1. Synchronization Algorithms for OFDM Systems
Denis Yusim
Evgeniy Priziment
Presented by:
Supervised by:
Dr. Shlomo Greenberg
Mr. Ron Bercovich

2. Synchronization problems in OFDM Synchronization algorithms
Outline
IEEE background
OFDM basics
Synchronization problems in OFDM
Synchronization algorithms

3. Escalating business demand for:
IEEE Background
Escalating business demand for:
rapid Internet connection
integrated data, voice and video services
BWA is an affordable alternative to wired solutions which can be easily and quickly deployed
BWA was limited due to lack of a universal standards
IEEE standard was developed to make BWA more widely available
The standard specifies the PHY and the MAC layers for wireless metropolitan area networks

4. Global Wireless Standards
IEEE Bluetooth
WAN
MAN
LAN
PAN
ETSI HiperPAN
IEEE WirelessLAN
ETSI HiperLAN
IEEE WirelessMAN
ETSI HiperMAN & HIPERACCESS
IEEE
(proposed)
3GPP, EDGE (GSM)

5. T1+ LEVEL SERVICE ENTERPRISE FRACTIONAL T1 FOR SMALL BUSINESS
IEEE Applications
RESIDENTIAL & SOHO DSL
ALWAYS BEST CONNECTED
802.11
802.11
802.11
802.16
T1+ LEVEL SERVICE ENTERPRISE
BACKHAUL FOR HOTSPOTS
INTERNET BACKBONE
FRACTIONAL T1 FOR SMALL BUSINESS

6. Synchronization problems in OFDM Synchronization algorithms
Outline
IEEE background
OFDM basics
Synchronization problems in OFDM
Synchronization algorithms

7. OFDM Orthogonal Frequency Division Multiplexing OFDM - What is it?
Multi - Carrier Transmission technique
Subcarriers are spaced by 1/Ts
ORTHOGONALITY - The peak of each signal coincides with trough of other signals

8. OFDM – Principles
FFT is the basic technique to generate overlapped orthogonal OFDM tones
dN-1
SN-1
Data In
Modulation
(QPSK,QAM,
BPSK, etc)
S/P
IFFT
D/A d0 S0
Baseband
OFDM signal
Data Out
De-Modulation
(QPSK,QAM,
BPSK, etc)
P/S
FFT
A/D

9. Adaptive Modulation
Modulation is chosen based on the channel conditions
Closer the range – higher order modulations (QAM16, QAM64)
Further the range – lower modulations (BPSK)

10. OFDM Advantages High Spectrum Efficiency
Guard bands f
FDM
OFDM
High Spectrum Efficiency
The information is spread over frequencies, without any frequency guard bands
Equalization is very simple compared to Single-Carrier systems

11. OFDM Advantages (cont.)
OFDM has excellent robustness in multi-path environments
Cyclic prefix preserves orthogonality between subcarriers

12. Synchronization problems in OFDM Synchronization algorithms
Outline
IEEE background
OFDM basics
Synchronization problems in OFDM
Synchronization algorithms

13. The goal is to determine where to start the FFT
Time Offset
Symbol # n-1
Symbol # n
Symbol # n+1
Optimal FFT window
Late FFT window
Early FFT window
The goal is to determine where to start the FFT
FFT window can start from any sample within the Guard Period due to FFT cyclic properties
When FFT window spans samples from two consecutive OFDM symbols, inter-OFDM-symbol interference (ISI) occurs

14. Frequency Offset
Frequency offset is a multiple of a sub carrier spacing
Sub carriers are still orthogonal
Received data symbols are in the wrong position in the demodulated spectrum, resulting in a BER degradation
Frequency offset is a fraction of sub carrier spacing
Loss of the orthogonality
BER of the system deteriorates

15. Synchronization problems in OFDM Synchronization algorithms
Outline
IEEE background
OFDM basics
Synchronization problems in OFDM
Synchronization algorithms

16. Uplink Synchronization
The synchronization is based on the uplink preamble symbol
Preamble
Data
From Subscriber # k
From Subscriber # k+1
The preamble consists of one OFDM symbol utilizing only even subcarriers
The time domain waveform consists of 2 times 128 samples preceded by a CP

17. Symbol Timing Estimation by M. Schmidl and C. Cox
Time θ X X
= M(θ)
R(θ)2
2|P(θ)|2

18. Symbol Timing Estimation by M. Schmidl and C. Cox
The estimator is defined as:
where
is the proposed metric
is the correlation between the samples in two halves
is the energy of the entire symbol

19. Symbol Timing Estimation by M. Schmidl and C. Cox

20. ML Algorithm by M. Sandell, et al.
The estimator is defined as:
where
is the proposed metric
is the energy of the signal sample
is the energy of the noise sample

21. ML Algorithm by M. Sandell, et al.

22. Modified ML Algorithm for IEEE 802.16 uplink preambleCP …
The algorithm is based on calculation of six correlations between three windows , and and correlations of two windows and In addition the energy within these five windows is calculated. These windows slide through an observation window.

23. Modified ML Algorithm for IEEE 802.16 uplink preamble

24. Frequency Offset Estimation
is shifted in frequency by
is a sampled signal at intervals

25. Frequency Offset Estimation (Cont’)
Within the preamble symbol and are identical
Averaging over all pairs and calculating an argument

26. Frequency Offset Estimation (Cont’)

27. Simulation Results
For Cyclic Prefix 1/16

28. Simulation Results (Cont’)
For Cyclic Prefix 1/32

29. Questions?
Thank you!