HA NOI UNIVERSITY OF SCIENCE AND TECHNOLOGY

----

PROJECT REPORT

IMPLEMENTING CHANNEL ESTIMATION FOR OFDM SIGNAL USING

NORMAL METHODS VIA HIPER STANDARD

Instructor: Assoc. Prof. Nguyễn Văn Đức

Group 3:
Đỗ Thu Hà 20161213 EEE K61

Tạ Thị Thanh Lâ m 20162333 EEE K61

Ha Noi,
11/2020

1
CONTENTS
TABLE OF FIGURES................................................................................................3
PREFACE................................................................................................................4
TABLE OF TASKS FOR MEMBER............................................................................5
CHAPTER 1: INTRODUCTION.................................................................................6
1.1 OFDM system..............................................................................................6
1.2 Priciple of pilot symbols insertion in time domain and frequency domain 8
1.3 Channel estimation through normal methods..........................................10
1.4 Interpolation techniques for channel estimation.....................................11
a. Interpolation using linear function, SI function and polynomial function
11
b. Interpolation using Wiener filter...........................................................13
1.5 Channel equalization for OFDM systems..................................................15
CHAPTER 2: SIMULATION CHANNEL ESTIMATION FOR OFDM SYSTEM ON
MATLAB SOFTWAVE...........................................................................................17
2.1 Code in Matlab..........................................................................................17
a. OFDM.m....................................................................................................17
b. Pilotsymbol.m........................................................................................19
c. OFDM_Modulator.m..............................................................................20
2.2 Simulation results......................................................................................22
CONCLUSION......................................................................................................24
REFERENCES........................................................................................................25

2
 TABLE OF FIGURES
Figure 1: OFDM system........................................................................................................ 6
Figure 2: Pilot symbol in time domain...........................................................................8
Figure 3: Pilot average subsystem.................................................................................. 9
Figure 4: Relationship between Doppler effect and channel................................9
Figure 5: Interpolatic techniques.................................................................................. 12
Figure 6: SI interpolation and cubic interpolation................................................13
Figure 7: Wiener filter........................................................................................................ 13
Figure 8: BER......................................................................................................................... 22

3
 PREFACE
With the extremely strong development of information technology in
general and electronics and telecommunications technology in particular,
the demand for electronic and telecommunication services is growing
rapidly and strongly. Therefore, it has design requirements for the
manufacture of system with low noise , less affected by external and low
BER that still meet the required function that is suitable for current use
needs. The subject “Digital Communication” helps us understand more
about Filters ,Telecomunications , channel estimations in general and
OFDM systems in particular. We would like to send our most sincere thanks
to Assoc. Prof. Nguyen Van Duc for his help us during the course. Due to
limited knowledge and limited time to learn practical experience, so our
project is inevitable shortcomings.

Finally, we would like to send our sincere thanks to the teachers in the
Institution, the students of previous year and especially to Assoc.Prof.
Nguyen Van Duc who enthusiastically guided and explained in detail
throughout the course of study to help us complete this topic.

We sincerely thank you!

4
 TABLE OF TASKS FOR MEMBER
No. Task Assign Support
Find Theorems and
1 Public articles relating to Lâ m Hà
project
Research the following
2 Hà Lâ m
documents and paper
3 Study the whole system Lâ m Hà

4 Study detailed block Hà Lâ m

Simulate channel
5 estimation for OFDM Lâ m, Hà
system in Matlab

6 Write presentation Hà Lâ m

7 Write report Lâ m Hà

5
 CHAPTER 1: INTRODUCTION

1.1 OFDM system

Bit source
Baseband Insert pilot
IFFT GI insertion DAC
Modulation symbol

Wireless
channel
Gaussian
noise
(AWGN)

Baseband Channel GI
FFT ADC
demodulation equalization separation

Separate
pilot symbol

Channel
estimation

Figure 1: OFDM system

Orthogonal Frequency Division Multiplexing (OFDM) is a transmission
technique used to achieve very high data rates. OFDM is the technology of
choice for all major wireless systems including Wireless LAN – 802.11,
WiMAX – 802.16, digital audio/video broadcast systems, and the air
interface evolution of 3G Wireless systems based on 3GPP and 3GPP2.
OFDM facilitates higher data rates over a wireless medium, which is very
exciting to wireless operators who are eager to deploy multimedia rich
Internet content over a wireless medium with seamless access anywhere,
anytime. This course describes key OFDM concepts and terminology. It
explains the challenges of radio propagation and describes how OFDM
overcomes these challenges to offer high data rates in a spectrally efficient
manner, and steps through the key OFDM operations in an end-to-end
transmission.

The OFDM overview system is shown in Figure 1. The bit source is

modulated by baseband modulation through some methods such as QPSK,
Mary-QAM. Pilot symbols are inserted into signal source, and are
6
modulated into OFDM signal by IFFT and inserting protective chain. Digital
Signals are converted to analog signals by using DAC before being
transmitted to Tx antenna. The signals through wireless channel are
affected by fading and additive Gaussian noise – AWGN. The populated
resource grid represents several subframes containing data. This grid is
then OFDM modulated and passed through the model of the propagation
channel. Channel noise in the form of additive white Gaussian noise
(AWGN) is added before the signal enters the receiver. Once inside the
receiver the signal is OFDM demodulated and a received resource grid can
be constructed. The received resource grid contains the transmitted
resource elements which have been affected by the complex channel gains
and the channel noise. Using the known pilot symbols to estimate the
channel, it is possible to equalize the effects of the channel and reduce the
noise on the received resource grid.

The channel estimation algorithm extracts the reference signals for a

transmit/receive antenna pair from the received grid. The least squares
estimates of the channel frequency response at the pilot symbols are
calculated as described in On Channel Estimation in OFDM Systems . The
least squares estimates are then averaged to reduce any unwanted noise
from the pilot symbols. Because it is possible that no pilots are located near
the subframe edge, virtual pilot symbols are created to aid the
interpolation process near the edge of the subframe. Using the averaged
pilot symbol estimates and the calculated virtual pilot symbols,
interpolation is then carried out to estimate the entire subframe. This
process is demonstrated in the following block diagram.

Pilot symbols are known in advance on the transmitter and receiver ,

and are transmitted together with a useful signal source for various
purposes such as channel estimation and channel synchronization.

The receiver performs the reverse functions as done with the

transmitter. However, in order to estimate the transmitted signal, the
channel transmission function of the wireless channel is performed via the
received pilot symbols at the receiver side. The signal received after OFDM
demodulation is divided into two signal flows. The first signal flow is the
useful signal fed to the channel equalizer. The second flow of signals is the

7
pilot symbol that is fed into the channel estimation. The estimated channel
is also fed into the channel equalizer to estimate the signal

In the next part, the principle of performing communication channel

recovery via the pilot symbols will be presented

1.2 Priciple of pilot symbols insertion in time domain and frequency

domain
The pilot symbols in OFDM system are assigned positions within a
subframe depending on the SNRindB cell identification number and which
transmit antenna is being used, as shown in the following Figure 2.

Time domain

Df
Frequency domain

Pilot Symbol

Dt

Figure 2: Pilot symbol in time domain

The unique positioning of the pilots ensures that they do not interfere
with one another and can be used to provide a reliable estimate of the
complex gains imparted onto each resource element within the transmitted
grid by the propagation channel.

8
 P P P’

P P P’

…..
P P P’
Frequency

P P P’

Time

Figure 3: Pilot average subsystem

All the pilot symbols found in a subcarrier are time averaged across all
OFDM symbols, resulting in a column vector containing the average for
each reference signal subcarrier, P’

cir

Doppler
t

fd

-fd

Figure 4: Relationship between Doppler effect and channel

9
 Pilot symbols can be inserted with useful symbols that are useful both
in frequency domain and time domain as shown in the figure 1 and figure 4.
However, the distance between two consecutive pilot symbols must comply
with the sampling rules in both frequency domain and time domain. In the
frequency domain, the variation of the wireless channel depends on the
maximum transmission delay of the channel tmax ( maximum propagation
delay). With rf is oversampling rate in frequency domain, fs is consecutive
distance between two sidelobes, the distance between two pilot symbols Df
must satisfy the following condition:
1
r f= ≥ 1 (Eq 1)
Df f s t max

The minimum sampling ratio in the frequency domain, rf , should be 1. This

ratio may be greater than 1, when there is more pilot symbols required and
the channel is oversampling. In case that the distance between the two pilot
symbols does not satisfy the sampling conditions as in the equation (Eq 1),
that is, the channel cannot be completely estimated through the pilot
symbol.

As in the frequency domain, the time-domain distance of two consecutive

pilot symbols should also satisfy the time-domain sampling standards. The
time-domain variation of the wireless propagation function depends on the
Doppler frequency fD,max . According to the sampling standards in the
frequency domain, the distance Dt must satisfy the following conditions :
1
rt = ≥ 1 (Eq 2)
2 f D ,max Dt (T S +T G )

The ratio rt is called the time domain sampling rate. In the case the
condition at the (Eq 2) is not satisfied, the wireless channel transmission
function cannot be fully estimated on the Receiver .

1.3 Channel estimation through normal methods

Assuming that on OFDM symbol i’ and on sub carrier n' the pilot
symbol Si’,n’ is transmitted. After demodulation on the Receiver , this
symbol is represented in the presence of additive Gaussian noise – AWGN
as follows

10
 Since the pilot symbol is known on the Receiver, the channel is easily
estimated through the following two steps

Step 1: The channel factor at the pilot symbol Ȟ i ' ,n ' is estimated by
dividing the received pilot symbol by the transmitted one. The result is as
following:
Ri ', n'
Ȟ i ' ,n ' = (Eq 3)
S i ', n '

However, only the wireless transmitting function at the location of the

pilot symbol is estimated, while the transmitting function at the location of
the useful symol remains unknown. This problem is solved through
interpolation algorithms

Step 2: The transmission channel coefficients at positions of useful

symbols through interpolation algorithms from the estimated channel
coefficients as obtained in step 1

Ȟ i ' ,n ' =¿ t erpolation of { Ȟ i ' ,n ' } (Eq 4)

There are many interpolation methods that can be used such as linear
interpolation, interpolation using the SI function, cubic polynomial
function, or interpolation using the Wiener filter. The following parts
introduce the properties as well as how to implement these interpolation
algorithms

1.4 Interpolation techniques for channel estimation

a. Interpolation using linear function, SI function and polynomial

function
Some of the most common interpolation techniques are linear function
interpolation, interpolation using function Si(x) = sin(x)/x, or cubic
polynomial interpolation.

In linear interpolation, the transfer function at the position of the

useful symbol is interpolated only through two adjacent points of the two
pilot symbols. However, in polynomial interpolation, the transfer function
of the useful symbol is interpolated through many different points of the
pilot symbol. Therefore polynomial interpolation has better quality than
linear interpolation but the complexity is higher.

11
 Figure 4 illustates a linear interpolation method where each point to
be interpolated is the mean of the two adjacent points.

Figure 5: Interpolatic techniques

Figure 6 illustrates two methods of polynomial interpolation and SI
interpolation. Both of these methods are commonly used in practice. SI
interpolation is essentially the use of a low pass filter in the frequency
domain. In the time domain it is represented by the SI function.

12
 Figure 6: SI interpolation and cubic interpolation
b. Interpolation using Wiener filter

….. …..

1 2 Ntap

-
∑
+

Figure 7: Wiener filter

13
 The Wiener filter is also known as MMSE - minimum mean sqare error.
Wiener filters are widely used in signal balancing or channel estimation
techniques. The filter structure is depicted in Figure 6. The filter input
filter Ȟ i ' ,n ' channel coefficient values at the routing messages obtained at
(Eq 3). The values of the transmission channel are multiplied by the filter
coefficients wi’,n’,i,n as shown in the equation below:
❑
Ȟ i ,n = ∑ wi ', n ', i ,n Ȟ i ' , n ' (Eq 5)
∀ {i ' ,n ' }∋ P

At (Eq 5), the set P is the set of all the values for i 'and n'. That is, a Ȟ i ,n
value is interpolated from different Ȟ i ' ,n ' elements in both the frequency
domain and the time domain. Then, people call the interpolation is two
dimentional Wiener interpolation (2D Wiener interpolation). This
interpolation is highly accurate but highly complicated. Peter Hoehet
demonstrated that a two-way Wiener filter can split into two one-way
Wiener filters (one operates in the time domain and one performs in the
frequency domain). Thanks to that, the complexity of implementing the
filter is greatly reduced, but the filter signal quality is not significantly
reduced

If we represent the input values Ȟ i ' ,n ' as a column vector as follows:

Ȟ 1,1 (1)
.
.
.
ȟ=⌊ Ȟ i' , n' ( k) (Eq 6)
.
.
.
Ȟ (l¿¿t −1)D +1 ,(l¿¿f −1)D +1(N
t f ) ⌋¿
tap

and the filter coefficients wi’,n’,i,n in the form of the vector dong

WTi,n = (w1,1,i,n,…wi’,n’,i,n,…, w(l¿¿ t−1) D +1 ,(l¿¿ f −1) D +1 ,i , n ¿¿ (Eq 7)

t f

then (Eq 5) a is represented as follows:

Ȟ i ,n =¿ WTi,nȟ (Eq 8)

14
 In Figure 6, Ntap is the coefficients of the filter, which are equivalent to
the number of Ȟ i ' ,n ' input signals used to interpolate for an output value. If
the filter coefficients are designed optimally, the squared error between all
the interpolation results Ȟ i ' ,n ' and the ideal Hi,n value is minimal. The
calculation of the optimal coefficients for the filter is based on the equation
of Wiener-Hop as shown below

Equation of Wiener-Hop

Equations of Wiener-Hop used to calculate the filter coefficients. The

purpose of the filter is to minimize the normal error between the ideal
factor of the channel and the coefficient estimated using the filter.
Estimated using the ^ H n ,i filter

WTi,n = PTi,n R (Eq 9)

(Eq 9) is called pt winer-hop which allows calculating the filter factor

vector W such that the mean squared error is minimal. The condition for
calculating the filter coefficients is that the matrix R and the positive
channel vector must be known in advance

PTi,n = E[= Hi,n ȟ (Eq 10)

r ( 0)… r ( N tap−1)
r (−1)… … … … …. ⌋
R =⌊ ………….
(Eq 11)
r ( N tap−1) … r (0)

1.5 Channel equalization for OFDM systems

In this part it is assumed that the transmission channel remains
unchanged (or almost unchanged) over a period of the OFDM signal pattern
and within a frequency range the widths of two consecutive subcarriers.
This means in the time domain
H ( jw , t)=H ( jw , kT )with kT ≤ t ≤(k +1)T (Eq 12)

and frequency domain

H ( jw , t)=H (nj ws , t) with(n−1/2)w s ≤ w ≤(n+ 1/2) w s (Eq 13)

Then the transfer function coefficient corresponding to the nth

subcarrier and the kth OFDM symbol is represented as

15
 kT ≤ t ≤( k +1) T
H ( jw , t)=H (nj ws , t)with
{(n−1/2) ws ≤ w ≤(n+1/2) ws }
(Eq 14)

The transmitted signal is recovered through division of the signal after

demodulation with a transmission function factor as follows
1 ~
d k , n= d (Eq 15)
H (nj ws , kT ) k ,n

The above equation shows that the equalizer for the OFDM system is
performed in a very simple way when the radio channel transmission
function has been restored. The channel equalizer is performed simply by
dividing the received signal by the transmission function factor of the
channel

16
CHAPTER 2: SIMULATION CHANNEL ESTIMATION FOR OFDM SYSTEM
ON MATLAB SOFTWAVE

2.1 Code in Matlab

a. OFDM.m

17
18
b. Pilotsymbol.m

19
c. OFDM_Modulator.m

20
21
2.2 Simulation results

Figure 8: BER

22
23
 CONCLUSION
In this project, our group with 2 members practice Matlab software to
design and simulate channel estimation for OFDM systems from basic to
complex. We learn more detail about channel estimation for OFDM systems
applications which is very popular in telecommunications.. Base on all the
fundamental knowleges, we can find the following major in the future. The
durability of product will be improved so that it can work in industrial
environment.

In the process of completing the project, the group received the help of
Assoc. Prof. Nguyễn Vă n Đứ c and the classmates. Through the
implementation of the project, we have accumulated a lot of practical
knowledge in the field of Digital communications and will certainly learn
more.

Lastly, once again we would like to thank Assoc. Prof. Nguyễn Vă n Đứ c

who has helped us get closer to Electronics industry.

We sincerely thank you!

24
 REFERENCES
1. Lý thuyết và cá c ứ ng dụ ng củ a kỹ thuậ t điều chế OFDM, Nguyễn Vă n
Đứ c, NXB Khoa họ c và kỹ thuậ t
2. Thô ng tin vô tuyến, Nguyễn Vă n Đứ c, Vũ Vă n Yêm, Đà o Ngọ c Chiến,
Nguyễn Quố c Khương, Nguyễn Trung Kiên, NXB Khoa họ c và kỹ
thuậ t
3. Cá c bà i tậ p Matlab về thô ng tin vô tuyến, Cheng-Xiang Wang,
Nguyễn Vă n Đứ c, NXB Khoa họ c và kỹ thuậ t
4. Channel Models of the Doppler Effect , Nikolaos Tsakalozos,
Konstantinos Drakakis and Scott Rickard
5. https://www.mathworks.com/matlabcentral/fileexchange/41634-
channel-estimation-for-ofdm-systems

25