SatSig topic:

Weather Outages

(Read 14632 times)

Jul 10^th, 2007 at 10:00pm

Considering weather outages, does antone know the availibilty of the Ka-Band Wildblue service in New York? Can I excpect service to be availible 95%,98%,99% of the year or other?

Jul 11^th, 2007 at 10:28am

Some figures are shown below for a similar 29.9 deg elevation angle and climate. Attenuation is due to water vapour, cloud and light rain. Percentage of 3 years of time.

% 20 GHz 30 GHz
99% 2.2 dB 2.8 dB
98% 1.8 dB 2.3 dB
95% 1.5 dB 1.9 dB

The Wildblue system is designed to cope with far worse attenuation than this, so you can expect no problems at all, at the percentages shown.

If you want details in the region between 99.99% and 99.5% you need to ask Wildblue as they have uplink power control and, I believe, reactive modulation and coding methods to minimise outages in heavy and very heavy rain. Note that it may take significant resources to answer the question and the accuracy of the result will still be uncertain - predicting the percentages of time for different rainfall rates is a problem with the weather nowadays.

I found this image which may be of interest or it may just confuse. The single value of 25mm/hr rainfall rate in the title confuses me but at least the image looks right. Also, I think the graphs refer to rain only and ignore the steady loss of about 1.5 dB due to light cloud, moisture and water vapour. Shift the graphs 1.5 dB to the right to allow for this.

Your elevation angle at 30 deg is higher than 10 deg and higher than 20 deg so your slant path length through the atmosphere and therefore your attenuation are less than in the image.

If you search Google for words like satellite attenuation slant path Ka band etc you will find much more to read.

There are recognised CCIR formulae for working out attenuation for various frequencies, elevation angles, site height, latitude and CCIR climate type (A - Q). I have considered writing a page that will do these calculations but it would be about two weeks solid work to do this, for me at least at my rate of progress! Are there any university students out there interested in writing a page in javascript for this site.?

Best regards, Eric.

Jul 11^th, 2007 at 12:42pm

Just to clarify a technical point, Wildblue customer hardware employs adaptive uplink power control. The gateway terminal provides signal quality feedback over the customer connection. Based upon this input, the customer modem varies the X-band output amplitude in 0.1db increments. This in turn regulates RF output power at the transmitter up to a maximum of 3 watts.

Short version; customer transmitters automatically increase power to compensate for rain attenuation. It doesn't eliminate weather related outages, but it does reduce the impact. There's still ice and wet snow to contend with in your area.

Take network availability times with a grain of salt however. The figures Eric gave you are much more realistic. Provider figures are based upon year round 24/7 operation, not upon how many hours a day you may actually use the connection. Reason not to take them at face value is the word NETWORK. There are too many variables to accurately predict individual CONNECTION availability. Wildblue has two other fudge factors that can inflate their network availability as well; multiple satellites and multiple gateway terminals. So technically every one of their gateways would have to be down on both satellites before they'd log even one second of NETWORK downtime. To fudge the figures even more in their favor, many providers do NOT consider scheduled downtimes in their availability figures either.

Of primary concern to you will be the reliability factor of the gateway terminal to which you're assigned (quite probably Syracuse), combined with your personal effectiveness in removing wet snow/ice from your dish. Short version; feel very lucky if you average 99.0% percent reliability, expect less.

//greg//

Jul 11^th, 2007 at 6:49pm

Thanks Eric and Greg,
I would imagine the WildBlue would have to maintain availibilty far greater than 99%. After all 99% availibilty works out to 87.6 hours of no service (365x0.01x24hrs.). This seems like an extremely long period. Even at 99.8% availibilty, uplink rain fade alone is on the order of 13dB. At this % availibity, service is out for 17.5 hrs. Even with an uplink power control of maybe 5dB (1W going to 3W) its hard to imagine that this would be enough to compensate for the 13dB of uplink fade. Does Wildblue use adaptive modulation and coding (i.e DVB-S2) on the customers return uplink. Does the modulation type and FEC change to something more tolerant during rain fades?

Jul 11^th, 2007 at 9:15pm

I guess we are looking at this from different angles. 87.6 hours a year works out to 14.4 minutes per day. You might not experience 14.4 minutes EVERY day, but I don't believe it's unusual for Wildblue customers to accumulate 5000+ minutes of outage in a year's time.

I'm talking percent of connection availability. Not just weather-related, I'm considering ALL outage categories. When the connection is not available for the purpose you pay for, it's an outage.

As I stated above, Wildblue employs adaptive uplink POWER to counter rain attenuation. It's HughesNet that uses adaptive (error correction) coding.

//greg//

Jul 13^th, 2007 at 4:00pm

I'm in southern Ontario so I'd imagine we have similar availability. I've been using Xplornet's Ka band service for not quite eight weeks now. As a purely subjective guess I'd say we lose about 1 day a month, total.

Not a full day outage -- but whenever there's a significant rain storm or thunderstorm we'll lose service for a few hours. And this is the stormy season here. In fairness, I must admit that we shut everything off when a thunderstorm arrives.

I've been told by a level 2 tech that we can expect to lose service whenever there is a storm here or at the gateway. So whatever adaptive measures they're using, they can't handle a heavy storm.

Jul 16^th, 2007 at 3:04pm

It's interesting to note that WB has filed with the FCC for numerous 0.38m terminals with EIRPs using various data rate (160Ksps, 320Ksps, 640Ksps under rain fade conditions) and 1280Ksps under clear sky. Sounds to me like dynamic variable data rates to mitigate rain fade.

Jul 16^th, 2007 at 5:46pm

No, those numbers represent symbol rate - not data rate. And I think you're looking at a superceded application anyway, because they're currently only selling 3 upload speeds; 128k, 200k, 256k. Or else they have an unadvertised 512k upload rate. Did you check the actual filing date?

Either way, that document you cite is almost certainly related to four different subscription plans - where the more you pay, the faster your uplink. For example; Value Pak, Select Pak, Pro Pak, Enterprise Service.

//greg//

Jul 16^th, 2007 at 7:06pm

Greg,
Filing where in April 2007. The uplink is QSPK, and I assumed that a FEC of 1/2 would mean a more fade tolerant Eb/No requirement (lower Eb/No given a fix BER). For QPSK R1/2, the data rate would equate to the symbol rate. If am wrong about the FEC than you are correct, those number represent the symbol rate. The filing "seeks to modify its existing blanket earth station authorization to (1) add a new antenna model for use with enterprise customers, (2) increase the number of antennas authorized and (3) increase EIRP and EIRP density". Has there been a more recent filing? The filing includes a 0.38m, 0.7m and a 0.98m antenna. Upon closer inspection, all three antenna models have the faded and clear-sky symbol rate as previously listed. Do you know which is the new antenna model?

Jul 16^th, 2007 at 7:50pm

The "sps" stands for symbols per second. No way to accurately extrapolate data rate without knowing the specific FEC rate. Best guess is the associated uplink rates associated with those numbers you provided are 128k, 256k, 512k, 1024k (although I'd be danged surprised to ever see 1024 via 66cm @ 3 watts). Might actually be 64k, 128k, 256k, 512k

Not sure where you got the 38cm figure, that would represent less than a 15" parabola. Their mainstream dish is a 66cm equivalent (about a 26" equivalent).

The current consumer dish is the Raven Corvus. Visually it's the same at the earlier version G66, tech specs have undergone a recent but minor tweak.

I don't have any data on the larger dishes, they're typically exclusive to unusually high precipitation areas and to enterprise solutions

//greg//