Timing synchronization under multipath fading channel

Hi
I am looking for robust timing and frequency synchronization algorithm
based on preamble detection for a single carrier non ofdm(orthogonal
frequency division multiplexing) system.most of the papers on
synchronization deal with ofdm system, which is based on auto-correlation
of identical short training symbols proceeded by cyclic prefix.currently I
am trying to perform timing synchronization over the received noisy signal
using simple cross-correlation and then frequency offset estimation using
dual-PLL(phase locked loop), however this technique fails when the channel
is multipath fading and frequency selective channel  e.g HF communication,
so I would appreciate your help if you could suggest me some robust timing
synchronization algorithm and frequency offset estimation algorithm which
can estimate large Doppler shift.

Thanks

asia wrote:

> Hi
> I am looking for robust timing and frequency synchronization algorithm
> based on preamble detection for a single carrier non ofdm

Classics: Costas Loop. Gardner Loop. BETR for complex signals.

(orthogonal
> frequency division multiplexing) system.most of the papers on
> synchronization deal with ofdm system, which is based on auto-correlation
> of identical short training symbols proceeded by cyclic prefix.currently I
> am trying to perform timing synchronization over the received noisy signal
> using simple cross-correlation and then frequency offset estimation using
> dual-PLL(phase locked loop), however this technique fails when the channel
> is multipath fading and frequency selective channel  e.g HF communication,
> so I would appreciate your help if you could suggest me some robust timing
> synchronization algorithm and frequency offset estimation algorithm which
> can estimate large Doppler shift.

Until the frequency offset is less then a baud rate, all you need is a 
Costas Loop. It gets tricky if the offset is comparable with the baud; 
then you have to estimate the carrier frequency beforehand.


Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com

asia wrote:


> I would appreciate your help if you could suggest me some robust timing
> synchronization algorithm and frequency offset estimation algorithm which
> can estimate large Doppler shift.

The estimation of the large frequency offset of the data signal in the 
presence of noise could appear pretty heavyweight problem. This could be 
done by n-th power device (n = order of the phase modulation) and then 
an estimator of the frequency. If the SNR in the bandwidth (baudrate + 
shift) is high enough (say, ~10 dB), then a simple discriminator would 
do. If the SNR is low, then you will need better estimator (i.e. FFT or 
MUSIC). As the problem exhibits a threshold behavior, you may have to 
accumulate a lot of data to get useful estimate.

Alternatively, you can sweep your receiver up or down in the frequency 
and detect the moment of locking on the signal. Or run multiple 
receivers shifted in the frequency in parallel. Or sweep multiple 
receivers (maximum likelihood). This is not going to break the 
theoretical limits however it could be more or less preferable from the 
application standpoint.


Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com

On Jun 18, 5:19=A0am, "asia" <hilwa_24@n_o_s_p_a_m.yahoo.com> wrote:
> Hi
> I am looking for robust timing and frequency synchronization algorithm
> based on preamble detection for a single carrier non ofdm(orthogonal
> frequency division multiplexing) system.most of the papers on
> synchronization deal with ofdm system, which is based on auto-correlation
> of identical short training symbols proceeded by cyclic prefix.currently =
I
> am trying to perform timing synchronization over the received noisy signa=
l
> using simple cross-correlation and then frequency offset estimation using
> dual-PLL(phase locked loop), however this technique fails when the channe=
l
> is multipath fading and frequency selective channel =A0e.g HF communicati=
on,
> so I would appreciate your help if you could suggest me some robust timin=
g
> synchronization algorithm and frequency offset estimation algorithm which
> can estimate large Doppler shift.
>
> Thanks

What sort of channel model are you assuming, i.e., what is the
multipath model?

hi
I am using frequency selective Rayleigh fading channel with some Doppler shift in independent paths, for example I have used the Math lab function rayleighchan(8000,100,[0 0.0025 0.005 0.0075],[0 -.155 -.301 -1]) and my baud rate is 400, where 8000 is sampling frequency and 100Hz is Doppler shift and the first four values stand for delay between discrete paths and the second figure is their corresponding path gain in dB. so in my math lab simulation i have generated random binary data and I have added PN sequence bits as preamble for timing synchronization purpose in receiver. then after pulse shaping I have modulated the frame using BPSK and I finaly passed this signal through the channel model and I want to perform timing synchronization or frame detection on the recieved signal however timing synchronization using correlation fails when the signal faces multipath fading along with Doppler shift, so I would like your sincere suggestion on this 
thanks  

---
frmsrcurl: http://compgroups.net/comp.dsp/Timing-synchronization-under-multipath-fading-channel