Hello!

Channel Equalizers tries to restore distorted received signals back into the original message through filtering based on the estimated channel parameters. But this is an estimate. If I increase the number of observations, then it gets a better estimate. Is there anyway to improve the accuracy further?

I am debating with myself if I can decrease the processing and receiver delay it can improve the accuracy since there are less uncertainties for error. I was also thinking if the length of the equalizer is as short as possible then the filter won't be able to filter what it shouldn't be filtering (the original message).

I want to make sure my ideas are correct.

Thank you very much in advance!

 

Hello there, welcome to AAC!

want to make sure my ideas are correct

Could you please elaborate on your idea.
With all humility I have no idea what what you are talking about. Respectfully.

 

Hello there, welcome to AAC!

Thanks for the welcome!

Could you please elaborate on your idea.
With all humility I have no idea what what you are talking about. Respectfully.

Sorry I am new at this and English is not my first language so I probably didn't explain it very well ; ; I was trying to understand this but it got too technical for me. Basically, what I picked up is:

When signals are transmitted, they are distorted by the physical medium (channel) in which they are sent through. This distortion is mainly ISI. In order to decode the signals after they are received, I need to know the channel characteristics so distortion from here can be filtered out of the original signal. But if I don't know the channel characteristics, then I can use the channel estimator to identify by computing/estimating its model parameters. For example, I can try to generate known signals, transmit it to the channel with unknown parameters, and look at the received signals and compare (Error Rate Calculation). From here I can estimate channel parameters. I can mimic channel behavior by using a discrete FIR filter with coefficients proportional to channel length (coefficients can be [0.1 0.2 0.3] for a channel with length of 3). Going through the process introduces processing and receiving delays. If I increase the number of observations, then the estimation gets more accurate. I was wondering how to further increase the accuracy. From what I said above, I was thinking if minimizing the delays as much as possible increases accuracy of the estimation. This can be done for example by using minimum phase IIR filters instead of linear phase FIR filters since FIR filters have higher latency. Also, if I use a channel with the least possible length, I can also improve the accuracy since it reduces the overall processing delay. I added screenshots below from the link above to see if my understanding is correct.

Thanks for the reply and clarification!
--

--

 

Hello there again! My apologies for answering a question with a question.
Ok.Do you wish to know the the difference between FIR (Finite Impulse Response) and IIR (Infinite Impulse Response) filtering also known as "Biquad" filters,used in DSP (digital signal processing)?
Because this does not sound like homework to me. And right now I am very confused as to what forum should address your question.

 

Hello there again! My apologies for answering a question with a question.
Ok.Do you wish to know the the difference between FIR (Finite Impulse Response) and IIR (Infinite Impulse Response) filtering also known as "Biquad" filters,used in DSP (digital signal processing)?
Because this does not sound like homework to me. And right now I am very confused as to what forum should address your question.

Hello!

No, I just want to make sure I understood the lesson completely. I just want to ask and confirm if everything I said is right because I usually make a lot of mistakes in understanding topics that are new to me. My homework is to describe the relationship of the accuracy of estimation and the processing delay, number of observation, and length of equalizer. I want to confirm if the estimation really does become more accurate if I decrease delay by choosing different kinds of filter, and if I minimize the length. Sorry for the confusion.

Thanks for the reply

 

Possibly asking:
Is FIR EQ latency an issue for accuracy compared to a moving average in digital audio ?

Yes, it is accurate depending on how you define accurate when solving this with mathamagic.
For a pure single frequency at first look it is accurate not sure that an audio composition with harmonics and distortion effects would be accurate to a composer or accurate on low res equipment but future equipment don't hold your breath.
See FFT of a FIR at fast forward time and see the beginning about what the experiment is doing.

 

I just want to ask and confirm if everything I said is right because I usually make a lot of mistakes in understanding topics that are new to me.

I can confirm.You did not make a lot of mistakes. There is a distinct advantage for each soft computing algorithms that are used to estimate the filter weights for their adaptive channel equalizer. PSO, GA, WDO and BFO. For FIR, WDO is more robust than the others by comparing all to the benchmark soft computing algorithm LMS. By using coefficients proportional to channel length you reduce bit error rate and distortion, minimizing latency or processing delays.For reliable data transmission in communication systems, the goal is to estimate the weights of equalizers installed at the front end of the receiver, as quickly
as possible instead of good average potential solution that is ... using minimum phase IIR filters. Future
services demand high data rate and quality. Thus, it is necessary
to define new and robust algorithms to estimate equalizer weights, so as to reduce the effect of noise in the communicated data.Population based soft computing algorithms such as these are envisioned to receive increasing attention due to its
reliability and accuracy.

 

Does a poor frequency response have distortion?
Equalization is used to make a low distortion audio signal have a flat frequency response if something wrongly cut or boosted bass, mid-range or treble frequencies. Some people like to wrongly boost or cut some audio frequencies.

Equalization can also cut mains hum (then it also cuts adult male voices and bass sounds) and equalization can cut hiss (and it also cuts consonant sounds in speech and harmonics of music).
I guess cutting the high frequencies with an equalizer also cuts some distortion, but audio should not have any distortion except acid metal rock (music?) has lots of distortion (overdrive and fuzz).