How can I perform root-raised cosine filtering on very large data sets using the RCOSFLT function without memory issues using the Communications Toolbox 3.4 (R2006b)?

I am using the function RCOSFLT to filter out a very large data set. The set is so large that I face memory problems when I try to use the function. I want to know if there is a way that I can break up the data set and filter each set and then remember the state of the filter for the next data set to maintain continuity.

Answer 1

[MathWorks Support Team]

This is currently not possible for the RCOSFLT function in Communications Toolbox 3.4 (R2006b).

To work around this issue, you can design the root raised cosine filter using function in the Communications Toolbox and then use the FILTER function to filter each of the many data sets and return the filter state each time. The FILTER function is able to return its states.

Answer 2

[Tasos Giannoulis]

You can also use the corresponding System objects: comm.RaisedCosineTransmitFilter, comm.RaisedCosineReceiveFilter. System objects inherently retain state, so you can break down your large input and pass it to the System object's step() method bit by bit. State will retained between successive step() calls.