On Thu, May 10, 2012 at 9:15 PM, Bruce Perens <br...@perens.com> wrote: > David, Mel, and I have been having a discussion of PAPR: peak vs. average > power ratio, off of the list. This is also called "crest factor". [snip] > Should we compress in software before feeding the amplifier, so that > clipping is more graceful than the hardware might make it?
Generally the PAPR of an uncorreced N-carrier systems is roughly N, 13.5 sounds a bit high— but one of the FDMDV carriers is hotter. There are a lot of crazy ideas people have thrown at this problem, including pre-distorting the signals and enumerating the set of all possible concurrent-multicarrier symbols and puncturing the modems codebook to eliminate the ones with the worst PAPR (but alas, a 2^28 entry codebook would be a little insane, unless it turned out that the bad symbols had some nice structure). Another hack I've seen is adding some dummy carriers which the modem would pick values for which reduced the PAPR (yuck). It should be noted that the OFDM modem with N-PSK carriers modulated with a golay complementary sequence (a rate 1/2 FEC) provides a PAPR of 3dB, which I think is optimal. (And, moreover, errors which change the energy of the signal are exactly the ones best corrected by the FEC— so I'm not even sure if the loss reduction in codeword separation from not using QAM carriers even hurts). I'd certainly suggest looking into that for power and/or linearity limited channels. Even when not peak limited, it generally seems to me that modulation schemes with more constant energy should perform better when operating near the noise floor for a given total output energy in any case: The energy spent in parts of the signal that are (say) 13dB over the noise would be better spent lifting the parts of the signal at 0dB over the noise floor above it. ------------------------------------------------------------------------------ Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ _______________________________________________ Freetel-codec2 mailing list Freetel-codec2@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/freetel-codec2
