On 2/8/2011 12:15 PM, Llew Roberts wrote: > After pulling the old reference materials out, I was hoping to be able to > provide some quantitative data here, but I can't find even one reference > that talks definitively about potential throughput using varying > bandwidths/signal separations on HF frequencies. It's a fun theoretical > question but putting together a system to test any potential answers would > be expensive to build and illegal to use. > > As a side note, I've learned one interesting thing that may seem obvious > when you take a moment to think about it: one of the main factors that > limits data transfer rate over long distances using HF frequencies is > propagation delay.
It doesn't seem obvious to me, largely because it isn't true. The propagation delay of an HF radio channel is only marginally higher (and in some cases lower) than an equivalent fiber path. The limitation in data transfer rate is simply that there's limited bandwidth (and a fairly high noise floor). Even if you use all of DC-30 MHz, that's only 30 MHz of occupied bandwidth. That, and the noise, can be plugged right into Shannon's big contribution to this field to give you the answer. Propagation delay has no effect whatsoever on throughput... only when you get the answer. > Propagation delay increases when the physical distance > between nodes increases and imposes a hard limit on what your potential > transfer rate can be. Not true. It does impose a limit on how fast an unmodified TCP stack's AIMD grows, because it contributes to the delay*bandwidth product. > The longer the distance, the longer the propagation > delay, the smaller your potential transfer rate can be. Nope. > Because of that (and the fact that radio doesn't provide a clean enough > signal to work with at HF frequencies) I don't think it's possible to "go as > fast as you want" using HF. You can't "go as fast as you want" but only because of the 30 MHz of bandwidth and the noise floor. > The total available signal bandwidth below 30MHz is... well... 30MHz. That > includes everyone's frequencies, not just amateur radio frequencies. When > you consider that the bandwidth of just one 802.11 wifi channel is 22Mhz and > there are 11 of them (in most countries), you can see that it takes a > tremendous amount of bandwidth even when you have an excellent signal to > transfer a significant amount of data at high speed. There just isn't > enough bandwidth available in the HF bands to do that. Aha. Now you're getting somewhere. > > Here's a good summary of digital modes that are available on HF. According > to this and most of the other resources I've looked at today, 300 baud is > about the best you can expect reliably at HF frequencies over long distances > although 1200 can be workable. > > http://www.qsl.net/yo5ofh/doc/digital%20hf%20radio%20operating%20modes.htm There's lots of OFDM (and other) modes that can go much faster than this, they're just not in common use by folks who have legal limits on occupied bandwidth. Matthew Kaufman _______________________________________________ p2p-hackers mailing list p2p-hackers@lists.zooko.com http://lists.zooko.com/mailman/listinfo/p2p-hackers
