Anda di halaman 1dari 6

2011 Wireless Advanced

Comparison of the Rayleigh and Nakagami Fading Channels


MIMO Multicarrier System
Vahid Fotohabady, Fatin Said,
Centre for Telecommunications Research, King's College, Strand London, WC2R 2LS, UK,
[vahid.fotohabadi, fatin.said,]@kcl.ac.uk
Abstract - The combination of orthogonal frequency division
multiplexing (OFDM) signal processing and multiple-input
multiple-output (MIMO) is regarded as a promising solution for
enhancing the data rates of next-generation wireless
communication systems operating in frequency- selective fading
environments. This alteration of parameters is based on the active
monitoring of several factors in the external and internal radio
environment, such as radio frequency spectrum, user behavior and
network state. It may be tempting to suggest that the solution of
this problem lies in simply increasing the transmission power level
of each subserviced transmitter. However, increasing the
transmission power level of any one transmitter has the
undesirable effect of also increasing the level of interference to
which the receivers of all the other transmitters are subjected. In
this paper we have explained and simulated several fading
channels such as Nakagami fading, Ricean fading and Rayleigh
fading channel. Also we have compared path gain level in the
Rayleigh and Nakagami fading channels. The results show that the
consumed power in Nakagami sub-channels is less than consumed
power in Rayleigh sub-channels.

work remains to investigate the performance of MIMO systems in


practical scenarios and to find receiver structures which offer an
acceptable trade-off between complexity and performance [4]. The
main challenge in developing reliable high data rate mobile
communications systems is to overcome the detrimental effects of
frequency-selective fading in mobile communications channels.
These previously proposed OFDM transmitter diversity systems all
require a cyclic prefix to be added to the transmitted symbols to
avoid ISI and inter channel interference (ICI) in the OFDM
symbols, and the number of cyclic prefix symbols has to be equal
to or greater than the order of the wireless channels [4]. So
combining of these two schemes offers immunity to ISI and high
capacity. STBC-OFDM assumes that channel gain is constant
during the n OFDM symbols in nTX transmit antennas system. It

has weakness under time selective channels; which can be


regarded as an advantage under frequency selective channels. The
remainder of this paper is organized as follows; channel model is
presented in section two, Fading Channel Rayleigh and Nakagami
are illustrated in section three, followed by impulse response
multi-path fading channel details in section four .The section five
of our paper describes the path loss of the channels such as
Keywords: MIMO, OFDM, Nakagami, Rayleigh, Fading
Rayleigh and Nakagami, the section six is dedicated to the
channels
simulation results and finally in section seven our contributions are
I. INTRODUCTION
concluded.
II. CHANNEL MODEL
A wireless network consists of a group of nodes that communicate
with each other over a wireless medium of propagation. When the In this section we consider a MIMO channel. A point-to-point
channel between two nodes source: Signal Radio Control (SRC)
communication system of NT transmitter and N R receiver
and destination: Destination host (DST) of such a network is in a
deep shadow-fading state, increasing the transmit power by the antennas is shown in Figure 1.
SRC can be too power consuming and result in interference for
other co-channel receivers. In this situation a Collaborating node
(CN) may be used to relay the packets between the SRC and the
DST [1]. We consider a protocol in which the collaborating node
is selected from a number of nearby nodes to the SRC, such that it
has the host propagation channel to the DST. Recently several
authors have considered the problem of imperfect channel state
information and its effect on bit and power allocation we show that
when an extra node becomes involved in transmitting the power,
the total average power consumed in the collaborative network is
lower than that of a standard wireless network in the Rayleigh
channel. We consider two models for the channels in our analysis;
i) Rayleigh fading channels, ii) Nakagami fading channels
,shadowing channels. Let PR & PN be the power of the SRC in
the direct channel (single-hop) of a standard wireless network,
respectively, of the CN and the SRC in the collaborative network,
for the two systems to yield the same received power at the DST.
Also MIMO technology has attracted attention in Wireless
communications [2, 3] since it offers significant increase in data
throughput and link range without additional bandwidth or
transmit power. The advantage of MIMO systems has raised
prospects for achieving large increases in system throughput. The
theoretical performance gains are very large, but considerable

978-1-4577-0109-2/11/$26.00 2011 IEEE

295

Figure 1.point-to-point communication system of

NT transmitter and N R

receiver antennas

This system can be represented by following discrete time

model

th

receive antenna,
gain from the m transmit antenna to the n
and they are assumed to be independent and identically distributed
(I.I.D)circularly symmetric complex Gaussian random variables
with zero-mean and unit-variance, having uniformly distributed
phase and Rayleigh or Nakagami distributed magnitude. A
circularly symmetric complex Gaussian random variable is a
th

y1 h11 " h1Nt x1 n1


# = # %
# # + # (1)

y
h

h
x
n
NrNt Nt Mr
Nr Nr
Or simply as y = H x + n , here

x represents the N t dimensional random variable a + jb c (0,

) ), in which a and b

transmitted symbol, n is the N R dimensional noise vector, and H are I.I.D real Gaussian distributed as (0, / 2) . A commonly
used channel model in MIMO wireless communication systems is
is the Nt N r matrix of channel gains. h11 is the gain from transmit a block fading (also known as quasi-static) channel model where
antenna 1 to receive antenna 1 . We assume a channel bandwidth the channel matrix elements, which are I.I.D complex Gaussian
of B and complex Gaussian noise with zero mean and covariance (Rayleigh or Nakagami fading) random variables, are constant
over a block and change independently from block to block. We
2
2
matrix n I Nr , where typically n = N 0 B . For simplicity, given drop the index k for the channel gain. The elements of the additive
a transmit power constraint P we assume an equivalent model noise vector
T
2
with a noise power of unity and transmit power P / n = ,
n( K ) = [n (K ), n ( K ),..., nN ( K )]
(7)

where can be interpreted as the average SNR per receiver are assumed to be also white I.I.D complex Gaussian random
antenna under unity channel gain. This power constraint implies variables with zero-mean and unit-variance. From this
normalization of noise power and channel loss, the averaged
that the input symbols satisfy [5];
transmitted power which is equal to the average SNR at each

Nt
*
E [ xi xi ] =
i =1

receiver antenna is to be no greater than NT [4, 5]. Now we

(2)

consider the Nakagami fading model for the 2 transmitter and 2


In the system design, we considered both discrete Rayleigh and receiver cases which we use in our simulations. The generalization
Nakagami fast fading MIMO channel models, where NT to more than 2 transmitter or receiver antennas is straightforward.
The Ricean channel model is given by [6]
and N R
are the number of transmit and receive antennas
respectively. In the transmitter, a data stream is demultiplexed

H=

into NT independent sub streams. Each sub stream is encoded into


transmit symbols using a modulation scheme (e.g. BPSK, QPSK,
M-QAM, etc.) at symbol rate 1/T symbol/sec with synchronized
symbol timing [4, 5].
The Rayleigh flat fading model is described here. The baseband

N R dimensional received signal vector


.

T
r ( K ) = [ r1 ( K ), r2 ( K ), ...rN R ( K ) ]
at sampling instant k may be expressed as
r ( K ) = H .x ( k ) + n ( K )
where, the formula is too crunched

th

e j 1 1
HF =
j 2 1
e

e
e

j 1 2

j 2 2

(8)

(9 )

(3)
(4)

fading matrix as described before, for instance the matrix element


hij denotes the Nakagami flat fading channel coefficient from

(6)

The transmit symbol vector with equally distributed transmitted


power across the transmitted signal. Here, the superscript T is
transposition. H denotes the N T N R
channel matrix, whose
elements h m n at the

where,

is a fixed matrix (i.e. fixed for each block of a block fading


channel) consisting of phase elements corresponding to line-ofsight (LOS) component and
is the Nakagami or Rayleigh flat-

h11 " hN 1
r1(K)
n1(K)
T

H= # % #
,.r(K) = # X(K) =,n(K) = # .(5)
h " h
rNR(K)
nNR(K)
NTNr
1Nr

T
X ( K ) = [ X1 ( K ), X 2 ( K ),..., XNT ( K )

11
e 12
K
K e
1
1 h11 h12
HF +
HV =
j21
j22 +
K +1
K +1
K +1
e e K +1 h21 h22

row and nth column is the channel

transmitter i to transmitter j. K is the K-factor of the Nakagami


distribution which is proportional to the strength of the LOS
component. The Nakagami model was suggested before as a first
order approximation to the Nakagami channel. We have decided to
use this model in our simulations because under certain reasonable
assumptions it gives an accurate approximation to the Nakagami
channel. When using this model the following assumptions are
valid: (i) the distance between the antennas is sufficiently large;
(ii) the subscriber unit is mobile and possibly changing
orientations.
With respect to the base station, under these assumptions it is
reasonable to state that the phases of the LOS component arriving
at the different receiver antennas are random and uncorrelated.
In this work the multipath delay and fading are generated based
on the Jakes fading model. Using this model we composed the

296

channel impulse response of complex samples using uniformly


distributed Rice and Nakagami fading channels. The channel
impulse response can be described as:

f (r ) =

hk = N (0,
) + jN (0,
),
2
2

1
2
= E [ R ], m =

2
2
E[ R ] 2

(13)

(10)

where,

2
2
k = 0 exp(- kTS / Trms )
2
0 = 1 exp( TS / Trm s )

2 m me r 2 m 1 ( m / ) r 2
e
( m ) m

In M fading Nakagami channel when m is equal to 1, the system


becomes equivalent to Rayleigh model. The parameter can also
be regarded as the average power fading.

(11)

III. FADING CHANNEL RAYLEIGH AND NAKAGAMI:


FADING CHANNEL RAYLEIGH,

Rayleigh channel has probability density function with the


Rayleigh distribution in formula (12)

P( r )

r2

r e 2 2
= 2
0

(0 r )

(12)

(r < 0)

In (12) is average power when the received signal before


the detection, is root mean square (rms) of the value of the
X
Figure 3. Probability density function-Nakagami
received signal before the detection and r is amplitude cover of
the received signal. [4] The Rayleigh distribution probability
Figure (3) shows the distribution probability density function
density function is shown in Figure 2,
Nakagami for m = 1.5. Formula (13) and finally Figure 4 shows
the Comparison Probability Density Function- for Rayleigh and
Nakagami channel.
2

r
Figure 2.Rayleigh distribution probability density function

Figure 4.Nakagami probability density function (-) k=1/2,


(--) k=1(Rayleigh), (-.-) k=2, (o) k=3

FADING CHANNEL NAKAGAMI


IV- IMPULSE RESPONSE MULTI-PATH FADING CHANNEL
In this section the M fading Nakagami channel is described. Phase
of the unknown channel is considered. Fading Nakagami model is The impulse response of multi path channel with complex low
best fit model for multi-path channels in urban area [4].
pass filter is [7]

297

h (t , ) =

(t ) e j k ( t ) ( k (t ))

conventional radio channels are used to shape can form of


exponential decrease equation (18) described

(14)

Where k ( t ) 2 f c k ( t ) shows domain and path delay time


in k th channel, k

(t ) and k (t )

h ( k ) = Ce

k
rms

(18)

are path phase and time Let assume the channel sample period ( s ) is in discrete time.

delay in the k th channel. In the communication systems where f c The above equation can be rewritten as shown in equation (19)
are very large, a little delay causes a great phase rotation. In each

k
path, the signal with the local instruments scattered around the
rm s
(19)
h ( k ) = C e
k = k
reflective approach and makes it appear to the recipient. Since this
s
scatters are random process, so amplitude and phase of each path
can be supposed as a randomized variable and therefore the where the delays k uniformly distributed on the
channel
impulse response channel h (t , ) is also considered as a random sample period. Delay range can cause symbol interference
process. Fading random process of channel characteristics is such that, digital symbols become adjacent to the delay . The
important to determine channel treated [8]. A conventional method different paths have been interference overlaid and are over
in estimating the channel randomly supposing scatters number is lapping. Number of symbols found in a single system single
high therefore could be using the central limit theorem. This carrier interference is the following [10]
process for some complex Gaussian is obtained channels impulse
response. As mentioned if there is a direct path between sender

and receiver there is no channel amplitude impulse response with


N ISI , Single Carrier = max
(20)
Rayleigh distribution, and the range of distribution channels with
Ts

Rice will be in phase with both uniformly distributed in the


interval (0, 2 ) is. Assuming arrest Wide Sense Stationary (WSS)
of channel process, the correlation function as its second equation where Ts is the symbol period. Since is The value of Ts is
very small in a very high rate, the effect of ISI increase. Thus
(15) is defined
the complexities of implementation mechanisms are
1
compensation, to achieve optimal performance as a significant
h( 1, 2,t)= E[h*( 1,t) h ( 2,t +t)]
(15)
increase.
If the symbol period delay channel posts bigger than
2
maximum of range ( max ), the Channel ISI can create less. In
(3.9)
the multicarrier posts method, each period symbol with the
number of carriers is increased therefore the ISI is reduced.
Thus the number of interfering symbols In the multicarrier
systems is as follows

t = 0 .The Wide Sense

Let assume the distribution for

Stationary un-correlated Scattering (WSSUS) is a model


for multi-channel path .Here lose of the power delay
spectrum or the
intensity profile is called multi-path
channel as shown in (16)

h ( ) = k 2 ( k )

(16)

This function, in terms of average power output channel delay


determines . Find the limit of Opposed to its value is zero range
delay and says a few channel path average values, maximum and
rms it is important. Channel parameters using profile are
identified include [9]:
p0

: Normalized power received to the ratio of power that can be


sent,
K: Rican coefficient of line of said
rms : The amount of delay range multi path,
: profit power path, that using the following relationship is
reached:

rms =
2

P0 = k
k

mi2i

m=
i

P0

k2,max
k=
P0 k2,max

m=1, 2
(17)

N ISI , Single C arrier = m ax


N C Ts

A little remaining ISI can be removed by using guard interval


(GI)[11].

V- CELLULAR LAYOUT PATH LOSS OF THE CHANNELS


SUCH AS RAYLEIGH AND NAKAGAMI
A wireless communications network simulation scenario with
7 cells has been considered. This scenario is shown in Figure
5. In this arrangement, a central cell and six hexagonal cells
around it are assumed. In the central of any cells there is a
base station and also in any cells there is a mobile node as
well. The distance between mobile node and base station in
each cell are different. The radius of each cell is assumed to be
500 meters and also distance between any two adjacent base
stations is set to 2 km. The frequency channels in adjacent
cells are assumed to be different and non neighbors cells are
assumed as co channel. Channels fading are set as Rayleigh
and Nakagami. Gain of each mobile node can be calculated as
following:

Normally the first signal with zero delay at the receiving point is
considered as a maximum range. Delay profiler for a

298

(2 1 )

g i = 0.0097 / d i4 .where the

di is distance

between mobile node and its base station. We also indicate all
base stations in the system by a unique index

b {0,..., B 1} and all users in the system by a unique


index K {0,..., K 1} . The cellular layout spans the interval

N 0 = 2 * 1 0 1 5 considered. According to the previous formula, if

the distance between users with their base station in each cell is
500meters, the power path gain will be a minimum at

[0 B] in the real line, where one base station is located at the


spatial coordinate b+1/2 and user k is located at a point
belonging to a regular uniform grid of points such that the
number of users per cell is K B
for all cells and users are
symmetrically located in each cell. The channel-path gain for a
transmitter-receiver pair located at U ,V [0, B] is given by

g (u, v) =

G0
1 + dB (u, v) / )

around ( 0 .00 0 16 * 10

) . Meanwhile, if this distance is 100

meters, the power path gain will be s around (0.097*109 ) . In


Figure 6 and 7 it can be seen that the mobile node in cell 2 has the
path gain level of Profile can route one path gain better than the
rest of users and user index than other users, the farthest distance
of the station per cell itself and thus losses and more profile can be
a less gain path power. Therefore by game theory we can adjusting
and optimize the resource allocation in the system.

(22)

where we define the modulo distance

dB(u,v) = min zEZ u v + Bz

(23)

This has the advantage of eliminating the boundary effects.


The parameters G0 , and in (22) denote the SNR at the
cell centre, the path-loss exponent and the loss distance. The
function (22) can approximate path loss models used where

gains are 1 / d but are clipped to a maximum value for


d below some minimum distance [12]. All systems operate
under strict per base station power constraints, where
pb[ f ] 0 is the maximum power that station b can use on
subchannel F. The powers are relative to the SNR given by
(22), assuming a normalized noise power spectral density
equal to 1, where

1
F

F 1
f =0

pb[ f ] = 1 .

Figure 6. Path Gain of each subchannel in Channel fading


Rayleigh

Based on the model explained in section III and IV, Figure 7


shows simulation results of fading Nakagami channels with
different m values. If m=1, Nakagami fading channel becomes
Rayleigh fading channel as stated previously. For any other value
of m, the performance of Nakagami sub-channels is better than
Rayleigh sub-channels. Also according to the simulations
performed for M =2, consumed power in Nakagami sub-channels
is less than consumed power in Rayleigh sub-channels. This means
that the overall power consumption is reduced. The other hand in
the equal condition, user can send value data rate by fading
channel Nakagami equal the fading channel Rayleigh but the
overall power consumption is reduced.
Figure 5.Wireless communications network simulation scenario.

VI- SIMULATION RESULTS


Figure 6 illustrates the simulation results for m=1 or Rayleigh
fading channel. We assume that the fading type is non-frequency
selective.
Background
noise
for
all
users
is

299

[4]

[5]
[6]
[7]

[8]
[9]

[10]

[11]

Figure 7. Path Gain of each subchannel in Channel fading


M=2

Nakagami

VII- CONCLUSION

[12]

[13]

For spatially uncorrelated MIMO systems, transmitter diversity


is one of the most effective techniques, in particular, for OFDM
based systems. We explained and simulated several fading
channels such as Nakagami fading, Ricean fading and Rayleigh
fading channel. We also compared path gain level in the Rayleigh
and Nakagami fading channels. The simulation is performed for
both M=1 and M=2, and the results showed promising overall
power consumption reduction. It can be concluded that the
consumed power in Nakagami sub-channels is less than consumed
power in Rayleigh sub-channels. Fading Nakagami model is the
best fit model for multi-path channels in urban area. Therefore the
employment of these transmission plus using the OFDM systems,
additional diversity gain the measured delay and fading caused
by multipath and Nakagami fading channels, improve system
performance by decreasing the consumed power.
ACKNOWLEDGEMENT
The authors would like to thank Dr. R.Delmaghani for giving
constructive feedbacks and would also like to thank Dr H.
Bovarshad for his suggestions.

REFERENCE
[1] X .Gao and M. Naraghi, Computationally Efficient Resource
Allocation for Multiuser OFDM Systems, IEEE, Wireless
Communications and networking conference, 2006.
Increasing uplink
[2] [2] A. Sendmaris, E. Erkip, and B. Aazhang,
capacity via user cooperation diversity, Proceedings of
IEEE
International Symposium on lnformation Thcory, pp. 156, 1998.
[3] Muayad S. Al-Janabi, Charalampos C. Tsimenidis, Bayan S. Sharif,

300

and Stephane Y. Le Goff , Simulated and Semi-Analytical


Throughput Evaluation for AMC-OFDMA Systems , IEEE
Communication Systems Networks and Digital Signal Processing
(CSNDSP) ,7th International Symposium,2010.
Marvin K. Simon and Mohamed-Slim Alouini, Digital
Communication over Fading Channels, Second Edition, A John
Wiley and Sons, Inc, 2005.
A. Goldsmith, Wireless Communications, published by
Cambridge University Press.2005.
Theodore S. Rappaport , Wireless Communications: Principles and
Practice, 2nd Edition, Prentice Hall, pp. 176, 2001.
S.M. Alamouti, A simple transmit diversity technique for wireless
communications, IEEE J. Sel. Areas Common., vol.16, no.8,
pp.14511458, Oct. 1998
M.R. Nakhai, Multicarrier transmission, IET Signal Processing,
vol. 2, no. 1, pp. 1-14, Oct. 2008.
Deniz Gunduz, Andrea Goldsmith and H. Vincent Poor_,
Diversity-Multiplexing Tradeoffs in MIMO Relay Channels , IEEE
Global Telecommunications conference, 2008.
King F. Lee, Douglas B.Williams Bandwidth Efficient OFDM
Transmitter Diversity Techniques IEEE Acoustice, Speeh and
Signal Processing (ICASSP) , International Conference,2010.
R. Zakhour and D. Gesbert, Optimized data sharing in multicell
MIMO with finite backhaul capacity, Jan 2011.
3GPP, 3GPP TR 25.996 V6.1.0: Spatial channel model for Multiple
Input Multiple Output (MIMO) simulations (Release 6). 3GPP, Sept
2003.
Zhu Han, Zhu Ji, and K. J. Ray Liu, Power Minimization for MultiCell OFDM Networks Using Distributed Non-cooperative
Game
Approach, IEEE Global Communications conference, 2004.

Anda mungkin juga menyukai