SERVING THE QUANTITATIVE FINANCE COMMUNITY

 
User avatar
Traden4Alpha
Posts: 23951
Joined: September 20th, 2002, 8:30 pm

FPGA library

February 13th, 2012, 3:20 pm

QuoteOriginally posted by: farmerQuoteOriginally posted by: Traden4Alphaeach successful communication yields very limited expected profits defined by the instantaneous liquidity and the price differential over the system's latency advantage.In forex and spu's, the instantaneous liquidity is more than a few dollars, no?Sure, but it's the price differential over the latency advantage multiplied by the local market depth and the probability that the received price signal represents a true move versus a price burble. If I get a ultra-speed communication that the New York price twitched by a couple pips, I can hit the prevailing bid or ask to capture it (and hope it was a sustained price change) but I can't expect to put on a huge order. And if the New York price just reverts, then I eat a round-trip transaction cost or wait (and hope) for a successfully-received communication of a twitch in the other direction.The point is that profits from any one successful communication won't be that large and the system's profit will be proportional to the transmission reliability.
 
User avatar
farmer
Topic Author
Posts: 13461
Joined: December 16th, 2002, 7:09 am

FPGA library

February 13th, 2012, 4:22 pm

QuoteOriginally posted by: Traden4AlphaThe point is that profits from any one successful communication won't be that large and the system's profit will be proportional to the transmission reliability.1 hop = tower_cost/40
 
User avatar
farmer
Topic Author
Posts: 13461
Joined: December 16th, 2002, 7:09 am

FPGA library

February 13th, 2012, 5:08 pm

Your cost rises linear with your number of towers, and your number of successful signals does not, while your latency goes up. So is not the sweet spot for profitability possibly one hop? I could see that your number of successful signals would more than double for adding a second hop, and a third hop. But I will take a wild guess you could get a very high transmission rate with 5 hops of 150 miles. So basically you need six towers with three frequencies each. For a total of 18 frequencies.land leasesolar panelsantennadeep cycle batteriescomputer stuffcamouflage to make your tower look like a moonshine still?$12,000 per antenna for the first year?
 
User avatar
farmer
Topic Author
Posts: 13461
Joined: December 16th, 2002, 7:09 am

FPGA library

February 13th, 2012, 6:46 pm

What technology and frequencies would you recommend for the regulated and gray sections?
 
User avatar
Traden4Alpha
Posts: 23951
Joined: September 20th, 2002, 8:30 pm

FPGA library

February 13th, 2012, 9:15 pm

QuoteOriginally posted by: farmerYour cost rises linear with your number of towers, and your number of successful signals does not, while your latency goes up. So is not the sweet spot for profitability possibly one hop? I could see that your number of successful signals would more than double for adding a second hop, and a third hop.Signal success is not linear in the number of towers because signal attenuation is an exponential process and LOS blockages are 100% attenuators. With less than N hops (where N is a function of distance, geography, and tower height), you will get ZERO success due to line-of-sight blockages. But even if you have enough towers to have LOS between them all, you can still have long periods of no signal due to rain or snow. For long chains of towers, the probability of having rain or snow on at least one link can be very high.No doubt, there is some optimum number of towers in which the cost of an added tower isn't justified by a sufficient increase in signal reliability. But the minimum number of hops may be more than five.
 
User avatar
farmer
Topic Author
Posts: 13461
Joined: December 16th, 2002, 7:09 am

FPGA library

February 13th, 2012, 9:48 pm

QuoteOriginally posted by: Traden4AlphaLOS blockages are 100% attenuators.Where did you get this LOS thing? Don't be silly, you don't need line of sight.
 
User avatar
Traden4Alpha
Posts: 23951
Joined: September 20th, 2002, 8:30 pm

FPGA library

February 13th, 2012, 10:18 pm

QuoteOriginally posted by: farmerQuoteOriginally posted by: Traden4AlphaLOS blockages are 100% attenuators.Where did you get this LOS thing? Don't be silly, you don't need line of sight.It's a pretty basic fact of radio engineering but you can read more here.
 
User avatar
farmer
Topic Author
Posts: 13461
Joined: December 16th, 2002, 7:09 am

FPGA library

February 13th, 2012, 10:23 pm

QuoteOriginally posted by: Traden4AlphaIt's a pretty basic fact of radio engineering but you can read more here.From your link:QuoteAt low frequencies (below approximately 2 MHz or so) radio signals travel as ground waves, which follow the Earth's curvature due to diffraction with the layers of atmosphere. This enables AM radio signals in low-noise environments to be received well after the transmitting antenna has dropped below the horizon. Additionally, frequencies between approximately 1 and 30 MHz, can be reflected by the F1/F2 Layer, thus giving radio transmissions in this range a potentially global reach (see shortwave radio), again along multiply deflected straight lines. The effects of multiple diffraction or reflection lead to macroscopically "quasi-curved paths".My cellphone is not in either of those frequencies, and it works inside my house. And I certainly cannot see any antennae inside my house!
 
User avatar
Traden4Alpha
Posts: 23951
Joined: September 20th, 2002, 8:30 pm

FPGA library

February 13th, 2012, 11:18 pm

LOS is a loose term and some materials (e.g., dry wood, wall paper, cloth curtains, plaster, vinyl siding, thin foliage) are transparent at cellphone frequencies. Cell phones can also get a signal via various short-range effects of how radio waves bounce around or deflect in the urban environment.
 
User avatar
farmer
Topic Author
Posts: 13461
Joined: December 16th, 2002, 7:09 am

FPGA library

February 13th, 2012, 11:25 pm

What about 20 Mhz bounced off the atmosphere from High Point State Park, 59 miles NW of Broad Street?
 
User avatar
farmer
Topic Author
Posts: 13461
Joined: December 16th, 2002, 7:09 am

FPGA library

February 13th, 2012, 11:45 pm

Three signals per tower, in the neighborhood of 1.5 Mhz, 15 Mhz, and 28 Mhz, bounced off the atmosphere between...Broad Street <-> High Point State Park <-> Mount Jewett <-> Battle Creek <-> South Bend <-> South Wacker Drive
 
User avatar
farmer
Topic Author
Posts: 13461
Joined: December 16th, 2002, 7:09 am

FPGA library

February 14th, 2012, 12:00 am

Here it is... Broad Street to Paternoster Square in one hop!
 
User avatar
farmer
Topic Author
Posts: 13461
Joined: December 16th, 2002, 7:09 am

FPGA library

February 14th, 2012, 10:41 am

Or you might believe this:QuoteA few weeks ago I had a contact with Charlie VK3TCV [in Australia]. He had just completed a homebrew helically wound yagi antenna for 40M. I was his first contact. He had a huge signal. Later that night I heard him talking to Pedro NP4A I Peurto Rico. Pedro was complementing Charlie on his big signal.Could I not amplitude modulate that thing? If I do 20 cycles high, 20 cycles low, that will be at least 1 bit per 40 cycles. 40 cycles = 1 microsecond, makes for 1000 bits per millisecond.Now if I do an array at both ends, 5 homemade yagi antennas per frequency, spaced at 1000 feet, can I not add up the signals from the array to get the signal straight from Chicago in one hop? These antennas are shown at a cost of a few dollars plus labor.
 
User avatar
farmer
Topic Author
Posts: 13461
Joined: December 16th, 2002, 7:09 am

FPGA library

February 14th, 2012, 11:01 am

QuoteOriginally posted by: outrunWhat I've read you can't get beyond 30 mile in a single hop (don't know why).Maybe because you need two-way communication to subscribe to a timeslot on a tower with multipler users and time-division multiplexing? Beyond 30 miles the delay puts your round trip outside the timing window for synchronized communication. And without synchronized communication, you can't reroute?QuoteFor example, if a MS were exactly 75.6km away from the BTS, then it would take 252µs for the signal to travel from the BTS to the MS. The MS would then synchronize with this timing and send its access burst on the RACH. It would take 252µs for this signal to return to the BTS. The total round trip time would be 504µs. So, by the time the signal from the MS arrives at the BTS, it will be 504µs behind the timing of the BTS. 504µs equals about 136.5 bits. The 68.25 bits of guard time would absorb some of the delay of 136.5 bits, but the access burst would still cut into the next time slot a whopping 68.25 bits
Last edited by farmer on February 13th, 2012, 11:00 pm, edited 1 time in total.
ABOUT WILMOTT

PW by JB

Wilmott.com has been "Serving the Quantitative Finance Community" since 2001. Continued...


Twitter LinkedIn Instagram

JOBS BOARD

JOBS BOARD

Looking for a quant job, risk, algo trading,...? Browse jobs here...


GZIP: On