125 CHAPTER 5 MIMO SC_FDMA WITH ICI ESTIMATION AND SUPPRESSION 5.1 INTRODUCTION Since past few decades different types of cellular networks were launched and went successful on the radio links such as WiMAX, that became very popular because of its high data rate (70Mbps) and support for providing wireless internet services over 50km distance. The Universal Mobile Telecommunication System(UMTS) Long Term Evolution (LTE) is an emerging technology in the evolution of 3G cellular services. LTE runs on an evolution of the existing UMTS infrastructure already used by over 80 percent of mobile subscribers globally. We have very limited resources in cellular technologies and it is important to utilize them with high efficiency. Single Carrier Frequency Division Multiple Access (SC-FDMA) & Orthogonal Division Multiple Access (OFDMA) are major part of LTE. OFDMA was well utilized for achieving high spectral efficiency in communication system. SC-FDMA is introduced recently and it became handy candidate for uplink multiple access scheme in LTE system that is a project of Third Generation Partnership Project (3GPP). The Multiple Access Scheme in Advanced Mobile radio system has to meet the challenging requirements, for example high throughput, good robustness, efficient Bit Error Rate (BER), high spectral efficiency, low delays, low computational complexity, low Peak to Average Power Ratio (PAPR), low error probability
28
Embed
CHAPTER 5 MIMO SC FDMA WITH ICI ESTIMATION AND …shodhganga.inflibnet.ac.in/bitstream/10603/16442/10/10_chapter5.pdf · OFDMA of LTE physical layer by considering different modulation
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
125
CHAPTER 5
MIMO SC_FDMA WITH ICI ESTIMATION AND
SUPPRESSION
5.1 INTRODUCTION
Since past few decades different types of cellular networks were
launched and went successful on the radio links such as WiMAX, that became
very popular because of its high data rate (70Mbps) and support for providing
wireless internet services over 50km distance. The Universal Mobile
Telecommunication System(UMTS) Long Term Evolution (LTE) is an
emerging technology in the evolution of 3G cellular services. LTE runs on an
evolution of the existing UMTS infrastructure already used by over 80
percent of mobile subscribers globally. We have very limited resources in
cellular technologies and it is important to utilize them with high efficiency.
Single Carrier Frequency Division Multiple Access (SC-FDMA) &
Orthogonal Division Multiple Access (OFDMA) are major part of LTE.
OFDMA was well utilized for achieving high spectral efficiency in
communication system. SC-FDMA is introduced recently and it became
handy candidate for uplink multiple access scheme in LTE system that is a
project of Third Generation Partnership Project (3GPP). The Multiple Access
Scheme in Advanced Mobile radio system has to meet the challenging
requirements, for example high throughput, good robustness, efficient Bit
Error Rate (BER), high spectral efficiency, low delays, low computational
complexity, low Peak to Average Power Ratio (PAPR), low error probability
126
etc. Error probability is playing vital role in channel estimation and there are
many ways to do channel estimation, like Wiener Channel Estimation,
Bayesian Demodulation etc.
This chapter investigates the performance of SC-FDMA and
OFDMA of LTE physical layer by considering different modulation schemes
(BPSK, QPSK, 16QAM and 64QAM) on the basis of PAPR, BER, power
spectral density (PSD) and error probability by simulating the model of SC-
FDMA & OFDMA. Additive White Gaussian Noise (AWGN) and
frequency selective (multipath) fading is introduced in the channel by using
Rayleigh Fading model to evaluate the performance in the presence of noise
and fading.
A set of conclusions is derived from our results describing the
effect of higher order modulation schemes on BER and error probability for
both OFDMA and SC-FDMA. The power spectral densities of both the
multiple access techniques (OFDMA and SC-FDMA) are calculated and
result shows that the OFDMA has high power spectral density. The
considered modulation schemes also have a significant impact on the PAPR
of both OFDMA and SC-FDMA such that the higher order modulations
increase PAPR in SC-FDMA and decrease PAPR in OFDMA. However, the
overall value of PAPR is minimum in SC-FDMA for all modulation schemes.
SC-FDMA is an OFDMA alternative technology. SC-FDMA is the
multiuser version of single carrier modulation with frequency domain
equalization (SC/FDE). The main objective of SC-FDMA is to introduce
transmission with lower PAPR than OFDMA but it is very sensitive to phase
noise and Carrier Frequency Offset (CFO), which will break the orthogonality
among subcarriers and generate Common Phase Error (CPE) as well as inter-
carrier interference (ICI) to distort the received signals. Also this is similar
situation in other OFDM communication systems. Thus the BER performance
127
and throughput are seriously degraded. To improve system performance and
throughput, it is necessary to analyze the effects of phase noise, CFO and then
suppress the interferences. To overcome this problem, the SC-SFBC scheme
used here with modified FFT Algorithm. Single Carrier- Space Frequency
Block Coding (SC-SFBC) is used to reduce the peak to average power ratio
(PAPR) of the Multiple-Input Multiple-Output (MIMO) SC-FDMA signal.
Modified FFT algorithm is used to reduce the memory references thereby
increasing the throughput. The coding method is different from the Alamouti
scheme, the four frequencies involved in SC-SFBC scheme do not use
successive subcarriers any longer. Thus it avoids the degrading interactions
between phase noise and CFO estimations. The proposed structure is
evaluated using performance parameters such as BER & SNR. Structural
realization and analysis pertaining to Timing, Power and Throughput are
implemented in Virtex-4 and analysis is carried out in Altera respectively
5.2 PAPR IN OFDM SIGNAL
OFDM signals have high PAPR and are thus not power efficient .
High PAPR values lead to severe power penalty problem at the transmitter,
which is not affordable in portable wireless systems where terminals are
battery powered. To avoid this problem, PTS approach is introduced by
achieving high PAPR reduction. However, it causes high system complexity
and computational complexity which make it difficult to employ for OFDM
system. So, the modified FFT technique (Nicola et al 2005) is proposed to
lower the computational complexity while maintaining the similar PAPR
reduction performance compared with the ordinary PTS technique.To reduce
PAPR further, SC_SFBC technique can be combined with modified FFT
scheme. MIMO systems may be implemented in a number of different ways
to obtain either diversitygain or capacity gain. A modified FFT combined
with SC-SFBC technique to improve PAPR reduction performance in OFDM
128
and MIMO-OFDM is presented in this chapter. The performance of the
modified FFT combined with SC_SFBC is evaluated to maximize.
In OFDM modulation technique, a block of N data symbols, ,nX
0,n 1, …, 1N , is formed with each symbol modulating the corresponding
subcarrier from a set ,nf 0,n 1, …, 1N where N is the number of
subcarriers. The N subcarriers are chosen to be orthogonal, i.e. nf n f ,
where 1/f NT and T is the original symbol period.
The resulting baseband OFDM signal x t of a block can be
expressed as
11 2
0n
N j f tx t X enN n, 0 t NT (5.1)
The PAPR of the transmitted OFDM signal x(t), is the ratio of the
maximum to the average power and given as
0 0
0
2 2( ) ( )m ax m a x
2 1 2t N T t N T
N T
x t x tP A P R
E x t x t d tN T
(5.2)
The PAPR of the continuous-time OFDM signal cannot be
precisely computed at the Nyquist sampling rate, which corresponds to N
samples per OFDM symbol. So, in discrete-time systems, instead of reducing
the peak of the continuous-time signal i.e. max ( )x t , it is better to reduce the
maximum amplitude of LN samples of ( )x t , where parameter L denotes the
oversampling factor. This is due to the fact that the PAPR of a continuous
time OFDM signal cannot be precisely described by sampling the signal using
N samples per signal period where some of the signal peaks may be missed.
129
So oversampling is usually employed ( 1)L for better approximation of true
PAPR. The case 1L is known as critical sampling or Nyquist rate sampling.
Oversampling is implemented by LN -point IFFT of the data block with
( 1)L N zero padding.
The continuous PAPR of ( )x t is approximated by its discrete LN
samples i.e. nTxLN
, which is obtained from the LN –point IFFT of nX with
( 1)L N zero-padding can be represented as
11 2 /( )0
N j kn LNx n X enN n, 0,1,..., 1k NL (5.3)
To evaluate accurately the PAPR reduction performance from the
statistical point of view, the Complementary Cumulative Distribution
Function (CCDF) of the PAPR is used. It denotes the probability that the
PAPR of OFDM symbol exceeds a certain threshold 0PAPR .
( ( ( ))) Pr( ( ( ))) 0CCDF PAPR x n PAPR x n PAPR (5.4)
Due to the independence of the N samples, the CCDF of the PAPR
of a data block at Nyquist rate sampling is given by
0( ( ( ))) 1 (1 )PAPR NCCDF PAPR x n e (5.5)
Therefore, the CCDF of PAPR of L-times oversampled OFDM
signal can be defined as
00( ( ( ))) Pr( ( ( )) ) 1 (1 )PAPR LNCCDF PAPR x n PAPR x n PAPR e
(5.6)
130
5.3 SPATIAL BLOCK CODES
5.3.1 STBC
Space Time Block Coding (STBC) is the most straight forward
approach for applying Alamouti coding. In this case, precoding involves four
frequency components ( ( ), ( ), ( )and ( )) to be sent to the
antennas over four successive time intervals. As presented in Table 5.1,
Alamouti STBC is performed between k-th frequency component ( ) of
FFT output vector ( ) at time 4j, k-th frequency component ( ) of
FFT output vector ( ) at time 4j+1, k-th frequency component ( )
of FFT output vector ( ) at time 4j+2 and k-th frequency component( ) of FFT output vector ( ) at time 4j+3. As a result for each
subcarrier, the precoded components are mapped onto four consecutive SC-
FDMA blocks and the frequency structure of the signal from one block to
another is not altered. Because complex conjugation and or sign changes do
not break the low PAPR, the signals sent to the four transmit antennas are
single carrier signals.
Table 5.1 STBC Precoding
k-th subcarrier Time 4j Time 4j+1 Time 4j+2 Time 4j+3
Tx1 ,( )
= ( )
,( )=( ( ))
,( )=( ) *
,( )=( ( ))*
Tx4 ,( )=( )
,( ) =( ( ))*
,( )=( ( ))*
,( )=( ( ))*
Tx3 ,( )=( )
,( )=( ( ))*
,( ) =( ( ))*
,( )=( ( ))
Tx4 ,( )=( )
,( )=( ( ))*
,( )=( ( ))
,( ) =( ( ))*
131
5.3.2 SC- SFBC
Since Space frequency precoding scheme does not alter the PAPR
of SC-FDMA, SFBC will not be able to achieve diversity gain and improves
system capacity in MIMO SC-FDMA system. The coding scheme of SFBC
scheme is described in the Table 5.2.
Table 5.2 SFBC Scheme
Frequency 4k
Frequency 4k+1
Frequency 4k+2
Frequency 4k+3
Tx1 =X4k =X4k+1 =X4k+2 =X4k+3
Tx2 - X4k+1) (X4k)* -X4k+1)* =(X4k+2)*
Tx3 X4k+2)* -X4k+1)* (X4k)* =-X4k+1)
Tx4 =-X4k+3)* (X4k+2)* -(X4k+1) (X4k)*
SC-SFBC is used in MIMO SC-FDMA system, to overcome the
above problem. The coding scheme of SC-SFBC is entirely different from the
Alamouti scheme. It is described in Table 5.3. In SC-SFBC four frequencies
are involved, but they do not use successive subcarriers any longer. The
subcarriers K, K’, K’’ and k’’’= (p-1-k) mod S
where p is the even integer and S is the FFT spreading size.
Table 5.3 SC-SFBC Scheme
Frequency 4k
Frequency 4k+1
Frequency 4k+2
Frequency 4k+3
Tx1 =Xk =Xk’ =Xk’’ =Xk’’’
Tx2 -(Xk’) (Xk)* -(Xk’)* (X4k’’)*
Tx3 (Xk’’)* -(Xk’)* (Xk)* -(Xk’)
Tx4 =-Xk+’’’)* (Xk’’)* -(Xk’) (Xk)*
132
5.4 MIMO SC-FDMA
In MIMO SC-FDMA symbols are transmitted sequentially so that
the PAPR is reduced by spreading a symbol power over subcarriers. Also,
SC-FDMA in one mode introduces localized scheduling in which contiguous
subcarriers are assigned to an user. This makes mobile station more robust
of frequency offset than OFDMA, but the diversity order becomes lower
than OFDMA. The Block diagrams of MIMO SC-FDMA transmitter and
receiver are shown in the Figures 5.1a and 5.1b.
Figure 5.1a Block diagram of 4X4 MIMO SC-FDMA transmitter
133
Figure 5.1b Block diagram of 4X4 MIMO SC-FDMA receiver
The transmitter of an SC-FDMA system converts a binary input
signal to a sequence of modulated subcarriers. To do so, it performs the signal
processing operations like FFT shown in Figure 5.1a and Figure 5.1b. The
conventional N point FFT, requires (N/2)log2N multiplication operations and
Nlog2N addition operations. If it is N bit multiplier, it will generate 2N partial
products. Due to the partial products the delay is increased and it consumes
more power. Hence it will affect the throughput of the mobile system. Thus
there is a need for reducing the consumed power with decreased delay thereby
increase the throughput of the system. The proposed Modified FFT is the
combination of memory reference reduction technique, which eliminates the
redundant memory references, Binary scaling, converts floating point in to
fixed point and Radix 8 Multiplier, generates N/3 partial product.
134
Signal processing is repetitive in a few different time intervals and
resource assignment takes place in Transmit Time Intervals (TTIs). In 3gpp
LTE, a typical TTI is 0.5ms. The TTI is further divided into time intervals
referred to as blocks. A block is the time used to transmit all of subcarriers
once.
Let the Data sequence D= {d0, d1, d2...., dL-1,} be the input of FFT.
The Data spreads after the FFT. It is described in the Equation (5.7).
X = le-j2 kl /S= lpk, (5.7)
To improve the Channel capacity and diversity gain, the symbol
undergoes SC-FSBC encoding and it is described in the Equation (5.8).
Tx1 Tx2 Tx3 Tx4
Frequency X (-1)k+1Xk’ Xk’’ (-1)k+3Xk’’’*
Frequency k’ Xk’ Xk (-1)k’+2Xk’ Xk’’*
Frequency k’’ Xk’’ (-1)k’’+1Xk’ Xk (-1)k’’’+3Xk’
Frequency k’’’ Xk’’’ Xk’’ (-1)k’’’+2Xk’ Xk*
(5.8)
where XTx1, XTx2, XTx3 and XTx4 is the output of the SC-FSBC encoder for
Tx1, Tx2, Tx3 and Tx4 respectively
At the input to the transmitter, modulator transforms the binary
input to complex number Xn in one of several possible modulation formats
including Binary Phase Shift Keying (BPSK), Quaternary PSK (QPSK) and
16 level Quadrature Amplitude Modulation (16-QAM) The system adapts the
modulation format, and thereby increases the transmission bit rate, to match
the current channel conditions of each terminal. The transmitter next groups
the modulation symbols, Xn into blocks each containing N symbols. The first
135
step in modulating the SC-FDMA subcarriers is to perform an N-point Fast
Fourier transform (FFT), to produce a frequency domain representation XK of
the input symbols.
Several approaches for mapping transmission symbols XK to SC-
FDMA subcarriers are currently used. They are divided into two categories
namely distributed and localized as shown in Figure 5.2.
Frequency
Figure 5.2 Distributed and Localized Mapping
In the distributed subcarrier mapping mode, FFT outputs of the
input data are allocated over the entire bandwidth with zeros occupying the
unused subcarriers resulting in a non-continuous comb-shaped spectrum. As
mentioned earlier, interleaved SC-FDMA (IFDMA) is an important special
case of distributed SC-FDMA. In contrast with IFDMA, consecutive
subcarriers are occupied by the FFT outputs of the input data in the localized
subcarrier mapping mode resulting in a continuous spectrum that occupies a
fraction of the total available bandwidth. Distributed mapping makes use of
bandwidth spreading factor to introduce a parameter for interleaving the
136
allocated subcarriers of the user. Localized mapping maps the subcarriers
allocated to user, which are adjacent to each other.
XK then maps each of the N FFT outputs to one of the subcarriers
that can be transmitted. If all terminals transmit N symbols per block, the
system can handle Q simultaneous transmissions without co-channel
interference (where Q is the bandwidth expansion factor of the symbol
sequence) and N=(No of subcarriers)/Q. The result of the subcarrier mapping
is the set of XK (K=0,1,2,…..,M-1) complex subcarrier amplitudes. As in
OFDMA, an M-point inverse FFT (IFFT) transforms the subcarrier
amplitudes to a complex time domain signal XM. All the modulated symbols
are transmitted sequentially. The transmitted signal of each transmitter can be
written as
x1(n) = ej2 kn/N= ej2 kn/N
1) lpk,lej2 kn/N (5.9)
X2(n) = ej2 kn/N= ej2 kn/N
1) l*pk,l*ej2 kn/N (5.10)
X3(n) = ej2 kn/N= ej2 kn/N
1) l pk,l*ej2 kn/N (5.11)
X4(n) = ej2 kn/N= ej2 kn/N
1) l*pk,lej2 kn/N (5.12)
137
At the receiver, the received signal can be described in the
Equation (5.7).
y(n) 1 [xt(n)*ht,r(n) v(n)] ej[ (n)+2 fn] (5.13)
The receiver transforms the received signal into the frequency
domain, and then performs SC-SFBC decoding. Since SC-FDMA uses single
carrier modulation, it suffers from Inter Symbol Interference (ISI). The
equalized symbols are transformed back to the time domain via IFFT,
detection and decoding take place in the time domain.
5.5 ESTIMATION AND SUPPRESSION OF ICI
In MIMO SC-FDMA systems training based estimation is carried
out. In training based channel estimation algorithms, training symbols or pilot
tones that are known prior to the receiver, are multiplexed along with the data
stream for channel estimation. They rely on a set of known symbols
interleaved with data in order to acquire the channel estimate. The estimation
can be performed by either inserting pilot tones into all of the sub carriers of
SC-FDMA symbols, with a specific period, or inserting pilot tones into each
SC-FDMA symbol. This is accomplished with the help to two types of pilot
carriers. They are 1. Block type pilots 2. Comb type pilots
For a slow fading channel, where the channel is constant over a few
SC-FDMA symbols, the pilots are transmitted on all subcarriers in periodic
intervals of SC-FDMA blocks. This type of pilot arrangement depicted in
Figure 5.3, is called the block type pilot arrangement.
138
Figure 5.3 Block type Pilot arrangement
In this some sub-carriers are reserved for pilots for each symbol.
The channel estimates from the pilot subcarriers are interpolated to estimate
the ICI. The pilot arrangement used here is block type pilot arrangement
because channel is assumed to be constant over a single sub-carrier which is
very small. Based on block type pilot arrangement, ICI is estimated and
suppressed. ICI is calculated directly from the received pilot block by
reconstructing the ICI matrix and Least Square method, instead of estimating
ICI from the phase noise and CFO. Finally ICI is suppressed by taking the
inverse for ICI matrix.
5.6 LEAST SQUARE METHOD
The method of least squares is a standard approach for the
approximate solution of over determined systems. Least square means that the
overall solution minimizes the sum of the squares of the errors made in the
results of every single equation.
139
A regression model is a linear one when the model comprises a
linear combination of the parameter, i.e.,
f(xi, 1 j (xi (5.14)
Where the coefficients, j are function of (xi)
Letting
Xij( , ) (xi (5.15)
In this case the least square estimate (or estimator, in the context of
a random sample) can be seen, is given by
= (XT X) -1 XTy (5.16)
5.6.1 Estimation of ICI
Matrix representation of frequency domain signal vector is
Y = QM Rv + N (5.17)
Where QM is ICI matrix, Rv is received pilot block, N is AWGN noise.
Rv= [R0, R1, R2, ……, RN-1]T (5.18)
140
The ICI Matrix is given by
Q0 Q1 Q2 … QN-1
Q-1 Q0 Q1 … QN-2
QM = Q-2 Q-1 Q0 … QN-3
. . . … .
. . . .
. . . .
Q-(N-1) Q-(N-2) Q-(N-3) … Q0 (5.19)
By using FT property
Q-k j[2 (-k+ )n/N+ (n)]
j[2 (-k+ )n/N+ (n)+2 .n]
j[2 ((N-k)+ )n/N+ (n)]
= QN-k (5.20)
Now the ICI matrix is
Q0 Q1 Q2 … QN-1
QN-1 Q0 Q1 … QN-2
QM = QN-2 QN-1 Q0 … QN-3
. . . … .
. . . .
Q1 Q2 Q3 … Q0 (5.21)
QM is a circular matrix with only n different values. Frequency
domain signal vector of the received signal in matrix form is
Y = QM.Rv + N (5.22)
141
Q0 Q1 Q2 … QN-1 R0
QN-1 Q0 Q1 … QN-2 R1
QM = QN-2 QN-1 Q0 … QN-3 R2 + N
. . . … . .
. . . . .
Q1 Q2 Q3 … Q0 RN-1 (5.23)
QM consists of N unknown values, these values are easily calculated
by solving the equation set 5.23 and QM matrix can be reconstructed. By
multiplying the inverse QM matrix with the received signal, the Interference
caused by phase noise and CFO can be suppressed.
The above matrix equation can be rewritten as
Y = QM.Rv + N
= RM.Qv + N (5.24)
Where
R0 R1 R2 … RN-1 Q0
R1 R2 R3 … R0 Qv = Q1
RM = R2 R3 R4 … R1 Q2
. . . … . .
. . . . .
. . . . .
RN-1 R0 R1 … RN-2 Q N-1
(5.25)
N is the AWGN
142
where RM is a full rank matrix and Qv is the ICI vector, which contains the N
components in QM. Since the pilot carrier is known to the receiver, it can be
easily estimated from the received signal. Assuming that the channel
information is known to the receiver, the matrix RM and inverse matrix RM-1
can be calculated. Thus the ICI vector Qv is obtained by least square method
Qv = (RMT RM)-1 RM
T Y (5.26)
After that, the ICI matrix QM can be reconstructed from the
estimated Qv by very simple mapping process.
5.6.2 Suppression of ICI
During the pilot block Mapping the phase noise and CFO are
highly correlated, since the symbol period and frame size are usually very
short. Therefore phase noise and CFO of data block can be easily suppressed
by least square method
suppressed ( -1. .Y
R ( -1. (5.27)
5.7 RESULTS AND DISCUSSION
Code development and simulation was carried out on a system
running on Intel i3 configuration having 4GB RAM working in windows 7
Platform. The MIMO SC-FDMA Transceiver structure for determining the
constellation points and BER were coded & simulated in MATLAB the
results of which are presented & discussed in succeeding section. Structural
realisation of MIMO SC-FDMA Transceiver is implemented in Xilinx.
Analysis pertaining to timing and frequency is done in Altera. The
143
experimental results and analysis obtained using Matlab, Xilinx and Altera
are presented and discussed below.
The Modified FFT processor has been implemented using Xilinx
Virtex-4 FPGA. Based on the analysis, Device utilization summary is
tabulated in the Table 5.4. From the results, it can be observed that the
Modified FFT Algorithm requires only 10 percentage of slices, 11 percentage
of flip flops and 7.5 percentage of LUTs compared to the conventional FFT
Algorithm.
Table 5.4 Device utilization summary of FFT
Logic Utilization Conventional FFT
Algorithm Modified FFT
Algorithm
No of Slices 782 68
No of Slices Flip Flop 1040 75
No of 4 Input LUTS 1080 129
No of bonded IOBs 321 169
No of GCLKS 1 1
No of DSP 48s 32 12
The MIMO SC-FDMA has been implemented using Xilinx Virtex-
4 FPGA. Based on the analysis, Device utilization summary is tabulated in
the Table 5.5. From the results, it can be observed that the Modified FFT +
MIMO SC-FDMA requires only 24% percentage of slices, 5% percentage of
flip flops and 21% of LUTs.
144
Table 5.5 Device utilization summary MIMO SC-FDMA
Modified FFT +MIMO SC-FDMA
Logic Utilization Used Utilization
No of Slices 1489 24%
No of Slices Flip Flop 684 5%
No of 4 Input LUTS 2691 21%
No of bonded IOBs 25 21%
No of GCLKS 1 3%
The MIMO SC-FDMA described in section 5.4 has been analysed
using Altera Quarts II FPGA. Based on the results, the performance of the
system is compared and tabulated in the Table 5.6. Throughput of the mobile
system entirely depends on the power consumption and delay. The proposed
system consumes less power and short delay. Hence the Throughput of the