On Tue, 14 Aug 2018 10:09:14 -0700 (PDT),
=?UTF-8?Q?Mehtap_=C3=B6zkan?= <kurtulmehtap@gmail.com> wrote:

>Dear All,
> In page 8 of http://www.cs.tut.fi/kurssit/TLT-5806/Synch.pdf it is stated:
>All-digital symbol timing recovery scheme requires higher-than-symbol-rate sampling.
> If this statement is correct what should be the minimum ratio? We have also a constraint on the max sampling rate of the ADC. It is 2.2x symbol rate.

Using a polyphase filter as the matched filter, I've run systems to
very low oversampling rates with reasonable succes, e.g., around
1.2x-1.3x.    There may be some performance degradation due to
aliasing and other artifacts, but it can work.   

> If we sample at 2.2x symbol rate should we also adjust the Matched filter to this rate?

The matched filter has to be matched to the symbols, one way or other.
Not sure where your 2.2x number comes from, but it's not unreasonable.
It's also not absolute.   Depending on the pulse shape you may be able
to sample less than that with no issues.

> Are there any good references for symbol timing recovery with just a little bit hiher than symbol rate sampling? 

Not that I know of, as it is a little tricky and would be expected to
have problems.   As mentioned in the pdf you linked, one issue with
symbol-rate sampling is aliasing, since the signal will have
significant energy outside the bandwidth supported by that sample
rate.

>Thank you very much in advance.

Dear All,
 In page 8 of http://www.cs.tut.fi/kurssit/TLT-5806/Synch.pdf it is stated:
All-digital symbol timing recovery scheme requires higher-than-symbol-rate sampling.
 If this statement is correct what should be the minimum ratio? We have also a constraint on the max sampling rate of the ADC. It is 2.2x symbol rate.
 If we sample at 2.2x symbol rate should we also adjust the Matched filter to this rate?
 Are there any good references for symbol timing recovery with just a little bit hiher than symbol rate sampling? 
Thank you very much in advance.