Hi Phil, nothing at all. I'm still in debt with you since I looked into your Reed-Solomon Berleykamp-Massey decoder C source codes twenty years ago or so which, along with Peterson's Error Correcting Codes book, gave me the opportunity to learn something more than I was teached at university.
Sequences are an art, I agree. We spend a lot of time designing new codes but rather few people cope with details which are as important as coding. It's always frustrating trying to solve problems of exponential complexity. I love to think that we are still at the beginning of the third millennium and that some smart guy will solve the issue once for all within some few decades. I'll give a look to Levanon's book. For what concerns sequences personally I'm still stuck to an old ARRL handbook about Barker codes or similar things. Meanwhile computers speed has increased so much that it's not so difficult to find an adequate solution simply by a random search that optimizes some specific target cost function. We know that in this way we will never find the really optimal solution but if what we found is already very close to what can be expected from a theoretical point of view... well, even if it is not so elegant to say, who cares... Maybe if we only had some cheap quantum computers and suitable algorithms for them we would accept the idea that solutions arising from chaos are as elegant as those coming from finite fields algebra :-) Thanks for the book suggestion indeed. Nico / IV3NWV ___________________________________ Il giorno dom 16 mag 2021 alle ore 23:36 Phil Karn <k...@ka9q.net> ha scritto: > On 5/16/21 09:15, Nico Palermo wrote: > > Hi Phil, > > > > I don't know what Joe used in WSPR but, for instance, he told me that > > he generated the sync sequence used in Q65 by means of direct > optimization > > of its autocorrelation function, not in an algebraic way, so that its > > lateral peaks > > were minimum only in a limited time shift interval. > > I guess that he did something similar also with WSPR. > > Thanks Nico. That would actually make sense. Designing sync sequences > seems to be a real art form because it depends on just what you're > trying to optimize; you can't have it all at once. I have an entire book > on the subject. The title is "Radar Principles" by Levanon, but the more > you learn about radar the more you realize that radar is essentially the > correlation/synchronization problem. > > I asked because the various writeups mention a "pseudo random > synchronization word having good auto-correlation properties" and this > usually implies a simple linear feedback shift register or a combination > of them (e.g., Gold codes). On the other hand, 162 is not a power of 2 > minus 1 so it would have to be truncated from a longer sequence. > > Phil > > > > > > _______________________________________________ > wsjt-devel mailing list > wsjt-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/wsjt-devel >
_______________________________________________ wsjt-devel mailing list wsjt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wsjt-devel
