A Novel Method of Carrier Frequency Offset Estimation for OFDM Systems -Mingqi Li and Wenjun Zhang...

A Novel Method of Carrier Frequency Offset Estimationfor OFDM Systems

-Mingqi Li and Wenjun ZhangIEEE Transactions on Consumer 966 Electronics, Vol. 49, No. 4, NOVEMBER 2003

Advisor: Yung-An KaoStudent: Chian Young

outline

• Intoduction

• OFDM system Model

• Continual-based CFO estimation algorithm

• Scattered-based CFO estimation algorithm

• Performance evaluation

Introduction(1/2)

• This paper proposed a method using scattered pilots to estimate integer and fractional CFO, which are usually estimated by continual pilots.

• Ex: DVB-t pilot structure:

Introduction(2/2)

• The advantage of using scattered pilots to estimate CFO:– Scattered pilot are usually use to estimate

channel freq. response (CFR), using them at the same time to estimate freq. offset.

will not bring about additional cost for the system

Increasing spectral efficiency by saving the use of continual pilots

OFDM system model(1/2)

• The N samples of OFDM symbol l for OFDM demodulation(RX signal):

• al,k: tx signal at k-th subcarrier l-th OFDM symbol

Hl,k: channel freq. response

Ng: guard interval length N: FFT length

Ns= Ng +N

wl,k: complex value of AWGN noise

ξ samples of timing offset

Integer part of CFO

Fractional part of CFO

OFDM system model(2/2)

• The demodulation complex data k-th subcarrier of OFDM symbol l:

• attenuation:

offset by integer CFO

Continual-based CFO estimation algorithm-estimation of integer CFO

• The algorithm estimates integer CFO according to the offset of the correlating peak of continual pilots in two consecutive OFDM symbol:

• Accumulating all correlation samples xk belonging to the special set

k←C+m, integer CFO searching range :m← I, I=[-171,171]

• By searching the maximum absolute value |yl,m|, we obtain integer CFO at OFDM symbol l:

Pilot index

data index

Continual pilots-Based CFO estimation algorithms-estimation of fractional CFO

• After compensating the integer CFO, zl,k becomes:

• correlating continual pilots in demodulation output of two consecutive OFDM symbols l −1 and l (assume the channel and timing offset is constant during two OFDM symbol):

• the fractional CFO estimate at OFDM symbol l :

Scattered-based CFO estimation algorithm-estimation of integer CFO(1/3)

• The scattered pilots are known in the receiver. The four scattered

pilots vector:

• Moving correlating respectively with demodulation output:

• According to the maximum peak, it can be decided that which kind of pilot can be transmitted:

• Then according to the indices offset of the peak value, integer CFO estimate at OFDM symbol l :

Q: total number of scattered pilots

Scattered-based CFO estimation algorithm-estimation of integer CFO(2/3)

• In the matched case, the peak value is:

to eliminate timing offset

the peak value:

Sl,n+Sl,Q-n+1 is constant for all n:

the peak value:

Scattered-based CFO estimation algorithm-estimation of integer CFO(3/3)

To reduce the effect of CFR

the peak value:

The sum item is less effect by CFR

Scattered pilots-Based CFO estimation algorithms-estimation of fractional CFO

• assume that receiver is correct sampling and the received two consecutive OFDM symbols have same timing offset

• correlating scattered pilots in demodulation output of two consecutive OFDM symbols l −1 and l:

m=(lmod4),l=0,1…∞, ⊕denotes addition module 4

• the fractional CFO estimate at OFDM symbol l:

Simulation environment and parameters

• Channel:– AWGN channel– Fixed multipath Rayleigh fading channel with 20 paths– Time varying multipath channel (GSM TU with six taps Jacks fa

ding model)

• OFDM parameters:– DVB-T system 2k mode– 1/16 guard interval– 16 QAM modulation– Timing offset is 10 samples ,but no ISI

Integer CFO estimation performance

• CFO=10.3Δf, 5000 OFDM symbols for the simulation• Prob. of failure :

Prob. of failure in AWGN and fixed multipath channel Prob. of failure in time-varying channel

Fractional CFO estimation performance

• The initial CFO is set to 0.1Δf

Tracking performance of fractional CFO In time varying channels

Fractional CFO estimation performance

BER performance of fractional CFO In time varying channels with initial CFO=0.1Δf

BER performance of fractional CFO In time varying channels with initial CFO=0