ICI avoidance in OFDM

Started by Mannai_Murali 5 years ago8 replieslatest reply 5 years ago180 views

In standards like DVB-T2 there is a iterative decoder to mitigate ICI.Since the effect of high doppler is to introduce ICI due to frequency offset why we can NOT avoid ICI before it occurs by using a PLL and tracking frequency offset.

Anyhow OFDM receivers has frequency offset estimation in post FFT and correction before FFT can't we get away with ICI even in high doppler environment by PLL.

Thanks for the reply in advance.

[ - ]
Reply by shafie7July 5, 2019

It sounds like you are assuming every single carrier in the OFDM signal is shifted by fixed amount and that is known to the receiver.  Is that correct understanding of your question?

If no, please elaborate.

The Doppler frequency shift occurs over the air/channel and could be time variant.  The Doppler frequency shift is dependent on relative speed of transmitter and receiver, their relative position, and is dynamically changing.  The receiver does not know the amount of the frequency shift in advance.

The following Fading Manifestations and Mitigation Techniques blog may help.


Best regards,

Shahram Shafie


[ - ]
Reply by SlartibartfastJuly 5, 2019

The Forward Error Correction is not just for ICI, but for all impairments which includes noise, synchronization errors, unmitigated multipath effects, interference, etc., etc.   ICI is just one part of that, and is probably not the driving impairment most of the time.

Synchronization techniques, including frequency estimation and tracking, are implementation dependent so some may be much better than others.  In other words, if ICI is a problem, employing better methods for offset estimation and tracking may be doable. 

Usually frequency offset can be removed enough in an OFDM system that ICI is less of an impairment than noise and multipath, especially in a system like DVB-T2 where there plenty of pilot tones to help out.

[ - ]
Reply by Mannai_MuraliJuly 5, 2019

Is it because with multi path each path will have different frequency shift due to different arrival angle and hence different velocity.Frequency offset estimation with post FFT estimation will only work for single time varying shift and not multiple shifts for each path which vary slowly with time.


[ - ]
Reply by SlartibartfastJuly 5, 2019

What is the mechanism by which one subcarrier will have a significantly different frequency offset than another?  All of the subcarriers go through the same multipath profiles.

[ - ]
Reply by Mannai_MuraliJuly 5, 2019

Since Doppler in Hz=V*Cos(Theeta)/Lambda

v-->velocity;Lambda-->Wave length;Theeta -->Angle for different paths

Each resolvable path having different (theeta) angle will have different Doppler.So each sub carrier will be subjected to more than one frequency shift Deltaf1,Deltaf2,...DeltafN.For N resolvable paths.So 'i'th sub carrier will shift by fi+Deltaf1,...fi+DeltafN.

The Frequency offset estimation post FFT expects to single shift rather than 'N' shifts.Here all carrier undergo by 'N' different shifts as


[ - ]
Reply by SlartibartfastJuly 5, 2019

How is theta different for each subcarrier?  All of the subcarriers come through each propagation path, so all experience the same thetas and the same dopplers, no?

Lambda is very slightly different for each subcarrier, but the widest DVB-T2 channel is only 10MHz, so I doubt that would make a significant difference compared to the subcarrier spacing.   What is the mininum subcarrier spacing being used in DVB-T2?

[ - ]
Reply by Mannai_MuraliJuly 5, 2019

Suppose there are two paths.Path1 will have theeta1 as angle of arrival.path2 will have theeta2 as angle.

The doppler for path1 will be deltaf1.For path2 it is deltaf2.

Subcarrier i will shift by fi+deltaf1 due to path1 and fi+deltaf2 due to pathf2.

subcarrier fj will shift by fj+deltaf1 due to pathf1 and 

fj+deltaf2 due to path2.

This is path dependent not dependent on sub carrier frequency since this is narrow band channel unlike wideband channel like under water communication.

Thanks for your reply

[ - ]
Reply by thirugvasJuly 5, 2019

The Doppler spread is instantaneous and it cannot be predicted as it depends on the velocity of the mobile. Using adaptive equalizers the frequency offset can be avoided. It is better and less complex to cancel the offset post FFT rather than doing this pre FFT.