|
From: | George Nychis |
Subject: | Re: [Discuss-gnuradio] FIR filters and set_history() |
Date: | Wed, 06 Feb 2008 13:01:36 -0500 |
User-agent: | Thunderbird 2.0.0.6 (X11/20071022) |
George Nychis wrote:
Using the FIR with time reversed complex conjugate of the desired input signal:http://cyprus.cmcl.cs.cmu.edu/matched_filter/graphs/mf_fir_conjugate.jpgUsing the "matched filter block" that uses the complex/real-wise inner product:http://cyprus.cmcl.cs.cmu.edu/matched_filter/graphs/mf_block.jpgThe FIR filter method seems to have much better properties, creating a much larger magnitude in the output.
This is probably a much better comparison: http://cyprus.cmcl.cs.cmu.edu/matched_filter/graphs/mf_compare.jpgThe difference between what I would consider the maximum noise level of the inner product method and the correlation peak is: 160-45=115
For the FIR filter, 225-75=150. Just approximations. - George
[Prev in Thread] | Current Thread | [Next in Thread] |