search

sdrforengineers / bookerrata

This is the issues and errata for the "Software Defined Radio for Engineers" book
4 stars 0 forks source link

Code 6.1 Pg 198: Clarification regarding decimation #12

Open solomonbstoner opened 3 years ago

solomonbstoner commented 3 years ago

In Code 6.1, there is a decimation of 4 in the receiver matched filter. Why is that so? Changing it to 1, I get the following animation, which looks the same to me as using decimation of 4 (except it slower). What is the purpose of the decimation?

Also, why does decimation=1 result in a slower change than decimation=4? I was expecting it to be the same since tau is a fraction of a symbol period

code6 1_decimation_1 Decimation = 1 is above ^ . Decimation = 4 is below v . Freq =2400000000Hz, SampleRate =1000000Hz 2021-09-24 22-44-56

tfcollins commented 3 years ago

The Tx RRC filter always interpolates by SamplesPerSymbol but the RRC only decimates by the supplied Decimation factor. The remaining symbols (SamplesPerSymbol - Decimation) can be used to better estimate the timing error. This is likely slower since you are processing 4x the data when setting decimation to 1.

solomonbstoner commented 3 years ago

Ah that makes sense. But just to clarify, should it be SamplesPerSymbol - Decimation or SamplesPerSymbol/Decimation?