[time-nuts] PN sequence generation using GPS

jimlux jimlux at earthlink.net
Wed Feb 23 15:05:49 UTC 2011


On 2/23/11 4:48 AM, Bob Camp wrote:
> Hi
>
> Hop rates below 1,000 per second are far more common in simple systems than anything faster than that. At VHF, you are looking at a everybody being within one hop of each other. That makes the idea of a GPS based code start fairly reasonable.
>
> Bob
>

yes.. back in the day, people had proposed various "time of day" sync 
strategies, but they all seemed require more complexity and mass than it 
was worth.  (you could add a manpack sized GPS receiver to a manpack 
sized radio) There were a lot of interesting schemes proposed for things 
like clock syntonization so the hops/chips stay in phase.  And, given 
the usual application of SS links in the 70s and 80s (Skywalker sound 
excepted) there was a lot of thought to the jammability and/or 
jam-resistance of such schemes.  Most SS systems have a generalized 
vulnerability during acquisition, since you don't have any process gain 
until you've acquired sync.  (parallel sync schemes help a huge amount, 
because you might be able to reject "false locks", while a serial sync 
scheme might decide to lock on the wrong thing if it hits it first in 
the search... sort of a global vs local minima thing)


These days, though, with ubiquitous GPS and very tiny, low power GPS 
receivers, it would be very attractive.  Getting 1 microsecond sync 
would be trivial (after computing the nav solution).


And, if you want to be clever, if you've already implemented hardware 
correlators and code space searchers, you could use GPS to get an 
initial approximate phase for a fast Direct Sequence system, and then 
use the same hardware to acquire the signal of interest.

By the way, the usual terms of art for frequency hoppers are "slow" and 
"fast" hoppers, differentiated by whether the hop rate is slower or 
faster than the symbol rate.  For an example of exactly in the middle, 
there were (are?) a raft of schemes for MFSK where the constellation was 
much larger than needed for the number of bits per symbol (say you're 
encoding 4 bits per symbol, and you have 2000 possible frequencies).





More information about the Time-nuts_lists.febo.com mailing list