit as
outage constraint has been estimated in the downlink scenario of
long as SINR requirements at all the primary receivers are met
a standard hexagonal type frequency planned cellular network.
in the overlay approach.
The impact of shadowing correlation on the average outage
probability is also analyzed for a wide range of SIR threshold. When the primary system is a frequency planned network
Suitable power scaling is seen to have comparable performance (e.g. a cellular system), it is mostly configured optimally with
as compared to Distance Based Power Allocation (DBPA). respect to frequency reuse distance, cluster size and channel
allocation in a cell. According to predictions in [9], there will
Index Terms— Cognitive radio, Outage probability, underlay be massive utilization of frequencies below 25 GHz in year
approach, DBPA algorithm. 2017. Consequently, the frequency planning of cognitive based
new radio equipments will be more draconian due to the
I.INTRODUCTION deficiency of untapped frequencies below 25 GHz and thus
would emphasize more on shared use of spectrum between
Technological areas like OFDM, MIMO, Cooperative different systems and technologies. Use of license-exempt
communications, DPC (dirty paper coding), Multi-hop equipment, and – most importantly – the new UWB
communication are gearing the development of 4G wireless for technology, along with the increasing number of devices in
improved spectral efficiency. Spectrum scarcity however many applications, increase the need for resources and
continues to be a bottleneck in the whole process of coping up frequency planning. Use of the frequency bands below 25 GHz
with multimedia traffic demands and ever increasing number of is expected to increase significantly by 2017. New radio
users. The expensive natural resource of radio spectrum in spite systems using the frequency bands below 25 GHz must be able
of scarcity has been paradoxically characterized by inefficient to share their frequency use with existing equipment utilising
utilization associated with fixed spectrum licensing policies the same frequency band. Allowing other systems to use the
and several measurement campaigns throughout the world have frequency bands below 25 GHz requires some of the existing
reported about it [1,2,3]. This imbalance between the spectrum use to be moved to higher frequencies. One of the major
scarcity and spectrum underutilization has propelled many challenges is to opportunistically deploy secondary users to
researchers towards the emerging field of Cognitive Radio/ access the primary radio spectrum. This requires satisfying a
Dynamic Spectrum Access Technology (CR/DSA) [4,5,6]. The PU outage constraint so as to guarantee non-interference to
first worldwide standardized air interface based on CR PUs. In this paper, we have considered the effect of
techniques developed for the opportunistic use of TV bands on interference not only on primary receiver (PRx) but also on
a non-interfering basis is the IEEE 802.22 WRAN [7]. In a cognitive receivers (CRx’s). This requires maintaining outage
general cognitive radio network (CRN), both primary or constraint at both PU and CU. The potential source of CRx
licensed users (PU) and unlicensed secondary access points interference is simultaneous transmission of other opportunistic
(CUs) such as short-range devices (SRD), devices based on devices operating within the same band. Moreover, the sharing
new UWB technology, radio frequency identification (RFID) becomes simpler when both PUs and CUs belong to the same
devices, are allowed to operate where CUs dynamically access service operator and the use of underlay paradigm will ensure
the spectrum licensed to primary users without compromising that the secondary network is automatically assigned a lower
the incumbent primary user service. The spectrum priority in maintaining the interference limits at the receiving
underutilization and the existence of “holes” in the spectrum points of the primary network. A vital performance parameter
such as in cellular telephony , TV broadcasting etc may be is the transmit power which aids in controlling the interference
locally reused by another network of cognitive users in either at primary receivers. The idea of how the transmit power of CU
of three ways that are underlay, interweave and overlay [8]. In
must be scaled to meet a certain interference constraint at a
A
D  d2  d 2  2d 
 1  0  A 1
primary receiver is explored in [10]. In the context of the 2 
limitations discussed in [11], the authors have suggested that D  D
subject to increasing the tolerance limit of primary users, the (1)
challenging task of the deployment may give useful results. A  0 signifies the restriction on cognitive users to operate
This aspect has been investigated in this paper to study the anywhere while A  1 signifies that CUs would be allowed to
scope of underlay CU deployment as a function of a wide range operate everywhere. This area of cognitive operation plays a
of SIR threshold to take into account the different tolerance significant role in determining the relative size of the cognitive
levels of the primary network under the influence of scaled area where they are allowed to operate efficiently.
secondary user power. Also, in terms of capacity and for
systems implementing link adaptation on shared downlink
channels the whole SIR distribution is usually needed for
evaluation of deployment issues. So it is felt that the
repercussions of co-channel interference estimated by SIR
Outage Probability characterized on the downlink of a cellular
system can serve to be a good measure for analyzing the
system in the context of PU-CU scenario. The system model
similar to [10] is investigated in [12] where they emphasized on
the essence of properly designing the network parameters to
guarantee a certain performance limit to PUs. Theoretical
bounds on average aggregated interference power at primary
receiver are derived which aids in examining the primary
outage constraint and obtaining the relationship between
primary exclusive region (PER) radius, transmit power and
other network parameters useful in designing cognitive
networks with PER. Obviously, it is desirable to be able to Fig 1: System Model
predict whether the required SIR threshold is achievable using
the adopted power control technique. By controlling the In other words, it is a deciding parameter for choosing the
transmit power of CUs satisfying a certain PU constraint, the feasible values of d , where d is the minimum distance
deployment scope can obviously improve. In addition, the required for placing CUs away from primary receiver so as to
number of CUs that the system can support is also an important ensure minimum interference to primary receiver.
aspect to take care of as very less CUs signifies the system Each CU pair consist of a cognitive transmitter (CTx) also
inefficiency due to unutilized spectral holes while too many called as access point and a CRx which is placed within a
CUs causes system breakdown due to collisions amongst CUs. certain range say ‘ R ’ around CTx. The distance to first tier of
Interfering signals affected by shadowing are usually correlated co-channel interferers is D  3n r where n is the cluster
as the obstacles in the vicinity of mobile station block the size and r is the cell radius. With the primary BS located at
desired and interfering signals similarly [14]. Shadowing origin, the other six co channel Base Stations are specified at
correlation significantly affects system performance and is a
crucial component in achieving more realistic channel
coordinates  
D cos  [ j  1] p / 3 , D sin  [ j  1] p / 3
propagation models [15]. The system model in the paper has where j  1, 2,...,6 . Each of the base-stations transmits at
therefore considered correlated shadow fading path loss for the power Pt and let each CU transmit with power PCU . As the
Monte Carlo estimation of the outage over several CU
users are distributed over a wide geographic area so the signal
positions satisfying the primary user constraint as well as a
propagation model assumed is distance-dependent path loss
distance based power control. Section II describes the system
and lognormal shadowing for which the local mean power
model and its simulation methodology, the results of which are
discussed in Section III. The salient features of the analysis are level, denoted by  ( d l ) is given by
summarized in the concluding section.  (d l )  d l  10  / 10 ,  ~   0,  
II. SYSTEM SIMULATION MODEL (2)
where  is path loss exponent,  is the standard deviation in
A downlink scenario in the standard hexagonal type frequency
reuse with the effects of first tier co-channel interference has dB and d l is the Tx-Rx separation distance. This local mean
been considered as the system model here as shown in Figure power level follows the normal variation when expressed in dB
1. There exists two types of users where a primary user is and is characterized by two parameters namely, area mean
located inside the primary cell and N CU cognitive user pairs value and standard deviation. The power received by a desired
are randomly distributed in the outer annular region i.e. within   ri 
primary receiver (PRx) from BS0 is a realization of Pt �
the two circles centered at BS0 having radii d and D  d . where ri , 0 �ri �r is the distance between the i th PRx of
The area of cognitive operation as used in [11] is given as interest and BS0 located at origin.
For analyzing such scenarios where resources are dynamically
shared amongst wireless users, Signal-to-Interference Ratio
SIR denoted by ‘  ’ plays as an important performance total interference measured at PRx and CRx’s respectively can
metric. We have considered the interference channel model as be computed using the Wilkinson method [12, 13]. The
shown in Figure 2 which takes into account the effect of determination of mean and s.d of total interference experienced
interference on PRx as well as on CRx. This requires by PRx is given in Appendix where similar methology could be
estimating SIR at both PRx and CRx. The accurate estimation adopted for CRx’s. The mean and standard deviation of SIR
of interference level at any specific location requires
consideration of the shadowing effects upon the interfering
 measured at PRx m
P , P  and CRx’s m
S ,  S 
signals and the net interference received at a desired PRx and respectively is determined as:
CRx respectively. m  mS  mI dB (5)
       2   s I
2
s
2
I (6)
where m s and  s are the mean and standard deviation of the
desired received signal respectively and  is the correlation
factor. These moments facilitates in determining the outage
probability, which is a significant performance evaluation
metric for capacity and coverage issues. It measures the
probability of failing to attain a desired SIR threshold suitable
for a satisfactory performance. Outage probability is defined
as:
Poutage  th   Pr    th 
 th
1    x  m  2 
 exp   dx
Fig.2. Interference channel model with a single PRx and multiple CRx’s. 0 2p    2 2 
   m 
As can be seen, a PRx is subject to interference from first tier  1  Q th 
co-channel cell BS’s and N CU cognitive access points so    (7)
SIR at i th PRx can be expressed as: where Q (.) is the Q-function. Note that here the outages occur
due to the path loss, primary system BSs transmission and
Pt    ri  random position location of CUs in the considered system
P  N CU
model.

 Pt    d j    PCU    d m 
6
The mean sum capacity of CUs required for estimating the
(3) impact of densities of CU transmitters on system is defined as:
j 1 m 1 N CU
1
where d j and d m are the respective distance of the PU of C 
mean 
N CU
C m
interest from j th co-channel BS and mth CU. Similarly, as a m 1
(8)
CRx is subject to interference from other N CU  1 CTx’s,
where C m is the individual capacity of the m th CU
primary BS0 and co-channel cell BS’s so SIR at th
m CRx characterized by the following relation:
can be expressed as: 
C m  log 2 1   Sm  (9)

 
m
m

PCU   d CTx  Rx
 The evaluation of this capacity is necessary before deployment
and mainly relies on the characterization of interference
S 7 NCU described in terms of SIR  S .
 Pt    d k   P CU    ds  (4) The essence of controlling the transmit power comes from the
k 1 s 1,s  m fact that the CUs have the ability to optimize their
m transmissions with varying radio environment. We have used
where d CTx  Rx is the distance of m CRx from their
th
the criteria for controlling the transmit power of each CU
respective transmitters. d k is the distance between mth CRx which is the allocation of the transmit power according to
and BS’s of primary network which comprises of 6 co-channels Distance Based Power Allocation (DBPA) algorithm instead of
BS and a primary BS situated at center of cell. d s is the allocating equal power to each CU. Similar analogy can be
found in [12], where they theoretically derived the lower and
distance between mth CRx and other N CU  1 CTx’s upper bounds of aggregate interference considering this
causing co-channel interference to desired CRx. criterion and stated that there is equivalency between a network
The randomness of the signal and interference power levels due with power scaling and a network without power scaling where
to various phenomena like RF propagation effects, user the latter case is nothing but the gradual decay of power with
mobility and traffic variation, makes  P and  S random distance. Further, they have shown that for the interference to
variables. The mean m I and standard deviation (s.d)  I of stay bounded, the path loss factor need to be greater than 2
while in [10], they showed that a viable power control metric Three different cases have been envisaged to obtain the average
can be the local signal-to-noise ratio (SNR) of primary users primary user SIR outage for a wide range of SIR thresholds: (i)
which is used as a proxy for distance. According to DBPA No CUs exist and the interference is only due to the co-channel
algorithm, the distance between the PUs and CUs are computed BSs (ii) CUs transmit with same power scaled with respect to
to scale the CU transmit power according to the following PU (iii) CUs transmit with power controlled according to
equation: DBPA.
  PCU  
PCU Figure 3 computes the average outage probability in % while
maintaining PU interference constraint and shows the result for
(10)
  1 and N CU  6 where  denotes the ratio of the
 secondary and primary user transmitter power. There is a

 R if    min
significant improvement in the average outage probability with
a simple power control strategy and more involved ones could
reduce it further. This is quite obvious as the probability of
   max occurrence of outages get reduced by effectively controlling the
transmit power of CUs.
min
where  if    min Figure 4 shows the effect of the correlated lognormal
interferers on the average outage probability for all the three


 Rmax

cases and for three values of  as defined in (7). The general
trend of decrease of Poutage with increasing  is most
is real positive constant, is the distance between PU and
prominent when CUs are deployed with power control. The
CU and  min is threshold distance which is assumed for outage performance is further observed to change drastically
simplication as distance between the center of primary cell and and randomly as the number of secondary user is changed.
the midpoint of annular ring. Rmax is the maximum radius at
which the CUs are placed, which in our case is D-d.
The simulation methodology comprises of first defining the
system parameters and then generating the mobile user’s
locations required for determination of distances between: a)
PU and co- channel BSs , b) PU and CTx’s and c) CTx’s and
CRx’s. Using SIR statistics, the outage probability is
determined for different SIR threshold.

III. RESULTS AND DISCUSSIONS


The system model is developed in MATLAB and 10 4
~
snapshots were run to estimate N s , the average number of

Fig.4. Average outage probability at different SIR thresholds for different


correlation factors   0.0 , 0.5, and 0.9

Fig.3. Average outage probability vs. SIR threshold satisfying PU interference


constraint for   1 and N CU  6 .
cognitive transmitters that can coexist satisfying a particular
PU interference constraints. The parameters used are: primary
cell radius r  500m , R  100m   6dB , n  7 ,
  4 and A as in equation (1) is 25%. The transmit
power of each base station is taken as 1W .
Fig.5. Average outage probability vs. power of cognitive users (dBW) Figure 5 depicts the effect of increasing CUs power on average
satisfying PU interference constraint for N CU  6 . outage probability. As can be seen, the increase in CUs power
implies increased interference from CU Tx's to PU and hence,
in turn leads to increase in average outage probability. It is interference and hence leads to decrease in outage probability.
observed that for the scenario when CUs wish to transmit at
same power level as that of PUs say 0dBW, the outage
probability is reduced by 20% using DBPA technique as
compared to uniform power allocation. However, when CUs
desires to transmit at very low power levels say -30dBW, the
performance of CUs with and without power allocation is
almost same which is quite obvious. Thus, the benefit of power
allocation can be realized by suitably scaling CUs power Pcu.

Fig.7. Average outage probability vs. number of cognitive users, for   1 .

As can be observed from figure 7, the increase in number of


CUs leads to increased congestion in network, this motivates to
compute the average number of cognitive transmitters that can
coexist with licensed network satisfying PU outage constraint.
~
Fig.6. Average outage probability vs. SIR threshold satisfying PU interference Through extensive Monte Carlo simulation, N s the mean
constraint showing the impact of varying area of cognitive operation for   1 number of secondary transmitters that could coexist in the
and N CU  6 . primary network was estimated. Table I presents the results for
threshold SIRo in the range of 1 to 40dB, and two particular
Figure 6 shows the impact of varying area of cognitive
operation on the average outage probability for the two values of Poutage , namely 25% and 45%. Apparently, a larger
scenarios. It can be seen that for the uniform power allocation value of outage tolerance limit signifies the enhanced
scenario, the increase in A signifies reduction in minimum permissiveness of primary network leading to increased
distance (1) at which the CUs are allowed to be placed, which number of CUs deployment. The analysis for various ranges of
in turn leads to increased interference from CUs to PU causing SIRo is relevant as depending on traffic conditions, a system
increased outage probability. However, for the case of power designed for high SIRo could negatively influence the
allocation, the power of CUs is dependent on Rmax which
financial feasibility while designing for low SIRo would
increases with increase in A, which in turn leads to decrease in
aggravate the quality of service.
CUs power P CU scaled according to DBPA, causing less

TABLE I. FOR PRIMARY OUTAGE CONSTRAINT


[ ] [ ] [ ]
 I 2   I CCI   I CU  2 cov I CCI , I CU 
2 2

We can see that with the secondary users transmitting at about (13)
50% of the primary power,   0.5 , approximately the same Assume that I CCI and I CU are log normally distributed.
number of CUs can be deployed as that with power control. Once mean and s.d of individuals I Co , j and I C , m are
This aspect is important as by suitably scaling SU power, the
deployment issue can be handled even without power control in determined then we can apply Wilkinson method to determine
an optimized frequency planned environment. mean and s.d of summand I CCI and I CU . This in turn will
Scheduling of the cognitive transmitter however needs to be enable us to determine the mean and s.d of net interference I
done as per QOS requirement constraint of the secondary by using equations (12) and (13).
network, if any. The work is being extended in this regard with The local mean power I Co , j of j signal is expressed in
th
encouraging results.
decibel units as:
II. CONCLUSION  CO , j  10 log10 I CO , j  m X CO , j   CO , j dBW (14)
The paper has, by numerical simulation, estimated the number
where m X CO , j is the area mean power or path loss and
of secondary transmitters that can be deployed in a frequency
planned cellular network in the downlink scenario, taking into  CO, j is the factor described by random shadowing effects
account different interference tolerance levels of the primary which is a zero-mean normally distributed random variable in
network. The influence of a wide range of primary threshold dB with s.d.  X CO , j . The area mean power is given by
SIR and two specific primary outage has been quantified for
three cases considering the SU deployment with and without following equation:
power allocation. Suitable power scaling is seen to have m X CO , j  Pt , j  Gt , j  10 log10 d j (15)
comparable performance as compared to DBPA. The impact of N CO
shadowing correlation on the outage is also analyzed for the
entire range of SIR threshold. The results are significant for Also,
I CCI  I
j 1
CO , j

developing underlay spectrum allocation models and further (16)


scheduling of secondary users as per their capacity constraints.
It is well known that the distribution of summand I CCI can
III. APPENDIX be approximated by another lognormal distribution which can
be mathematically expressed as:
For estimating the mean m P and s.d   P of SIR at PRx,
X CCI  10 log10 I CCI (17)
 P , the challenging task is to calculate the statistics of net
interference encountered by PRx due to simultaneous where X CCI follows a normal distribution. The assumption
transmission of both co-channels interferers and cognitive of the sum I CCI being log-normally distributed allows for the
access points. Mathematically, this net interference is expressed computation of mean m X CCI and s.d  X CCI of X CCI with
as :
the aid of Wilkinson’s method. For easy derivation, it is
I  I CCI  I CU (9) favorable to use natural logarithm in place of base 10 logarithm
where I CCI is total co-channel interference power and I CU to define normal r.v that is similar to a log-normal r.v. This
is total interference power generated by CTx’s. The total co- normal r.v is defined as:
channel interference power is simply the resultant of the YCO , j  ln I CO , j
individual co-channel interference signals I Co , j whose local (18)
mean power levels undergoes lognormal variations, expressed which has mean mYCO , j and s.d  YCO , j given by :
as:
mYCO , j   m X CO , j and Y    YCO , j
I CCI   I Co , j   Pt    d j 
Nco CO , j

(19)
(10)
where   ln(10) 10 . This leads to YCO , j   X CO , j .
j j 1

where N CO  6 for first tier of co-channel interferers. Using equation (16) and (18), we have got the following
Similarly the total interference received at PRx due to multiple relation:
CTx’s is: YCO N CO
N CU I CCI  e YCO1  e YCO 2      e  e Z CCI  10 X CCI / 10 (20)
I CU   I C ,m   PCU    d m  where both Z CCI (in logarithmic units) and X CCI (in dB) are
(11)
m m 1 having normal distribution. Moreover, Z CCI   X CCI .
For calculating the statistics of SIR at PRx the mean and
standard deviation (s.d) of this net interference I can be After determining the mean and s.d of summands YCO , j , one
determined by using following relations: can compute the mean and s.d of Z CCI or X CCI by using
Wilkinson Method. This is done by matching the first and
[ I ]  [ I CCI ]  [ I CU ] (12) second moments of I CCI with those of
I CO1  I CO 2      I CO N CO . For determining the first N CU
I CU   I C ,m
moment which gives mean value of Z CCI , one needs to take Also,
m 1
(31)
expectation of equation (20) from which we have:
  e 
Again, approximating the distribution of I CU by another
e Z CCI YCO1
e YCO 2
   e
YCO N CO
lognormal distribution described as:
(21) X CU  10 log10 I CU (32)
For a normal r.v  with mean m and variance   and any
2
In natural logarithms:
integer value say  , the  th moments are evaluated from the YC , m  ln I C ,m
following expression : (33)

   1 
 e   exp m   2 2 
(22)
where YC , m is a normal r.v with mean mYC , m and s.d
 2  Y C ,m
given by :
With the help of above relation, the first moment of Z CCI is mYC , m   m X C , m and  YC , m   YC , m
[ ] 1
 e Z CCI  exp m Z CCI   Z2CCI 
2

(23)
(34)
  where   ln(10) 10 . This leads to YC ,m   X C ,m .
and
Now again defining I CU in terms of Z CU , we get:
 
N CO
�  e CO NCO � �exp mYCO , j   Y2CO , j 2
Y
eYCO1  eYCO 2  �
�� YC N CU
� � j 1 (24) I CU  e YC 1  e YC 2      e  e ZCU  10 X CU / 10 (35)
Using equation (23) and (24), we have: We define the first and second moments of Z CU as similar to

1,CCI  exp �
1 � NCO
mZCCI   Z2CCI � �exp mYCO , j   Y2CO , j 2
(25)   equation (25) and (28), we have:
� 2 � j 1
1,CU  exp mZ   
N CU

Now, for computing the second moment of Z CCI we have: CU


 2
Z CU 2   exp mYC ,m   Y2C , m 2

 
m 1


Y 2
�
e 2 ZCCI �
� �  e
YCO 1
 eYCO 2  �
�� e CO NCO (36)
(26) and
and  2,CU  k 2  e 2 (37)
2,CCI
 2 mZCCI  2 Z2CCI 
e
 
N CU
(27)
Again, matching the second moments given in equation (26) where k 2   exp 2mY C ,m
 2 Y2C , m and
m 1
and (27), we have:

  exp m 
N CU 1 N CU
2,CCI  k1  e1 (28) e2  2 YC , m  mYC , p
  and
N CO
m 1 p  m 1
k1  �exp 2mY  2 Y2CO , j
where j 1
CO , j
1
 
 exp  Y2C , m   Y2C , p  2YC , mYC , p  YC , m  YC , p  
 
N CO 1 N CO
2 
e1  2 � � exp mYCO , j  mYCO ,k
j 1 k  j 1  1  1,CCI   1,CU (38)
 2  2,CCI  2,CU  21,CCI �
1,CU
1

2

 �
exp �  Y2CO , j   Y2CO ,k  2 YCO , jYCO ,k  YCO , j  YCO ,k �


 �
mZ I  �
1 �
2log 1  log 2 �
(39)

The similar calculation is done for computing the moments of � 2 � (40)


net interference generated by multiple cognitive access points.
 Z I  sqrt  log 2  2log 1  (41)
Therefore, the local mean power I C , m of m th signal is
expressed in decibel units as: Note that the Z I is in logarithmic units and so we need to
 C ,m  10 log10 I C ,m  m X C , m   C ,m dBW convert it into dB, using following relations:
(29)
m X I  1   m Z I
where m X C , m is the area mean power or path loss and  C , m
(42)
 X I  1   Z I
occurs due to shadowing effects which is a zero-mean normally
distributed random variable in dB with s.d.  X C , m . Thus, the (43)
area mean power is described by following equation: REFERENCES
m X C , m  PCU ,m  Gt ,m  10 log10 d m [1] Spectrum Policy Task Force, “Spectrum Policy Task Force
(30) Report,” Federal Communications Commission ET Docket 02-
135, 2002.
[2] J.M. Peha, “Approaches to spectrum sharing,” IEEE Commun.
Mag., vol. 43, no. 2, pp. 10-12, Feb. 2005.
[3] G. Staple and K. Werbach, “The end of spectrum scarcity,” [13] W. Tranter, K. Shanmugan, T. Rappaport and K. Kosbar,
IEEE Spectrum, vol. 41, no. 3, March 2004. Principles of Communication Systems Simulation with wireless
[4] S. Haykin, “ Cognitive radio: Brain-empowered wireless application: Pearson publication, Ist edition.
communications,” IEEE Journal on Selected Areas in [14] A.A. Abu-Dayya and N.C. Beaulieu, “Outage probabilities in the
Communications, vol. 23, no. 2, pp. 201-220, 2005. presence of correlated log-normal interferers,” IEEE
[5] Q. Zhao and B. Sadler, “A survey of Dynamic Spectrum Transactions on vehicular technology, vol. 43, no. 1, February
Access,” IEEE Signal Processing Magazine, vol. 79, no. 3, pp. 1994,pp.164-173.
79-89 (2007). [15] Szyszkowicz, S.S.; Yanikomeroglu, H.; Thompson, J.S., "On the
[6] I. Akyildiz, W. Lee, M. Vuran and S. Mohanty, “NeXt Feasibility of Wireless Shadowing Correlation
generation/dynamic spectrum access/cognitive radio wireless Models," Vehicular Technology, IEEE Transactions on , vol.59,
networks: A survey,” Computer Networks (Elsevier), vol. 50, pp. no.9, pp.4222,4236, Nov. 2010
2127-2159, Sep 2006. [16] R.Menon, R. R.M. Buehrer, and J. Reed, “Outage probability
[7] C. Stevenson and et.al., “IEEE 802.22: The first cognitive radio based comparison of underlay and overlay spectrum sharing
wireless regional area network standard,” IEEE Communication techniques,” in Proc. IEEE International Symposium of
Magazine, vol. 47, no. 1,pp 130-138, Jan 2009. Dynamic Spectrum Access Networks (DYSPAN), pp. 101-109,
Nov. 2005.
[8] X. Hong, C. Wang, H. Chen and Y. Zang, “ Secondary Spectrum
Access Networks,” IEEE Vehicular Technology Magazine, vol. [17] C.C. Chai and Y.H. Chew, “Power control for cognitive radios in
4, no. 2, pp 36-43, June 2009. nakagami fading channels with outage probability requirement”,
Conference Proceedings, IEEE Globecom, 2010.
[9] Finnish Communications Regulatory Authority (FICORA),
"Demand for radio frequencies in the future", 2010. [18] S. Singh, P.D. Teal, P.A. Dmochowski, and A.J. Coulson.
“Power allocation in underlay cognitive radio systems with
[10] N. Hoven and A. Sahai, “Power scaling for cognitive radio,” in feasibility detection”, In Proc. 13th Australian Comms. Theory
Proc. WirelessCom Symposium on Emerging Networks, Workshop, pages 135-139, Wellington, NZ, January 2012.
Technologies and Standards, June 2005.
[19] C.-X. Wang, X. Hong, H.-H. Chen, and 1. S. Thompson, "Uplink
[11] E. Larson and M. Skoglund, “Cognitive Radio in a frequency- capacity of cognitive radio networks with interference outage
planned environment: Some basic limits,” IEEE Transaction on constraints”, Proc.. IEEE ICCCAS 2008.
wireless communication, vol. 7, no. 12, December 2008.
[12] M. Vu, N. Devroye and V. Tarokh, “On the Primary Exclusive
Region of Cognitive Networks,” IEEE Transaction on Wireless
Communication,vol.8,no.7,July 2009.

Das könnte Ihnen auch gefallen