Study on High Efficiency OFDM Transmission System with Link
Adaptation Technique
Quan Ji
Master Course
Division of Electrical and Electronic Engineering Graduate School of Engineering
Mie University
March 2015
Table of Contents
1 Introduction
1.1 OFDM Transmission Technique………..1
1.1.1 OFDM Overview………..1
1.1.2 Multi-frequency Modulation and Demodulation Method………3
1.1.3 Fundamental of OFDM Technique………...5
1.2 Multi-path Fading Environments………...6
1.3 Role of Guard Interval………..……….8
1.4 CSMA/CA Mechanism………...9
1.5 Proposal Background………...11
1.6 Thesis Structure………...12
2 GAM-OFDM System………13
2.1 Channel Model and Channel Estimation………13
2.1.1 AWGN Channel………..…13
2.1.2 Multi-path Rayleigh Fading Channel………15
2.1.3 Time-varying Channel……...………. .…17
2.2 Adaptive Modulation Method……… …19
2.3 Grouping Method………..22
2.4 Frame Structure………...25
2.5 Convolutional FEC Codec………...…26
ii
3 Proposal of AOFDM-BMD System………..……..…28
3.1 System Model………..……..…28
3.2 System Structure………..………30
3.3 BMD Method………..………...…31
3.4 AFS Method……….……..35
4 Performance Evaluation……….37
4.1 Simulation Environment………....37
4.2 Transmission Quality………..39
4.3 System Throughput……….43
5 Conclusions………...….45
6 Bibliography………..47
iii
Acknowledgments
I would like to express the deepest appreciation to my supervisor Prof. Hideo Kobayashi, at the Communications Laboratory of Electrical and Electronics Department, Graduate School of Engineering, Mie University, for his full support, patience and expert guidance throughout my study, research and even my private life. In addition, I express my appreciation to Prof. Kazuo Mori and Associate Prof. Katsuhiro Naito for their precious advices and comments during my research.
I also thank Dr. Hiroharu Kawanaka for his reviewing opinions on this thesis and Mr. Yoshihiro Yamamoto for his help on technical troubles I ran into during my study. I was fortunately to have been studied with all other students at the communications laboratory. I give them my thanks for their support and friendship.
iv
Abstract
Today wireless communication technology is used in many applications which are well integrated into our daily life. From this fact, the demands for high quality and high data rate multimedia communication services keep increasing drastically. To realize these demands, we have to solve the significant degradation problem of Bit Error Rate (BER) performance in the multi-path fading which prevents the development of advanced wireless communication. The Orthogonal Frequency Division Multiplexing (OFDM) communication technique, which uses orthogonalized narrow band frequency domain subcarriers, has received considerable attentions, from the facts that OFDM has more efficient usage of frequency bandwidth as compared with the conventional single carrier transmission technique.
Since the frequency band of each subcarrier is narrowband, the time domain symbol duration becomes longer, OFDM can achieve better transmission quality under multi-path fading environments.
But channel conditions of received OFDM signal consisting of orthogonal subcarriers would fluctuate over a wide range under the multi-path fading channel. In conventional OFDM systems, to ensure the signal quality, systems set transmission parameters according to the worst channel condition, which leads to the fatal degradation of transmission efficiency. The link adaptation based on adaptive modulation OFDM system is well known as one of the solutions to overcome this problem [1]. Link adaptation is to improve the channel capacity by changing certain transmission parameters adaptively according to the channel conditions. The adaptive OFDM systems, which assign optimum modulation scheme to each subcarrier according to its channel condition, can improve transmission efficiency significantly. However, the transmitter is required to send Adaptive Modulation Information (AMI) of each subcarrier to the receiver for correct demodulation.
The number of AMI bits would increase proportionally to the number of subcarriers and modulation schemes. To solve this problem, Grouping Adaptive Modulation OFDM (GAM-OFDM) system was proposed. The GAM-OFDM system divides one OFDM symbol to several subcarrier groups and assigns constant modulation scheme to the subcarriers within one group to reduce the number of AMI bits that required to transmit. Results in [2] show that the GAM-OFDM decreases the AMI significantly under static environment. However, in mobile communication systems, the transmission of AMI bits becomes more sophisticated and frequent. The transmission of AMI bits would still lead to a high complexity of transceiver and decrease the transmission efficiency. In addition, due to the Doppler frequency shift, channel conditions change symbol by symbol in one frame while the channel condition of subcarriers at the same place of each symbol are constant under static environment.
v
Therefore, if the frame length is set too long, the transmission signal quality will decrease. On the other hand, if the frame length is too short, it takes more frames to transmit the same amount of data, which leads the longer time delay and possibility of collision grows since the media access control method, which will be introduced below, is applied in wireless communication systems.
To solve these problems, this thesis proposes a novel blind modulation detection method for OFDM systems with grouping adaptive modulation in time-varying channel. The salient feature of proposed method is to enable the detection of modulation scheme autonomously at the receiver by using the constellation of the received signal without any side information such as AMI. This thesis also proposes an adaptive transmission frame size method into grouping adaptive modulation OFDM (GAM-OFDM) systems which can achieve higher transmission efficiency in time varying fading channel by transmit more symbols in one frame. This thesis presents various evaluation results under static and mobile environments to verify the effectiveness of proposed systems. Simulation results show that the proposed system with adaptive modulation and transmission frame size methods can achieve higher transmission efficiency with keeping the required signal quality even under lower CNR conditions.
Chapter 1
Introduction
Recently, mobile communication technology is used in a lot of scenes, and it is considered that much more services based on mobile communication technology are coming to the stage from now on. Hence, the effective using of frequency resources and high speed transmission becomes to be a challenging. The OFDM technique has received a lot of attentions as one of the promising wireless transmission techniques which can meet these needs. In this chapter, the basic acknowledge of OFDM technique is introduced.
1.1 OFDM Transmission Technique
1.1.1 OFDM Overview
OFDM technique is a method of encoding digital data to modulated symbols by using Phase Shift Keying (PSK) or Quadrature Amplitude Modulation (QAM) on multiple carrier frequencies. OFDM technique has a long history, the basic concept of OFDM technique was proposed in the mid 1960’s.
Although the Discrete Fourier Transform (DFT) and Inverse Discrete Fourier Transform (IDFT) algorithms are employed to the modulation and demodulation part which simplified the receiver constitution, the computational complexity was still a problem. But the rapid development of digital signal processing technique has solved this problem, and OFDM technique has finally come to the stage.
OFDM technique is well known for its efficient usage of bandwidth and robustness to the multi- path fading [3]. From these advantages, OFDM technique has already been adopted as the downlink technique of the 4th Generation of mobile communications technology (LTE: Long Term Evolution).
The basic idea of OFDM is to divide the available bandwidth to several narrowband sub channels which are overlapped and orthogonal to each other. Fig.1.1 compares the spectrum overlap of Frequency Division Multiplexing (FDM) with OFDM. For FDM, the guard bands between subcarriers, which are the main cause of a low bandwidth efficiency, are used to avoid inter subcarrier interference
1 (ISI) and inter channel interference (ICI). On the other hand, the subcarriers in OFDM are overlapped with each other orthogonally. The orthogonality of subcarriers’ spectra and the narrow enough sub channels in OFDM can eliminate the effects of ISI and ICI without using guard band between subcarriers.
Guard Band
f
f (a) FDM
(b) OFDM
Fig.1.1 Spectrum Overlapping of FDM and OFDM
2 1.1.2 Multi-frequency Transmitter and Receiver
Fig.1.2 and Fig.1.3 show the transmitter and receiver of multi-frequency modulation system. At the transmitter side, an N size original data is modulated by an AM modulator which has the same central frequency interval ∆𝑓, individually. The modulated signal are added together by the transmitter and sent out to the receiver as eq.1.1 shows.
𝑆(𝑡) = ∑𝑁−1𝑛=0𝑎𝑛𝑒𝑗2𝜋𝑓𝑛𝑡 (1.1)
At the receiver side, the received signal, which shown by eq.1.1, is received and goes through a correlation detector. For k-th data, the receiver uses the same carrier frequency 𝑓𝑘, which used at the transmitter side, to do synchronous detection. After detection, the receiver integrates the detected signal over the transmitted data time. Fig.1.4 shows the constitution of synchronous detector, which can also be called matched filter.
t f j2 0
e
t f j2 1
e
t f j2 N-1
e
a0
a1
1
aN
1
0
e 2
)
( N
n
t f j
n n
a t
b
入力データ
AM変調器
送信信号
Fig.1.2 Transmitter of Multi-frequency Modulation AM Modulator
Input Data
Transmitted Signal
3 t
f j2 0
e-
t f
j2 1
e-
t f j2 N-1
e-
a0
a1
1
aN
1
0
e 2
)
( N
n
t f j
n n
a t
r
復調データ
相関検波器 受信信号
0Ts dt
0Ts dt
0Ts dt
Fig.1.3 Receiver of Multi-frequency Modulation
Fig.1.4 Constitution of Synchronous Detector Correlation Detector
Demodulated Data
Received Signal
4 1.1.3 Model of Basic OFDM System
Fig.1.5 shows the block diagram of a basic OFDM system with N subcarriers. In the figure, 𝑎𝑛 is the modulated signal of original data 𝑥𝑛, complex signal 𝑏𝑖 which is converted from an IFFT transform as Eq.1.2, is added by a GI and fed into P/S converter. The physical channel is experiencing the frequency selective fading whose impulse response is represented by h, and the additive white Gaussian noise n. On the receiver side, it removes the GI after S/P conversion, performs FFT transform and demodulate signal 𝑎𝑛∗ to get received data 𝑦𝑛∗.
𝑏𝑖 = ∑𝑁−1𝑛=0𝑎𝑛𝑒𝑗2𝜋𝑛𝑖/𝑁 (1.2)
Depending on the analyzed situation, imperfections in a real OFDM system may be ignored or explicitly included in the above model. Possible imperfections may include:
Time and frequency dispersion
Nonlinearities and clipping distortion
External interference
Corresponding to these imperfections, additional models and additional compensative techniques should be considered to improve the system performance [4][5][6][7][8].
𝑎1
𝑥1
n
𝑦𝑁∗ 𝑦1∗
𝑎𝑁∗ 𝑎1∗
𝑏𝑖∗ 𝑟𝑖
𝑏𝑖 𝑎𝑁
𝑥𝑁
MOD
MOD
IFFT +GI P/S
h
S/P -GI FFT
DEMOD
DEMOD Fig.1.5 Block Diagram of Basic OFDM System
5
1.2 Multi-path Fading Environments
In mobile communication, the signal offered to the receiver contains not only a direct line-of-sight radio wave, but also a large number of reflected radio waves. As showing in Fig.1.6, the signal at the receiver consists of a direct wave and a reflected wave which reflected by one building. These reflected waves interfere with the direct wave, which causes significant degradation of the system performance, and with the more reflected wave, the signal undulates more intensively. This kind of signal fading is called multi-path fading [9].
Fig.1.6 Multi-path Fading Environment
In fig.1.6, ρ, θ and τ indicate the amplitude, phase and delay time of reflected wave. The received time domain signal can be expressed as eq.1.3 if the frequency of carrier is assumed to 𝑓𝑐.
𝑟(𝑡) = {𝑠(𝑡) + ρ𝑒𝑗𝜃𝑠(𝑡 − 𝜏)}𝑒𝑗2𝜋𝑓𝑐𝑡 (1.3)
Eq.1.3 can be transferred to frequency domain by FFT as the equation below.
𝑅(𝑓) = 𝑆(𝑓 − 𝑓𝑐) ∙ {1 + 𝜌𝑒−𝑗2𝜋(𝑓−𝑓𝑐)𝜏∙ 𝑒𝑗𝜃}
= 𝑆(𝑓 − 𝑓𝑐) ∙ 𝐻(𝑓 − 𝑓𝑐) (1.4)
H(f) is the Channel Frequency Response (CFR) under multipath fading environment, which can be expressed as,
𝐻(𝑓) = 1 + 𝜌𝑒−𝑗2𝜋𝑓𝜏 ∙ 𝑒𝑗𝜃 (1.5)
from fig1.6, it can be seen that the CFR under multi-path fading environment can be expressed by using the phase, amplitude and delay time of delay path. In OFDM technique, information data are transmitted in frequency domain, therefore the frequency domain data after FFT at receiver side can
𝒔(𝒕)
𝛒𝒆𝒋𝜽𝒔(𝒕 − 𝝉) Direct Wave
Reflected Wave
6 be achieved by multiplying the transmitted data and CFR. Fig.1.7 shows an example of CFR of one OFDM symbol which is receiving the influence of multi-path fading. It can be observed from the figure, all sub-carriers are experiencing fluctuation over a wide range due to the multi-path fading.
Some subcarriers have good conditions while others experience deep fading that would dominate the whole transmission quality. It is required to estimate the CFR to realize the high accuracy demodulation of data, which would be introduced in chapter 2.
Fig.1.7 Frequency Response of Multi-path Fading
0 50 100 150 200 250
0 0.5 1 1.5
Subcarrier Number
Amplitude of Frequency Response
7
1.3 Role of Guard Interval (GI)
One of the most outstanding features of OFDM is that it performs excellently even under multi-path fading channels. Here, the GI plays an important role to ensure that transmissions do not interfere with one another. As Fig.1.8.A shows, system copies the last part of one OFDM symbol, which is called GI, to the front of the symbol. Generally, in OFDM systems the length of GI is decided in accordance with the maximum delay time of delay paths. GI makes the signal avoid ISI and ICI due to the multi- path fading without disturbing the periodicity of the signal. Although decreasing the transmission efficiency, adding GI to the front of symbol achieves much better transmission quality under multi- path fading environment.
Fig.1.8 Function of Guard Interval
At the receiver side, system can extract the data symbol part at any time point in GI duration with maintaining the orthogonality of frequency domain subcarriers. As an example, fig.1.8.B illustrates that system extracts the symbol length time τ earlier than the first data, the time domain signal can be expressed as the equation below.
𝑏𝑘τ = ∑𝑁−1𝑛=0𝑎𝑛𝑒𝑗2𝜋𝑛(𝑘−𝜏)𝑁 (1.6)
If we change the time domain signal above to frequency domain by FFT as eq.1.7.
𝑎̂𝑖 = ∑𝑁−1𝑘=0𝑏𝑘𝜏𝑒−𝑗2𝜋𝑖𝑘𝑁 = 𝑎𝑖𝑒−𝑗2𝜋𝑖𝜏𝑁 (1.7)
each subcarrier has a phase rotation due to the time τ in according to the eq.1.7. However, there is no interference from other subcarrier after demodulation, which means the orthogonality between subcarriers is maintained.
8
1.4 CSMA/CA Mechanism
In early area networking using Ethernet technology, a media access control method called Carrier Sense Multiple Access with Collision Detection (CSMA/CD) was used to ensure the transmission efficiency. CSMA/CD terminates transmission as soon as a collision is detected to shorten the time needed before a retry can be attempted. While in wireless communication systems, it is impossible to detect collision, instead of that, Carrier Sense Multiple Access with Collision Avoidance (CSMA/CA) is used to avoid collision as much as possible. Nodes attempt to avoid collisions by transmitting only when the channel is sensed to be "idle" [10][11]. If the medium is identified as being clear or received a Clear to Send (CTS) to explicitly indicate it can transmit data, it sends the frame. When they do transmit, nodes transmit their packet data in its entirety. It is particularly important for wireless networks, where the collision detection of the alternative CSMA/CD is unreliable due to the hidden node problem.
Fig.1.9 illustrates the basic transport process between two terminals in IEEE802.11 wireless network with CSMA/CA mechanism. Here, it is assumed that unicast communications between two terminals and no retransmission due to collision between packets. There is an exchange of a Request to Send (RTS) packet sent by the transmitter before transmission of data, and a CTS packet sent by the receiver after receiving the data packet according to the illustration. Thus alerting the other node to not transmit for the duration of the data transmission. This is known as the IEEE 802.11 RTS/CTS exchange. Implementation of RTS/CTS helps to partially solve the hidden node problem that is often found in wireless networking.
It can be seen from Fig.1.9 that the transmission time period T can be expressed by,
T=𝑇𝑅𝑇𝑆+𝑇𝐶𝑇𝑆+𝑇𝐷𝐴𝑇𝐴+3×𝑇𝑆𝐼𝐹𝑆+𝑇𝐴𝐶𝐾+𝑇𝐷𝐶𝐹 (1.8)
where 𝑇𝑅𝑇𝑆, 𝑇𝐶𝑇𝑆 and 𝑇𝐷𝐴𝑇𝐴 denote the transmission time periods of RTS, CTS and Data, respectively. 𝑇𝑆𝐼𝐹𝑆 and 𝑇𝐷𝐶𝐹 are the Short Interframe Space (SIFS) and Distributed Coordination Function (DCF) periods. Here, 𝑇𝐷𝐶𝐹 can be calculated by,
min
DCF DIFS slot time
T T CW T 2 (1.9)
where 𝑇𝐷𝐼𝐹𝑆 and 𝑇𝑠𝑙𝑜𝑡𝑡𝑖𝑚𝑒 denote the DIFS period and slot time duration. 𝐶𝑊𝑚𝑖𝑛 indicates the minimum contention window.
9 Fig.1.9 Basic Transport Process in IEEE802.11
10
1.5 Proposal Background
As mentioned in the above sections, although OFDM technique can provide better Bit error rate (BER) performance under frequency selective fading as compared to the conventional single carrier transmission technique, the instantaneous BER performance for each OFDM subcarrier will be fluctuated over a wide range according to the fading condition of each subcarrier as show in Fig.1.8.
In the conventional OFDM system, the lowest order modulation scheme has to be employed for all subcarriers in order to keep the overall objective BER performance. This leads the inefficient usage of frequency bandwidth and encountered in realizing higher data rate transmission.
The OFDM system conjunction with adaptive modulation technique is well known as one of the methods to solve the above problem [12][13]. In the OFDM systems with adaptive modulation technique, the optimum modulation scheme is assigned for each subcarrier according to its instantaneous fading characteristics. In other words, the system employs higher order modulation schemes so as to carry more bits per symbol for the subcarriers in good channel conditions, and employs lower order modulation schemes to carry one or even zero bit per symbol for those subcarriers affected by deep fading to make sure the transmission quality. By introducing such manner in OFDM system, the overall performance of the system including the signal quality and channel capacity can be improved significantly even under severe frequency selective fading conditions.
But, in the above system, transmission side is required to inform the modulation scheme employed for each subcarrier to the receiver side for the correct demodulation. These adaptive modulation information (AMI) bits are usually transmitted through the dedicated control channel or the traffic channel as the side information. The number of required AMI bits would be directly proportional to the number of subcarriers consisting of OFDM signal and the number of available modulation schemes considered in the system. Furthermore, these information bits are required to communicate between the transmitter and the receiver with very high quality. Therefore, the increase of adaptive modulation information bits will lead to degradation of overall system performance or unnecessary increase of both frequency bandwidth and transmission power.
In addition, in mobile communication systems, the transmission of AMI bits becomes more sophisticated and frequent since the channel conditions change symbol by symbol in one frame due to the Doppler frequency shift which means the transmitter has to transmit AMI for each group of data every several symbols. If the interval of transmitting AMI is set too long, the transmission signal quality will decrease since the modulation schemes applied to each subcarrier may not be able to be used to the subcarriers in later symbols. On the other hand, if the interval of transmitting AMI is too short, it takes more frames to transmit the same amount of data, which leads the longer time delay and possibility of collision grows since the CDMA/CA mechanism is applied in wireless communication
11 systems. The high quality transmission of AMI indicates that system has to apply extra methods to ensure the transmission quality of AMI such as FEC encode or transmitting them several times.
Therefore, the transmission of AMI bits would still lead to a high complexity of transceiver and decrease the transmission efficiency.
To solve the above problem, this thesis proposes a Blind Modulation Detection (BMD) method for Grouping Adaptive Modulation OFDM (GAM-OFDM) system to increase the system capacity in the time-varying channel [14]. The salient feature of proposed BMD method is to enable the detection of modulation scheme for each group autonomously at the receiver without any AMI bit. This thesis also introduces an adaptive transmission frame size method in which the number of symbols in one frame changes simultaneously according to channel conditions to improve the transmission efficiency with keeping the required signal quality [15].
12
1.6 Thesis Structure
The following of this thesis is organized as,
Chapter 2 illustrates the general description of grouping adaptive modulation system including system model, grouping modulation method and frame structure. This chapter also introduces the channel models and channel frequency response estimation methods used in these channels.
Chapter 3 presents the proposed Adaptive OFDM system with Blind Modulation Detection method (AOFDM-BMD) in detail. This chapter consists of several sections. The first section describes the system model of proposed system. Section 3.2 describes the block diagram of proposed system.
Sections 3.3 and 3.4 present the proposed Blind Modulation Detection (BMD) and Adaptive Frame Size (AFS) methods in detail.
Chapter 4 shows various results obtained from computer simulations including Packet Error Rate (PER) and channel throughput performances which indicate the advantage of proposed system.
Finally, some conclusions are presented in chapter 5.
13
Chapter 2
Grouping Adaptive Modulation (GAM)-OFDM System
As mentioned in the above chapter, adaptive modulation OFDM technique changes modulation schemes according to the instantaneous channel condition so as to improve the transmission efficiency.
In conventional adaptive modulation OFDM systems, optimum modulation scheme is employed to each subcarrier according to its fading characteristic. Hence, the transmitter has to inform the AMI bits with high quality for every subcarrier to the receiver for correct demodulation. To reduce the required number of AMI bits which are required to transmit to the receiver side for correct demodulation, GAM-OFDM system was proposed. This chapter introduces GAM-OFDM system in detail. Before that, the mathematic model and channel estimation methods of multi-path Rayleigh fading and Time-varying channels are discussed also.
2.1 Channel Model and Channel Estimation
2.1.1 AWGN Channel
The basic type of digital communication channels is the Additive White Gaussian Noise (AWGN) channel. AWGN channel is a basic and generally accepted model for thermal noise in communication channels. In an AWGN channel model, it is assumed that the received signal equals to the transmit signal plus some noise, where the noise is statistically independent of the signal and the value of the noise follows the Gaussian probability distribution function. The white Gaussian noise is a model for the thermal noise generated by random electron movement in the receiver. AWGN channel is a good channel for many satellite and deep space communication links. It is not a good channel for most terrestrial links because of multipath, terrain blocking, interference, etc. This model can be thought of as a wireline channel model, since there is effectively perfect transmission from transmitter to receiver.
The equation for the received signal R(t) in an AWGN channel can be expressed as,
𝑅(𝑡) = 𝑆(𝑡) + 𝑁̃(𝑡) (2.1)
14 where S(t) is the transmitted signal, R(t) is the received signal, and 𝑁̃ is a zero mean wide sense stationary random process with power spectral density S(jw)=No/2. The performances of the AWGN channel can be used to compare systems using different methods of modulation. Here, it is assumed that Binary Phase Shift Keying (BPSK) is used to transmit data as an example, the theoretical bit error rate can be expressed as
𝑃𝑏 = 12𝑒𝑟𝑓𝑐(√𝐶𝑁𝑅) (2.2)
Where, 𝑒𝑟𝑓𝑐(𝑥) = ∫𝑥∞√𝜋2 𝑒−𝑢2𝑑𝑢 (2.3)
Following is a figure which shows the theoretical bit error rate when changing CNR for BPSK signals going through an AWGN channel. It is very clear that the bit error rate drops as the CNR increases.
Fig.2.1 BER of BPSK under AWGN Channel
0 1 2 3 4 5 6 7 8 9 10
10-6 10-5 10-4 10-3 10-2 10-1
CNR dB
BER
15 2.1.2 Multi-path Rayleigh Fading Channel
Multi-path Rayleigh fading is a reasonable model when there are many objects in the environment that scatter the radio signal before it arrives the receiver. Multi-path Rayleigh fading models assume that the magnitude of a signal that has passed through such a transmission medium will very randomly, or fade, according to the multi-path reception as shown in Fig.1.6. In a multi-path environment, it is reasonably intuitive to visualize that an impulse transmitted from transmitter will reach the receiver as a series of impulses. Multi-path Rayleigh fading is most applicable when there is no dominant propagation between the transmitter and the receiver. If there is a dominant line of sight, Rician fading may be more applicable. Let the transmit bandpass signal be,
𝑠(𝑡) = 𝑅𝑒[𝑠𝑏(𝑡)𝑒𝑗2𝜋𝑓𝑐𝑡] (2.4)
where, Re[] means the real part of complex number, 𝑠𝑏(𝑡) is the baseband signal, 𝑓𝑐 is the carrier frequency.
The transmitted signal reaches the receiver through multi-path channels where the n-th delay signal has an attenuation 𝛼𝑛(𝑡) and delay 𝜏𝑛(𝑡). So the received signal is
𝑟(𝑡) = 𝑅𝑒{∑𝑛𝛼𝑛(𝑡)𝑠𝑏[𝑡 −𝜏𝑛(𝑡)]𝑒𝑗(2𝜋𝑓𝑐(𝑡−𝜏𝑛(𝑡)))} (2.5)
The baseband equivalent of received signal is,
𝑟𝑏(𝑡) = ∑ 𝛼𝑛 𝑛(𝑡)𝑒−𝑗2𝜋𝑓𝑐𝜏𝑛(𝑡)𝑠𝑏[𝑡 − 𝜏𝑛(𝑡)]
= ∑ 𝛼𝑛 𝑛(𝑡)𝑒−𝑗∅𝑛(𝑡)𝑠𝑏[𝑡 − 𝜏𝑛(𝑡)] (2.6)
where, ∅𝑛(𝑡) = 2π𝑓𝑐𝜏𝑛(𝑡) is the phase of the n-th path.
The impulse response can be expressed by
ℎ𝑏(𝑡) = ∑ 𝛼𝑛 𝑛(𝑡)𝑒−𝑗∅𝑛(𝑡) (2.7) When under static environment, ∅𝑛(t) is constant for each delay signal, the channel frequency response of each symbol stays the same during one frame, as shown in Fig.2.2, so the channel frequency response estimated by preamble, which inserted to the front of one frame, can be applied to all the symbols in one frame. In static systems like IEEE802.11a, one preamble symbol inserted before the transmission of data symbols is often used to estimate the channel frequency response, just as fig.2.3 shows [16].
16 Fig.2.2 Channel Frequency Response in Static System
Fig.2.3 Frame Format in Frequency Domain for Multi-path Fading Channel
0 100 200 300 400 500 600
0 20 40 60 80
0 0.5 1 1.5 2 2.5
Frequency Domain Time Domain
Amplitude
17 2.1.3 Time-varying Channel
In mobile system, phase ∅𝑛(𝑡) varies due to the movement of terminals, which results in amplitude and phase fluctuation of channel frequency response of OFDM signal. The time varying phase can be modeled by:
∅𝑛(𝑡) =2𝜋𝑣𝑡 cos 𝜃𝜆 𝑛+ ∅𝑛0 (2.8)
𝑓𝑑 =𝑣𝜆 (2.9)
where, λ is wave length, 𝜃𝑛 is reflect angle, v is the relative speed between terminals, 𝑓𝑑 is the Doppler frequency. This phenomenon is called Doppler Frequency Shift. In high time-varying channel, the channel frequency response of each symbol varies intensely and keeps changing due to the Doppler frequency shift, which means the channel frequency response at last symbol may be much different as compared with the first symbol in one frame, as shown in fig.2.4. Hence, it is required to estimate channel frequency response of each symbol. In the past years, various techniques for channel estimation in both time and frequency domains have been explored. One technique for channel estimation is to use pilots. To get the channel information, pilots on predetermined subcarriers are sent to the receiver as training signals in OFDM systems, and the channels for pilot subcarriers are directly estimated since the receiver knows the original data of pilot subcarriers. The channels for non-pilot subcarriers can be estimated through interpolation with the channels of pilot subcarriers.
Fig.2.5 illustrates the frame structure in the frequency domain for time-varying channel. Scattered pilot tones are inserted to the frame every certain number of subcarriers. At the receiver side, channel responses at all subcarriers are estimated by applying the maximum likelihood (ML) or Minimum Mean Squared Error (MMSE) method to the received scattered pilot tones and interpolating values between channels for pilot subcarriers.
18 Fig.2.4 Channel Frequency Response in Mobile System
Fig.2.5 Frame Format in Frequency Domain for Time-varying Channel
0 100 200 300 400 500 600
0 20 40 60 80
0 0.5 1 1.5 2 2.5 3
Frequency Domain Time Domain
Amplitude
19
2.2 Adaptive Modulation Method
In GAM-OFDM systems, the transmitter continually monitors the channel frequency response and assigns optimum modulation scheme for each group according to the channel conditions. All the subcarriers included in one group employs the same modulation scheme. Here it can be assumed that the channel is constant during the forward and reverse transmissions. Accordingly, the channel response of multi-path fading during the period can be treated as static. Therefore, the received signal to noise power ratio can be estimated from the signal to noise power ratio at the transmitter side. Here, the CNR of n-th subcarrier can be expressed as,
𝑐𝑛𝑟𝑛 = 𝜆 + 10𝑙𝑜𝑔10{|𝐻𝑛|2} (2.10)
Where λ is the signal to noise power ratio at the transmitter and Hn is the channel frequency response at the n-th subcarrier.
In the following investigations of grouping adaptive method, 4 modulation schemes of PSK and QAM are employed, whose theoretical BER performance under AWGN channel are shown in fig.2.6.
The following equations present the theoretical BER of PSK and QAM modulation schemes under AWGN channel.
𝑃𝑏/𝑏𝑝𝑠𝑘 =12𝑒𝑟𝑓𝑐(√𝐶𝑁𝑅) (2.11)
𝑃𝑏/𝑞𝑝𝑠𝑘 = 12𝑒𝑟𝑓𝑐(√𝐶𝑁𝑅2 ) (2.12) 𝑃𝑏/𝑀−𝑄𝐴𝑀 =𝑘4(1 −√𝑀1 )12𝑒𝑟𝑓𝑐(√2𝑥) (2.13)
where 𝑘 = 𝑙𝑜𝑔2𝑀 and 𝑥 = √3𝑘∗𝐶𝑁𝑅𝑀−1 . Modulation scheme for each subcarrier group can be assigned according to average CNR or average theoretical BER of this group.
For the average CNR method, the averaged CNR over all subcarriers including one group can be calculated by,
𝐶𝑁𝑅(𝑘) = 𝜆 +𝑃1∑𝑈𝑛=𝐿𝑘 10𝑙𝑜𝑔10{|𝐻𝑛|2}
𝑘 (2.14)
where λ is the signal to noise power ratio at the transmitter, 𝐿𝑘 and 𝑈𝑘 are the lower and upper subcarrier number in the group k, P is the number of subcarriers in one group and Hn is the channel
20 frequency response at the n-th subcarrier. The transmitter estimates the received averaged CNR for each group by using equ.2.14. The transmitter then calculates the theoretical BER of 4 modulation schemes by bringing the above average CNR into theoretical BER functions listed above and assigns the highest order modulation scheme whose theoretical BER performance meets the required BER performance.
Fig.2.6 Theoretical BER performance of 4 Modulation Schemes under AWGN Channel While modulating according to the average BER of one group, the transmitter estimates the received CNRs for each subcarrier in this group by using equ.2.10 and calculates the theoretical BER of each subcarrier within this group by using theoretical BER functions. After that, the transmitter averages the calculated theoretical BERs in one group and assigns the optimum modulation scheme to the group so that the average BER of this group can meet the required BER.
According to our simulation results, the two adaptive modulate method showed almost the same subcarrier payload. In the remainder of this thesis, the average BER adaptive modulate method will be adopted. In addition, the convolutional code with 1/2 rate is employed in the following systems.
Fig.2.7 which shows the theoretical BER performances of AWGN channel for coded BPSK, QPSK, 16QAM and 64QAM modulation schemes will be used instead of the theoretical BER functions stated above. For example, in the following simulations, the packet size is set to 32 bytes while the required PER is 10−2, so the required BER is 3.9e-5. The transmitter assigns the optimum modulation scheme which has the highest order from the modulation schemes that meet the required BER according to fig.2.7.
0 5 10 15 20 25
10-6 10-5 10-4 10-3 10-2 10-1 100
CNR dB
BER
64QAM 16QAM QPSK BPSK
21 Fig.2.7 BER Curve of 4 Modulation Schemes in AWGN Channel.
0 5 10 15 20
10-6 10-5 10-4 10-3 10-2 10-1 100
C/N, dB
BER
BPSK 16QAM QPSK 64QAM
22
2.3 Grouping Method
In GAM-OFDM systems, the GAM method divides OFDM channel into several subcarrier groups, and modulates each group of subcarriers adaptively according to the channel condition of subcarriers within this group. There are four modulation schemes used in the following research as listed in tab.2.1.
For every group of subcarriers, there are 5 modulation states including NULL, which means the AMI for one subcarrier group takes 3 bits. Fig.2.8 shows an example of grouping method, from which it can be observed that as the number of group grows, the CFR obtained by using GAM method is getting closer to the original CFR [17]. Fig.2.9 shows the loss of channel capacity when using different number groups in fig.2.8. Under low CNR conditions, almost all subcarriers in the frame are disrupted seriously so that system cannot transmit any bit. Therefore, nearly all of the channel capacity under very low CNR conditions lose no matter how many subcarrier groups are set. As the CNR increases, the effort of grouping method becomes apparent. Generally, with more subcarrier groups the system can achieve a higher channel capacity. Fig.2.10 shows the payload per subcarrier while changing the number of subcarrier group under different CNRs. The payload of each subcarrier is proved to increase while the number of subcarrier group grows. When the number of group equals to the number of subcarrier, which means every subcarrier is treated as one group and the CFR calculated by GAM method is the same as the original one, the GAM achieves the maximum subcarrier payload since each subcarrier is assigned optimum modulation scheme, but the number of AMI gets saturated as well.
Therefore, it is easy to see that as the number of subcarrier group grows, the number of AMI that required to inform to the receiver increases too. The optimum number of subcarrier groups that could achieve the minimum number of AMI bits with a minimum decrease of channel capacity may be depend on the fluctuation of channel response [18].
From Fig.2.9 and Fig.2.10, when the number of subcarrier group grows larger than 16, system can achieve the channel capacity with a slight loss. AMI is required to transmit with very high transmission quality, since it is used for correct demodulation at the receiver side. One bit of AMI error will lead to the mis-demodulation of one whole group which means a huge degradation of BER performance. One method to ensure the transmission quality of AMI is to modulate AMI by low order modulation scheme, for example BPSK and transmit them with Multi-Carrier Spectrum Spreading (MC-SS) technique. At the receiver side, Maximum Ratio Combing (MRC) method can be employed to compensate the channel fluctuation of each subcarrier. MRC is a special form of general diversity combining, by which multiple replicas of the same information bearing signal received over different diversity branches are combined so as to maximize the instantaneous CNR at the combiner output [19].
23 Tab.2.1 Modulation Scheme and AMI Code
Modulation
Scheme Number Modulation Scheme AMI code
0 NULL 000
1 BPSK 001
2 QPSK 010
3 16QAM 011
4 64QAM 100
Fig.2.8 CFR after Grouping Method
24 Fig.2.9 Loss of Channel Capacity vs. Averaged Received CNR
Fig.2.10 Payload per Subcarrier vs. Group Number
0 5 10 15 20 25 30 35 40 45
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Averaged Received CNR
Loss of Channel Capacity
1Group 8Groups 16Groups 32Groups 256Groups
0 50 100 150 200 250
0 1 2 3 4 5 6 7
Number of Groups
Bits Per Subcarrier
CN=40 CN=30
CN=20
CN=10
25
2.4 Frame Structure
Fig.2.11 shows an example illustration of GAM-OFDM system frame. As shown in the figure, GAM method divides one OFDM frame to N groups and assigns optimum modulation scheme to each group of subcarriers according to the average channel frequency response of these subcarriers. When the channel condition is good, high order modulation scheme such as 64QAM is employed to the group to transmit more data. While the channel condition is not good enough, low order modulation scheme such as BPSK is employed to ensure the required transmission quality. For each group of subcarriers, there is an AMI which takes 3 bits, indicates which modulation scheme is used in the group. The amount of AMI bits is decided by the group number and modulation scheme used to modulate AMI. In addition, AMI bits must be informed to the receiver with very high quality for the correct demodulation of data information because one bit AMI error will lead to the error of the whole group of data. Fig.2.11 shows an example of transmitting AMI which uses several number of symbols in front of the data symbols. AMI bits are usually transmitted to the receiver through the communication or the separate channels which leads to the complexity of transceiver and inefficient usage of frequency resources [20]. Especially in high speed mobile communications between vehicles, the transmission of AMI bits becomes more frequently and sophisticatedly due to the Doppler frequency shift.
Fig.2.11 Illustration of GAM-OFDM Frame
26
2.5 Convolutional FEC Codec
It is well known that (Forward Error Correction) FEC technique can decrease the effect of errors and achieve better communication performances. A convolutional FEC encoder can be represented as a finite state machine in terms of trellis diagram. An (n, k, m) convolutional code can be implemented with a k size input, n size output linear sequential circuit with m input memories. Typically, n and k are small integers with k < n. To convolutionally encode data, start with m size memory registers, each holding one input. Unless otherwise specified, all memory registers start with a value of 0. The encoder has n modulo-2 adders and n generator polynomials. An input bit is fed into the leftmost register at one time. Using the generator polynomials and the value in the remaining registers, the encoder outputs n bits coded data. Then bit shift all register values to the right and wait for the next input bit. If there are no remaining input bits, the encoder continues shifting until all registers have returned to the zero state.
The figure below shows a rate 1/2 encoder with constraint length of 3. Every time after the first bit of input sequence being fed to the register, the encoder generates 2 bits coded data in accordance with the one bit inputted and the two bits in registers. According to the figure, the encoder’s generator polynomials are 𝐺1= (110), 𝐺2= (111). Therefore, the output bits are calculated as follows,
𝑜𝑢𝑡𝑝𝑢𝑡1 = 𝑖𝑛𝑝𝑢𝑡 + 𝑚1 (2.15) 𝑜𝑢𝑡𝑝𝑢𝑡2 = 𝑖𝑛𝑝𝑢𝑡 + 𝑚1+ 𝑚2 (2.16)
Fig.2.12 Rate 1/2 Convolutional Encoder with Constraint Length 3
Data interleaving technique usually is also employed to randomize the burst errors due to fading channel and to prevent the error propagation characteristic of convolution FEC effectively. At the receiver side, the received coded sequence maybe differ from the transmitted sequence due to the noise in the channel. The process of deriving the original data sequence from the received code sequence is called decoding. The maximum likelihood decoding based on the Hamming distance is often used for
27 convolutional codes. The decoder calculates the Hamming distance between the received code sequence and all the sequences which is possible to generate the received sequence. The transmitted code sequence is the sequence which has the minimum Hamming distance with the received code sequence. The decoder then derives the original data from the calculated transmitted code sequence.
The Viterbi algorithm is probably the most widely used decoding method of convolutional codes. This algorithm is a maximum likelihood decoding algorithm, which searches through the trellis to find the path that is most likely to have generated the received sequence. If hard decision decoding is used, this algorithm finds the path that is at the minimum Euclidean distance from the received sequence.
This thesis employs a 1/2 rate convolution FEC encoder with constraint length 7 and hard decision Viterbi decoder to realize data codec. Although, the usage of channel resource decreases due to FEC codec, the error rates of signal improved significantly. The theoretical BER performance of such FEC codec is shown in Fig.2.13. From this figure, it can be observed that FEC technique can improve the BER performance significantly.
Fig.2.13 Effect of FEC under AWGN Channel
0 1 2 3 4 5 6 7 8 9 10
10-6 10-5 10-4 10-3 10-2 10-1
CNR (dB)
BER
No FEC FEC rate=1/2
28
Chapter 3
Proposal of AOFDM-BMD System
This chapter introduces the proposal of Adaptive OFDM system with Blind Modulation Detection (AOFDM-BMD) method. Section 3.1 and 3.2 present the system model and system block diagram.
Section 3.3 introduces the proposal of BMD method which used at the receiver side to detect modulation scheme of each subcarrier group without using any AMI. Section 3.4 discusses the proposal of Adaptive Frame Size (AFS) method which is adopted to improve the channel capacity of system in mobile systems. Section 3.5 introduces CSMA/CA mechanism.
3.1 System Model
Fig.3.1 shows the model of adaptive OFDM system with BMD and AFS methods. Here, it can be assumed that free space path loss and channel frequency response is common for both terminals because the variance of channel frequency response is sufficiently slow over the both way transmission time between user terminal 1 (UT1) and user terminal 2 (UT2). Accordingly, the channel response of multi-path fading during the period can be treated as static. Also assuming that the two terminals have the same transmission signal power level and antenna receiving gain. Therefore, the noise power at the receiver side can be estimated by using the noise power at the transmit side taking into account the channel frequency response. The noise power at transmit side and the channel frequency response can be estimated by using a preamble symbol.
When the UT2 has data information for UT1 as an example, the following procedures are taken for the data communications between UT2 and UT1.
(1) UT2 requests to UT1 for sending a preamble symbol, (2) UT1 sends the preamble symbol upon the request,
(3) UT2 estimates the channel frequency response (CFR) from the received preamble symbol and calculates the averaged carrier to noise power ratio (CNR) for each subcarrier in all groups taking into account estimated CFR,
29 (4) UT2 averages the BERs of all subcarriers within each group to achieve the average BER of each group by GAM method, and assigns the modulation scheme from among BPSK, QPSK, 16QAM and 64QAM, which can satisfy the required PER to each group and sends the data frame to UT1 by using optimal number of symbol per frame according to the AFS method,
(5) UT1 detects the modulation scheme for each group by using the proposed BMD method and demodulates the data information.
Fig.3.1 Proposed System Model
