Preparing timestamped data for fourier analysis

Greetings all,

I'm working on a project trying to apply fourier analysis to timestamped 
router logs, using R to perform the analysis.  The idea is to determine if 
any type of traffic (say, outgoing ICMP requests) has strong periodic 
features because it may indicate a compromise somewhere on the network.

The FFT requires all data points to be evenly spaced, but the recorded 
events do not occur at a consistent constant dt, so I need to "zero pad" 
the data.  I can do this for small data sets (say, one day's worth of 
traffic) in excel.

However, I am now attempting it on a large scale using a 25-day router 
log, which should have 2,160,000 records.  The log actually contains only 
56,725 records.

So, I'm looking for ways in which to pad the data in R.  If anyone could 
just point me at the right man pages to read that would be itself a great 
help.

Thanks in advance,

Pete


---------------------------------


[[alternative HTML version deleted]]

I believe that FFT is not appropriate. However Lomb-Scargle periodogram
could be used.