Schmidl & Cox OFDM synch.: Difference between revisions

From GNU Radio
Jump to navigation Jump to search
(Fix formatting issues, and missing words)
 
Line 3: Line 3:


; Input: complex samples.
; Input: complex samples.
; Output 0: Fine frequency offset, scaled by the OFDM symbol duration. This is  in [1]. The normalized frequency offset is then 2.0*output0/fft_len.  
; Output 0: Fine frequency offset, scaled by the OFDM symbol duration. This is <math>\hat{\varphi}</math> in [1]. The normalized frequency offset is then 2.0*output0/fft_len.  
; Output 1: Beginning of the first OFDM symbol after the first (doubled) OFDM symbol. The beginning is marked with a 1 (it's 0 everywhere else).
; Output 1: Beginning of the first OFDM symbol after the first (doubled) OFDM symbol. The beginning is marked with a 1 (it's 0 everywhere else).


The evaluation of the coarse frequency offset is done in this block. Also, the initial equalizer taps are not calculated here.
The evaluation of the coarse frequency offset is '''''not''''' done in this block.  
Also, the initial equalizer taps are not calculated here.


Note that we use a different normalization factor in the timing metric than the authors do in their original work[1]. If the timing metric (8) is \f[ M(d) = \frac{|P(d)|^2}{(R(d))^2}, \f] we calculate the normalization as \f[ R(d) = \frac{1}{2} \sum_{k=0}^{N-1} |r_{k+d}|^2, \f]
Note that we use a different normalization factor in the timing metric than the authors do in their original work[1]. If the timing metric (8) is <math>M(d) = \frac{|P(d)|^2}{(R(d))^2}</math>, we calculate the normalization as <math>R(d) = \frac{1}{2} \sum_{k=0}^{N-1} |r_{k+d}|^2</math> (N=fft_len),  
If the timing metric (8) is  we calculate the normalization as  i.e., we estimate the energy from half-symbols. This avoids spurious detects at the end of a burst, when the energy level suddenly drops.
 
meaning that we estimate the energy from '''''both''''' half-symbols. This avoids spurious detects at the end of a burst, when the energy level suddenly drops.


[1] Schmidl, T.M. and Cox, D.C., "Robust frequency and timing synchronization for OFDM", Communications, IEEE Transactions on, 1997.
[1] Schmidl, T.M. and Cox, D.C., "Robust frequency and timing synchronization for OFDM", Communications, IEEE Transactions on, 1997.

Latest revision as of 17:06, 27 February 2023

Schmidl & Cox synchronisation for OFDM.

Input
complex samples.
Output 0
Fine frequency offset, scaled by the OFDM symbol duration. This is in [1]. The normalized frequency offset is then 2.0*output0/fft_len.
Output 1
Beginning of the first OFDM symbol after the first (doubled) OFDM symbol. The beginning is marked with a 1 (it's 0 everywhere else).

The evaluation of the coarse frequency offset is not done in this block. Also, the initial equalizer taps are not calculated here.

Note that we use a different normalization factor in the timing metric than the authors do in their original work[1]. If the timing metric (8) is , we calculate the normalization as (N=fft_len),

meaning that we estimate the energy from both half-symbols. This avoids spurious detects at the end of a burst, when the energy level suddenly drops.

[1] Schmidl, T.M. and Cox, D.C., "Robust frequency and timing synchronization for OFDM", Communications, IEEE Transactions on, 1997.

Parameters

(R): Run-time adjustable

FFT Length
FFT Length
Cyclic Prefix Length
Length of the guard interval (cyclic prefix) in samples
Preamble Carriers
If true, the carriers in the sync preamble are occupied such that the even carriers are used (0, 2, 4, ...). If you use all carriers, that would include the DC carrier, so be careful.
Threshold (R)
detection threshold. Default is 0.9.

Example Flowgraph

This flowgraph can be found at [1].

Rx ofdm fg.png

Source Files

C++ files
[2]
Header files
[3]
Public header files
[4]
Block definition
[5]