Thesis Organization

The thesis is organized as follows. In Chapter 2, we explain the SCBT system and give the overview of IEEE 802.15.3c standard. Chapter 3 deals with the detailed mathematical explanation of the proposed FDE. The corresponding hardware design with some low-power architecture techniques and simulation results are presented in Chapter 4. Finally, Chapter 5 is the conclusion and the future work.

Chapter 2

Overview of IEEE 802.15.3c Standard

2.1 IEEE 802.15.3c Specifications

IEEE 802.15.3c is a wireless communication standard based on both SC and OFDM transmission. It uses Common Mode Signaling (CMS) and a preamble attached in front of the data stream. The CMS is specified to enable interoperability among different PHY modes. After the CMS, the frame payload is transmitted in different PHY modes. The preamble is added to aid receiver algorithms related to AGC setting, antenna diversity selection, timing acquisition, frequency offset estimation, frame synchronization, and channel estimation. With insertion of CP, it can reduce the impact of ISI. With channel coding, the system can correct the transmission errors. Furthermore, the standard defines interleaving, scrambler, unequal channel coding, and several modulation scheme to achieve better performance.

2.1.1 Basic Specifications

In this standard, there are three transmission modes: Single Carrier (SC) mode, High Speed Interface (HSI) mode, and Audio/Visual (AV) mode. HSI and AV mode use OFDM transmission, and SC mode is single carrier transmission. Some parameters are listed in Table 2-1 for only SC mode.

The RF band occupies 9 GHz bandwidth while the sampling rate is 1728 MHz as

shown in Fig. 2-1. The standard indicates that the RF band is divided into four sub-bands such that the Nyquist bandwidth of each sub-band is exactly 1728MHz. In addition, each of sub-bands has 432 MHz spacing to prevent the interference from each other. In this case, the RF band can support 4 transmission bands without any interference.

Fig. 2-1 RF band plan

Table 2-1 SC mode specifications

Description Value Unit

Sampling rate 1728 MHz

Sampling period ~0.579 ns

Subblock length 512 samples

Pilot word length 0 64 samples

Data samples per subblock 512 448 samples

Radio frequency band 57~66 GHz

Modulation schemes π/2 BPSK, π/2 QPSK, π/2 8-PSK, π/2 16-QAM

FEC type

RS(255,239), LDPC(672,336), LDPC(672,504), LDPC(672,588),

LDPC(1440,1344) Required frame error rate (FER)* 8%

*: FER is determined at the PHY Service Access Point interface after any error correction methods.

The measurement shall be performed in AWGN channel with a frame payload length of 2048 octets.

The standard supports two types of pilot word length: 0 or 64. Transmission with pilot words decreases the data rate, but there are benefits. The pilot word is designated for timing tracking, compensation for clock drift, and compensation for frequency offset error. With these additional supports, the receiver can achieve better performance in long term. The pilot word can also act as the guard interval and the cyclic prefix, which are very useful for the frequency domain equalization. On the other hand, transmission without any pilot word can make the data rate higher but comes with concern of performance loss. It’s a trade-off between the data rate and the performance.

The data is modulated by π/2 BPSK, π/2 QPSK, π/2 8-PSK, or π/2 16-QAM before transmitting in SC mode. The π/2 BPSK is a binary modulation with π/2 phase shift counter clockwise. The mathematical description is:

n*

n n

z

=

j d

(2.1)

, where dn indicate the data samples and zn are the constellation points.

As shown in Fig. 2-2(a), the π/2 BPSK is equivalent to MSK, which is the continuous phase modulation. The main purpose of the continuous phase modulation is to eliminate the discontinuity between the waveforms. The discontinuity will result in high frequency components in the waveform such that the transmitter requires a power amplifier with larger linear region. There are many applications using this special property of π/2 BPSK, such as symbol timing tracking [12] and differential receiver [13]. The π/2 QPSK, π/2 8-PSK, and π/2 16-QAM are also doing the π/2 phase shift after the mapping with gray encoding. The reason of the π/2 phase shift is to obtain a simple implementation aligning with the π/2 BPSK. The constellation

diagram is shown in Fig. 2-2(b) (c) (d).

Fig. 2-2 Constellation maps: (a) π/2 BPSK, (b) π/2 QPSK, (c) π/2 8-PSK, (d) π/2 16-QAM

2.1.2 Concept of Single Carrier Block Transmission

The standard indicates that the pilot words should be inserted into the data stream every 448 data samples, so the data stream is divided into several small subblocks, which lead to the block transmission, as shown in Fig. 2-3. These pilot words are used for timing tracking, compensation for clock drift, and compensation for frequency offset error. Furthermore, the pilot words act as the cyclic prefix and enable the

frequency-domain equalization.

Fig. 2-3 Frame format of block transmission

The inter-block-interference (IBI) and ICI affect the received signal of SCBT due to the channel distortion. Inserting the guard interval which length is longer than the length of the channel impulse response can eliminate the IBI since the previous subblock can not interfere the incoming subblock. In order to prevent the ICI while using the FDE, the guard intervals in the front and the back of the subblock should be the same, and this is so called the cyclic prefix. The CP can maintain each subchannel to be orthogonal and reduce the effect of ICI. Besides, the CP translates the linear convolution into the circular convolution, which results in a simple diagonal channel matrix in frequency domain. As the result, evaluating the filter coefficients is much easier than in time domain. We will explain the fact in the following paragraphs.

The received subblock rn and its frequency domain form Rk is described as follows, where dn is transmitted signal and h is CIR with length L and are extended to length N by inserting 0 into CIR and N is the length of the subblock.

1 1 2

Since dn is cyclic prefixed and periodic, the equation can be written as:

1 1 2

Because of the circular convolution, we can easily recover the transmitted data and evaluate the filter coefficients:

ˆⁿ { ˆ^k}

2.1.3 Equalization Related Specifications

In this standard, there are some well known data streams that are assigned to different specific purposes, i.e. MAC layer control signal, the piconet coordinate signal, or the performance improving signal. This section will introduce the specific signaling directly related to the equalization.

„ Common Mode Signaling(CMS)

The CMS is a low data rate SC mode and specified to enable the switching among different PHY modes. It’s also used for transmission of the beacon frame, sync frame, command frame, and training frame in the beamforming procedure. The forward error correction (RS(255,239)) and code spreading (spreading factor: 64) are used to ensure the correctness of the CMS. A PHY preamble is added to aid the receiver algorithms such as AGC setting, timing acquisition, frame synchronization, and channel estimation.

As shown in Fig. 2-4, the preamble is prior to the frame header and PHY Payload field. It consists of the Golay complimentary sequences a128 and b128 of length 128, listed in Table 2-2. The code u512

is constructed as below:

512 [ 128 128 128 128]

u

=

a b a b

(2.5)

, where the binary-complement of a sequence x is denoted as

x .

Fig. 2-4 CMS frame format and preamble structure

Table 2-2 Golay sequences

Sequence name Sequence value

a

128 C059950CC0596AF33FA66AF3C0596AF3

b

128 30A965FC30A99A03CF569A0330A99A03

The SYNC field, consists of 128 repetitions of a128, is used for frame detection.

The main purpose of SFD field is to validate the beginning of a frame. The CES field is assigned to do channel estimation.

As mentioned in Section 2.1.2, the SCBT is suitable for frequency domain equalization due to the CP. Although there is no periodic stream in u512, it becomes cyclic prefixed with

b in previous u

₁₂₈ 512. Combined with the leading

b , the first u

₁₂₈ 512

is also cyclic prefixed. As the result, there are 6 subblocks available in the CES field to do channel estimation.

„ PHY preamble

After the CMS, the system will switch to the designated mode and begin to transmit the data payload. In each beginning of transmission, the transmitter will send a PHY preamble to aid the receiver algorithms, just like the one in CMS. In SC mode, the preamble is transmitted at the rate of 1728 MHz. The PHY preamble structure is shown in Fig. 2-5.

Fig. 2-5 PHY preamble structure

Like CMS preamble structure, the PHY preamble consists of SYNC, SFD, and CES field. Each of the field functions like the one in CMS preamble: SYNC field for frame detection, SFD field for validating the beginning of the frame, and CES field for channel estimation. The Golay sequence

b and

₁₂₈

u ensures the cyclic prefix

₅₁₂

property and is also useful information for frequency domain equalization.

„ Pilot Channel Estimation Sequence(PCES)

The PCES insertion is an optional feature that allows the system to re-acquire the channel information periodically. To add the PCES, the data stream is divided into data blocks with each data blocks has 64 subblocks, as shown in Fig. 2-3. Each data block is followed by a PCES. The PCES is the Golay sequence a128 followed by the CES field in the PHY preamble mentioned above and is shown in (2.6). Since PCES contains the information of the CES field, this cyclic prefixed signal can provide the channel information periodically.

128 128 128 128 128 128

[ ]

PCES

=

a b a b a b

(2.6)

2.2 Channel Model

IEEE 802.15.3c standard is for an indoor, over GHz data rate, wireless communication system using 60 GHz RF band. In 60 GHz RF band environment, the channel model has some special properties that are much different from those below 10GHz RF band channel. These properties are listed below:

„ High Path Loss

While the EM wave passes through the medium, the medium absorbs the energy and limits the distance that the EM wave can travel. The more energy it lost, the shorter it can travel. The ratio of energy loss is mainly depends on the characteristic of the medium and the EM wavelength. The wavelength of 60 GHz wave is close to the length of the oxygen chemical bond, so the wireless communication in 60 GHz RF

band suffers the tremendously high path loss. As the result, the transmission distance is limited to about 10 m in maximum. Moreover, the effect of the multi-path fading is reduced since the non-line-of-sight (NLOS) wave travels more distance and loses more energy than the line-of-sight (LOS) wave.

„ Strong Directivity

The strong directivity means that the EM wave energy almost centralizes in the small angle. From the formula of the diffraction:

2 0

( ) sinc (

d

sin )

I θ I θ

=

λ

(2.7)

,where I is the intensity profile and basically is a sinc function related to the diffraction angle θ and the wavelength λ. For a small λ, the intensity drops rapidly while θ increasing. This phenomenon shows that the antenna can only transmits or receives the signal in the small angle. In conclusion, the NLOS path has lower path gain relative to the LOS path, and the multi-path fading effect is once more reduced.

The channel model is based on the golden set released by IEEE 802.15.3c group [14]. The golden set, shown in Fig. 2-6, 2-7, provides a set of the static channel models in 60 GHz RF band and the one with the largest RMS delay spread is chosen as the worst case channel model, which is 12.73 ns. To simulate the time-variant effect, the Jakes model is used as the Doppler Effect. Considering the human moving speed in indoor environment, the relative velocity is assumed to be 4.5 km/h, which is 250 Hz frequency shift according to (2.8):

0

f v f

Δ = −

c

(2.8)

,where v is for relative velocity between the receiver and the transmitter, c for velocity of EM wave, and f0 for the frequency of the carrier. The negative sign means the receiver is moving toward to the transmitter. The frequency shift is 7.41*10^-3% of the subcarrier spacing (3.375MHz).

Fig. 2-6 Channel impulse response

0 50 100 150 200 250 300 350 400 450 500

Fig. 2-7 Channel frequency response

2.3 Comparison of Time and Frequency Domain Equalizer

The main functional blocks of the baseband receiver are shown in Fig. 2-8. After the signal is sampled by ADC, we use synchronizer to detect the boundary of the frame. Then, the data is equalized to eliminate the channel effect and sent to channel decoding. The purpose of channel decoding is to correct the error bits using the algorithm of channel coding.

Symbol/Preamble

Fig. 2-8 Block diagram of receiver

We have overviewed the channel model in both time and frequency domain in Section 2.2. In this section, the discussion focuses on the comparison of time and frequency domain equalizer in terms of the computational and hardware complexity base on the channel model.

„ Time Domain Equalizer(TDE)

The basic structure of the TDE is the FIR filter, which performs the convolution between data stream and the filter coefficients. A simple illustration of the FIR filter is shown in Fig. 2-9.

………

D D ……… D

＊ ＊ ＊

∑

………

input

＊

W0 W1 W2 w_n

output

Fig. 2-9 FIR filter structure

The computational complexity of the convolution is proportional to the length of the filter coefficients, which is determined by the length of the CIR. From the channel model in Fig. 2-6, the filter coefficients must satisfy the mathematical property in Eqn.

(2.9). By this property, the filter coefficients can be derived as shown in Fig. 2-10.

[ ]

* 1 0 0 0

h w= (2.9)

From Fig. 2-10, the required number of the filter coefficients is 104, which is much longer than the wired communication system [15]. If the structure illustrated in Fig. 2-9 is used, then the number of the required complex multiplications is also 104.

Therefore, the operation time for the 104 complex multiplications is one sampling period. It’s almost impossible to implement the hardware with GHz sampling rate indicated by the standard. Although the parallel design can increase the throughput of TDE, the complexity grows linearly with the number of coefficients.

0 10 20 30 40 50 60 70 80 90 100 110

Fig. 2-10 FIR filter coefficients for TDE

„ Frequency Domain Equalizer(FDE)

A simple illustration of fully parallel FDE is shown in Fig. 2-11. The input passes through Serial-to-Parallel block and transforms to frequency domain by FFT. Then, the frequency domain data is multiplied with coefficients W and then transformed back to time domain by IFFT. Unlike TDE, the number of coefficients in FDE is fixed no matter how the length of the channel impulse response changes. The potential problem is when the length of the CIR is longer than the length of the CP. In that case, the circular convolution is ruined and FDE fails to equalize the channel effect.

However, the channel model shows that the maximum length of CIR is far less than the length of CP, so we do not consider the problem in the thesis.

The advantage of FDE is lower computational complexity than TDE when the length of CIR is long enough. Assuming that the computational complexity of

FFT/IFFT is NlogN, the total computational complexity of FDE is 2NlogN+N, where

N is 512 based on the standard. Therefore, the average computation on one sample is 2logN+1, which is 19. Moreover, the equalization can be reduced using the

pipeline-based FFT/IFFT as shown in Fig. 2-12. This pipeline-based design can be parallelized easily by adding multipliers of the equalization. Thus, it can be used in high sampling rate communication system without too much overhead. After some modifications, the FDE can also support OFDM mode, and the overhead of FFT is reduced.

Fig. 2-11 Structure of fully parallel FDE

Fig. 2-12 Structure of fully serial FDE

In summary, TDE does not need CP aided transmission, but it suffers from the

problem of large computational complexity in the hardware design. On the other hand, FDE is suitable for high sample rate since it’s easy to do parallelism. The drawback is that additional CP reduces the data rate. The comparison is listed in Table 2-3.

Table 2-3 Comparison between TDE and FDE

TDE FDE Coefficient number

Proportional to the length of

CIR (104) Subcarrier number (512)

Complexity

High (104 multiplications

on one sample)

Low (19 multiplications on one sample)

Data throughput

High (sampling rate*1)

Low

(sampling rate*448/512)

Storage requirement

104 coefficients 512 coefficients

Compatible with OFDM

No Yes

2.4 System Requirements and Design Considerations

The IEEE 802.15.3c standard indicates the system requirements mentioned in Section 2.1.1. One of the requirements is the 1728 MHz sampling rate. The sampling rate equals to the throughput of the system. The throughput is increased with the improvements of CMOS process and architecture, but over GHz throughput is still a challenge for hardware design. Moreover, high throughput also means that high power consumption in the digital circuit. Also, the choice of the architecture determines the power consumption. Hence, there two design considerations on high sampling rate:

architecture and power consumption.

The pipelined structure and parallel structure are commonly used for high throughput design. The pipelined structure can increase the clock rate by inserting the registers into the combinational circuit. However, the dynamic power is also increased

with clock rate. On the other hand, parallel structure increases the throughput by copying the structure without increasing the clock rate, but the area and static power grows with the number of copies. Hence, our considerations on architecture and power consumption mainly focus on how many copies we want and how fast the clock rate is.

Another system requirement is the bit error rate (BER). According to the FER and frame size, the required BER is 1.54*10^-4. Hence, the design consideration is the performance and the cost of computational complexity. First of all, we consider the algorithm that can be realized in hardware design. Since the throughput is very high, we should keep the complexity as low as possible. Then, we should consider the channel model. The channel model in Section 2.2 contains Doppler Effect, so it is time-variant. Thus, the algorithm of the equalizer must have the ability to update its coefficients with time. Third, the length of the training sequence is determined by the standard as mentioned in Section 2.1.3, so the algorithm should be ready within the training stage. Hence, we have to choose the reasonable computational complexity algorithm which satisfies the BER requirement.

Chapter 3 Fast Convergent Adaptive Frequency Domain Equalizer

3.1 Review of Frequency Domain Equalization

In Section 2.1.2, we derive the formula of circular convolution, which can be transformed into a simple multiplication in the frequency domain:

R H D

= ⋅ (3.1)

,where H is a diagonal matrix. To recover the transmitted data, we multiply the inverse of H on both sides of equation:

1 1

H

⁻ ⋅ =

R H

⁻ ⋅ ⋅ =

H D D

(3.2)

, where the inverse of H is also a diagonal matrix. After IFFT and CP removal, we can fully recover the transmitted signal dn.

The above equations describe the ideal case: no AWGN and time-invariant channel. In reality, the white noise always exists due to the thermal noise, and the channel varies with time due to many effects, such as related movement, air flow, or moving object. Thus, the equation should be:

k k( ) k k k

R

=

J t H D

⋅ ⋅ +

N

(3.3)

the subchannels. If we simply multiply the inverse of Hk all over the time, the time-variant effect will corrupt the data. Furthermore, to get the accurate inverse of Hk

is a difficult job under AWGN. To break through the predicament, the first thing is to overcome AWGN and get the inverse of Hk as accurate as possible. Then, an adaptive algorithm is performed to track the changes in the time-variant channel. In this way, the time-variant component Jk

(t) is no more a trouble in the equalization. Based on the

idea, the block diagram of the proposed adaptive FDE with channel estimation is shown in Fig. 3-1. The LS channel estimation evaluates the initial value of coefficients by using the CMS and the preamble as the training sequence. Then, the data payload is transmitted and equalized by FDE. The LMS adaptive algorithm updates the coefficients against the time-variant channel.

Fig. 3-1 Block diagram of the proposed FDE

3.2 Channel Estimation

In the beginning of the transmission, the transmitter sends the training sequence

u

512 located in CES field of CMS to assist the equalization as shown in Fig. 2-4. With the training sequence, we can easily estimate the channel matrix Hk, which is the

In the beginning of the transmission, the transmitter sends the training sequence

u

512 located in CES field of CMS to assist the equalization as shown in Fig. 2-4. With the training sequence, we can easily estimate the channel matrix Hk, which is the

在文檔中 應用於單載波室內無線接收器之快速適應頻率域通道等化器之設計 (頁 15-0)