UWB Synchronization

1
UWB Synchronization

• ???
• 2005/8/2

2
Outline

• Introduction
• Symbol-Differential UWB Systems
• Frame-Differential UWB Systems
• Proposed Algorithm
• Simulations
• Conclusions
• References

3
Introduction

• The synchronization schemes can be divided into
  two strategies as
• Serial search
• It consists of successive detection tests on
  consecutive locations of the received signal.
• This process is stopped when the correlation
  exceeds a predefined threshold.
• Parallel search
• This process will check all possible locations
  and select the maximum correlation value.

4

• The features of the serial search and the
  parallel search
• Serial search
• It has lower hardware complexity.
• However, it needs more search time.
• Parallel search
• It has higher hardware complexity due to the
  receiver architecture.
• It can reduce the total search time.
• Thus, according to different requirements of the
  system, a tradeoff between the hardware
  complexity and the total search time for the
  proposed receiver can be made.

5

• How to achieve synchronization?
• Generally, the receiver design is based on
  correlator-based receivers.
• By the observation of the correlation values can
  determine if the system achieves synchronization.
• In other words, how to capture the signal energy
  is considerable.
• The UWB impulse radio receiver design includes
  as
• Rake receiver
• Transmitted reference receiver
• Differential receiver

6

• The features for three types receivers
• Rake receiver
• It can take full advantage of the available
  signal energy.
• Although the full rake receiver can capture more
  signal energy, it has higher circuit complexity
• It has the challenge of the template signal
  design.

template signal
correlator
a(t-jTr-t0)
c0
r(t)
hj
rake combining
correlator
a(t-jTr-tL-1)
cL-1
7

• Transmitted reference receiver
• Pulses are transmitted in pairs, where the first
  pulse is denoted as the reference pulse and the
  second pulse is the data pulse.
• The reference pulse acts as a template signal to
  correlate the data pulse.
• But TR systems waste communication resources,
  i.e., power and time, to transmit reference
  signals.

data bit 0
data bit 1
D
8

• Differential receiver
• A differential UWB system doesnt transmit
  reference signal, but instead, the data signal in
  the previous is used as a reference signal.
• Since it does not transmit reference signals, it
  can reduce the transmitted power.
• It does not need to design a template signal.
• The differential receivers can be divided into
  two types
• Symbol-differential UWB receiver
• Frame-differential UWB receiver.

9

• Symbol differential approach
• Frame differential approach

symbol i1
symbol i
Delay for Ts
symbol i1
symbol i
D1
D0
10
Symbol-Differential UWB System

• The transmitted signal of the symbol differential
  scheme
• modulating the pulse polarity ( ).
• the transmitted pulse energy.
• the normalized pulse.
• the number of frames in one symbol.
• the frame duration
• the time hopped code.
• the chip duration
• Pulse polarities remain the same during each
  symbol, but will change in the next symbol if a
  -1 is transmitted.

11
Architecture of Symbol-Differential Receiver (
)

• Adaptive synchronization scheme can be divided
  into two steps
• Step 1 Symbol-level synchronization
• Step 2 Frame-level synchronization

12
Symbol-Level Synchronization

• To sample the integrator output twice per frame
  to achieve symbol-level synchronization.
• These sampled values are compared to find the
  highest absolute one, which corresponds to the
  coarse symbol boundary.
• Once symbol-level synchronization is achieved,
  the sampling rate can be decreased to be once per
  symbol.
• Assuming that the estimated start point of symbol
  is at , the sampled output of this symbol
  can be written as

13

• When and the
  SNR is not very low, the inaccuracy of
  symbol-level synchronization can be controlled
  within .
• Since we can not know the exact start point ,
  what we have is only an estimate with
  .
• So the uncertain region has a length of .

14
Frame-Level-Synchronization

• To exclude the noise-only region, we need to
  reduce the integration time in each symbol.
• Since each symbol is composed of frames.
• The new integration region would also be composed
  of sections, which are denoted as
  sub-integration-windows (SIWs).
• Each frame has a SIW, and each SIW has a same
  width at one integration, which is smaller
  than .
• Assuming that the exact start point of symbol
  is at , the integrator output can be
  expressed as the equation

15

• This step splits the continuous integration
  region over a whole symbol into SIWs, each
  with a width
• where
• Let the searching step size to be equal to
• Due to the uncertain region is equal to .
  Thus, altogether
• 
  times of search are needed.
• Assuming that the search starts from the
  symbol, at the search,
  the integrator output is
• with
• where is the estimated start
  place of the symbol.

16
For Example

• 1 Ts 3 Tf
• 1 Tf 4 Tc
• Th codes 0 2 1

Ts
Tf
Tc

• Step 1 Symbol-level synchronization
• Assuming that the symbol-level synchronization
  has been achieved.
• Step 2 Frame-level synchronization
• The uncertain region
• To split the continuous integration region over a
  whole symbol into
• SIWs.
• To find the highest integration output value.

17
Ts 3 Tf, Tf 4 Tc, Th codes 0 2 1
Tf
The estimated start place of the symbol
1st search
2nd search
3rd search
4th search
5th search
6th search
7th search
18
Frame-Differential UWB System

• What is frame-differential UWB system ?
• Each data symbol is modulated by
  .
• A known random sequence is
  differentially modulated on the time-hopped
  pulses, where .
• What is differential modulation scheme ?

For example
symbol i1
symbol i
19

• Here, super-imposing the user data and the
  frame-level binary code, the differentially
  modulated pulse-polarities are obtained as
• and

For example
symbol i1
symbol i
20
For example
symbol i1
symbol i

• The transmitted signal is written
• as the transmitted pulse shape
• The time-instants of the pulses
• Important for this scheme are the time shifts
  between consecutive pulses.

21

• The time shifts between consecutive pulses,
  written as
• 
  and
• The time hopped sequences are all the same for
  each symbol.
• If we know any one of the pulse position, we can
  find the following pulses according to the time
  shifts between each pulse.
• In my proposed algorithm, we let

symbol i1
symbol i
22
Proposed Algorithm Receiver Architecture
Decision
Delay
Delay
Delay
Delay
Delay
Delay
Delay
Delay
Delay
Delay
Delay
Delay
23

• The output value of the first branch can be
  expressed as
• where
• Therefore, each output value over any branch can
  be expressed as
• for

24
Uncertain Region

• Since time hopped sequences are defined as
• where the number of chips within one
  frame duration
• The maximum value of
• Where

25
Decision Rule

• The observation window of each branch is equal
  to one symbol duration.
• The uncertain region is equal to .
• We set the searching step size to be
  
• Thus, altogether times of search are
  needed.
• Each search has output values acquired from
  branches.
• Choosing the maximum output value between these
  output values.
• Finally, choosing the maximum output value
  between all searches.

26
A Reduced Complexity Scheme

• Since total correlators and delay
  elements are needed in this receiver
  architecture, we will try to reduce the number of
  them.
• In reduced complexity case, each branch needs
  correlators and delay elements.
  Generally, A lt 1.
• If A is equal to one, it is the same as the
  previous stated search scheme.
• It can save
  correlators and delay elements.

Reduced complexity
27
Simulations

• Simulation cases
• Multi-user environments
• Different shift step sizes
• Different width of SIWs
• A reduced complexity scheme
• Comparison with symbol differential
  synchronization scheme

28
Code correlation

• Each pulse energy 1/Nf 1/20 0.05

29
Channel model 1 (CM1)

• We can capture 92 total energy, it only takes
  20ns.

30
Channel model 4 (CM4)

• We can capture 92 total energy, it takes about
  66ns.

31
SIW 20, Shift 20
SNR 0 dB
SNR 10 dB
32
SIW 20, Shift 10
SNR 0 dB
SNR 10 dB
33
SIW 10, Shift 10
SNR 0 dB
SNR 10 dB
34
SIW 10, Shift 2
SNR 0 dB
SNR 10 dB
35
DER

• The detection error can also be called as the
  decision error, which describes the situation
  when the inaccuracy of estimated symbol boundary
  exceeds the SIW/2.

SIW/2
SIW/2
0
Timing offset
36
DER

• Purpose
• To compare the DER with different number of users
  (Nu 1, 5 and 10).
• Parameters
• CM1
• SIW 20 (chips)
• Shift step size 10 (chips)
• By the observation
• When the number of user is increased, it will get
  the worse DER.

37
DER

• Purpose
• To compare the DER with different widths of shift
  step size (Shift 20, 10 and 5).
• Parameters
• CM1
• Nu 1
• SIW 20
• By the observation
• When the width of shift step size is smaller, it
  has the better performance.

38
DER

• Purpose
• To compare the DER with different number of SIWs
  included in each branch during one symbol
  duration.
• Parameters
• CM1
• Nu 1
• SIW 20
• Shift step size 5
• By the observation
• When the number of SIWs is reduced, the DER will
  be worse.

39
DER

• Purpose
• To compare the DER with symbol differential
  scheme.
• Parameters
• CM1
• Nu 1
• SIW 20
• By the observation
• The performance of my proposed algorithm is
  better than the performance of symbol
  differential scheme.

40
BER

• Purpose
• To compare the BER with different number of users
  (Nu 1, 5 and 10).
• Parameters
• CM1
• SIW 10 (chips)
• Shift step size 2 (chips)
• By the observation
• When the number of user is increased, it will get
  the worse BER.

41
BER

• Purpose
• To compare the BER with different width of shift
  step size (Shift 2 and 5).
• Parameters
• CM1
• Nu 1
• SIW 10 (chips)
• By the observation
• When the width of shift step size is smaller, it
  has the better performance.

42
BER

• Purpose
• To compare the BER with symbol differential
  scheme.
• Parameters
• CM1
• Nu 1
• SIW 10 (chips)
• Shift step size 2 (chips)
• By the observation
• The performance of my proposed algorithm is
  better than the performance of symbol
  differential scheme.

43
BER

• Purpose
• To compare the BER with different width of SIWs
  in the corresponding synchronization inaccuracy.
• Parameters
• CM1
• Nu 1
• SNR 5 dB
• By the observation
• When the width of SIW is smaller, it get the
  worse performance.

44
BER
-10
45
BER

• Purpose
• To compare the BER with different number of SIWs
  for each branch during one symbol duration.
• Parameters
• CM1
• Nu 1
• SNR 5 dB
• By the observation
• When the number of SIWs is reduced, the BER will
  be worse.

46
Conclusions

• The differential receiver has an advantage that
  it does not need to spend extra efforts designing
  the template signal.
• An effective synchronization algorithm for
  multi-user environment is proposed in this
  thesis.
• In the architecture of proposed receiver, the
  uncertain region is smaller than two frame
  durations . It is helpful to reduce the search
  time.
• It is a tradeoff between the performance and the
  circuit complexity.
• The BER of our proposed algorithm is better than
  the symbol-differential synchronization
  algorithm.

47
References

• The references about this slides
• A.-J. van der Veen, A. Trindade, QH Dang, and G.
  Leus, Statistical Analysis of a
  Transmit-Reference UWB Wireless Communication
  System, ICASSP, Mar. 2005.
• N. He and C. Tepedelenlioglu, Adaptive
  synchronization for non-coherent UWB receivers,
  ICASSP, Montreal, CA, May 2004.
• M. Ho, V. S. Somayazulu, J. Foerster, and S. Roy,
  A differential detector for an ultra-wideband
  communications system, in IEEE Vehicular
  Tecnology Conference, VTC, Spr. 2002.
• K. Witrisal and M. Pausini, Equivalent system
  model of ISI in a frame- differential IR-UWB
  receiver, in IEEE Global Telecommunications
  Conference, GLOBECOM, Dallas, Dec. 2004.
• K. Witrisal, M. Pausini, and A. Trindade,
  Multiuser interference and inter-frame
  interference in UWB transmitted reference
  systems, in International Workshop on
  Ultra-Wideband Systems, IWUWBS, Kyoto, May 2004.

48
References

• Suggested readings for UWB synchronization
  research
• J. Oh, S. Yang, Y. Shin, A rapid acquisition
  scheme for UWB signals in indoor wireless
  channels, in IEEE Wireless Communications and
  Networking Conference, WCNC, Mar. 2004.
• L. Reggiani, GM Maggio, A reduced complexity
  acquisition algorithm for UWB impulse radio,
  UWBST, Reston (VA, USA), Nov. 2003.
• E. A. Homier and R. A. Scholtz, Rapid
  acquisition of ultra-wideband signals in the
  dense multipath channel, in IEEE Conference on
  UWB systems and Technologies, May 2002.
• S. Aedudodla, S. Vijayakumaran, and T. F. Wong,
  Timing Acquisition in Ultra-wideband
  Communication Systems, IEEE Transactions on
  Vehicular Technology, 2005. To appear.