To convert to more normal units, that is about 189 Mbits/second. Which sounds a little less impressive considering the peak 4G speed is supposed to be 100 Mbit/second.
I thought 83 GB/hr seemed like not much capacity, and indeed that really isn't much. I've seen these units (and from other carriers as well, IIRC) before and figured they were bringing massive amounts of extra backhaul to spots with big events.
I'm curious how these figures relate with what 'normal' towers have allocated. I've seen the rough 1 gbps figure before but don't know how accurate that is.
1Gbps assumes the tower has fiber. Since this was a temporary event we can assume they don't have fiber available, that and the article illustration shows a microwave backhaul.
Standard microwave backhaul in a licensed band typically tops out around 263Mbps as that's what you get with 256QAM in 40Mhz of spectrum.
A lot of operators are switching to higher frequency bands for their backhauls, I see a lot of Ubiquity airfibers popping up here in India. Airfibers are 800mbps full duplex
AirFiber radios (I'm quite fond of mine) and other 24Ghz, 60Ghz, and 80Ghz radios have severe rain fade so you need to keep the distance down to one or two miles unless you feel lucky. Now find a cell tower in that distance with fiber and available mounting space. I suppose that isn't a big problem in a more densely populated area like southern California, but still more challenging than parking a 11Ghz dish on a rooftop somewhere within 14 mile or so.
From what I've heard most carrier interest in high frequency bands is for links going a few city blocks to connect micropops back to a area hub site. That kind of caution with fade margin is more in line of what I'd expect out of at&t or a similar provider.
83 GB/hr is not capacity that's actual usage. Considering the overheads involved I would not be surprised if this was 2-5 times faster in terms of raw bandwidth.
So light on details.. could have mistaken it for a press release. And more amazing, they managed to get some of that wrong. In their graphic, they mislabeled the mobile command post as H, and G in the key... and "AT&T’s network is about 80% iPhones" then link to an article about _smartphone_ sales in _Q4 of 2012_
I tend to avoid articles on any 3rd party sites with the word "Mac" or "Apple" in their domain names for this exact reason. The effects of the RDF are just too transparent.
I ran a mobile gaming conference for about 250 people in a major city back when AT&T was the only game in town for the iPhone (~2009). Without calling us or us reaching out to them, AT&T contacted the venue and installed a micro-cell--free of charge--to ensure good coverage throughout the day.
Say what you want about the network or the company, but I was impressed. (It probably didn't hurt that we had some folks from Apple on the attendee list.)
I'm afraid I don't. We never actually found out from AT&T how this happened or who was responsible--someone got in touch with the venue manager and the next thing we knew, a micro-cell showed up.
I'm told one can rent them (it was a small rack of equipment), but I'm guessing it's rather expensive.
This is pretty surprising that AT&T engineers were the ones to design the antennas.
It is usually vendors such as Nokia-Siemens, Alcaltel-Lucent or Ericsson that provide the equipment and that have the antenna/radio know-how while on the operator side the engineers there focus more on optimization and RF planning, rather then the design.
If you watch the video of the two AT&T engineers, they basically say that they just had the idea; their (unnamed) vendor actually did the work to see if it was feasible and designed the antenna.
When you get to the level of "top 5 customer" (which AT&T almost assuredly is to whomever their equipment vendor is), the customer has quite a bit of input into the design of new products. Especially if the product is going to be sold primarily TO that huge customer.
I have an unlimited ipad data plan on AT&T, and for 1 month straight, I streamed video constantly just to test if they throttled or capped Unlimited Ipad plans. 128GB in 1 month and I was still getting 50+down/15up.
Evidence: My twitter images twitter.com/MichaelLargent
Leeching 128GB would only take 6 hours at the speeds you describe. I would certainly hope that they don't cap you after a period that could be as short as 6 hours!
I'm not sure what any of this has to do with the math. You claim 128GB over a 50Mb connection. That's less than six hours of constant streaming. Unless you want to qualify it with "during the times AT&T actually provided reliable service," your numbers seem off. A lot.
Maybe I wasn't clear (sorry). At the end of my testing after using 128GB (end of the month), I did the speed test to check if I was throttled.
And just because my ipad was capable of that high DL speed, my hosting server that I was constantly streaming HD movies from, isn't capable of that upload throughput (50Mbps uploading).
You're assuming that streaming video can provide data at the same rate as speedtest. In reality those numbers are nowhere close, and don't reflect the reliability of AT&T's service.
No, but the topic was network testing more than anything realistic. I missed the point, essentially. YouTube does have some 4K video though, which probably does hit that sort of bitrate.
10 Mbps is a good estimate considering modern video codecs, but it's not the upper bound. 1080p video at 32bpp and 60 fps could use as much as 3 Gbps, depending on the compression.
I've been researching something similar for a startup idea for a while now.
There still a couple of problems that still haven't been addressed (or just aren't included in this article)
1. 1 Cell site isn't enough to cover a densely populated area.
Although this is running on 850mhz which is much lower than wifi at 2.4ghz, multiple cell sites are usually better than 1 big one. If their peak is only 189mbps, then it should be fine using cat6 cable, however your limited when for range as you will run in trouble if you run cable over 100m without a repeater or booster. The other option is to use fibre, which is rather expensive but it wouldn't be that bad for AT&T considering their size.
2. Limited to the network beyond the cell towers.
This is coming from experience being at large festivals like Coachella, but although you can get signal on your phone most of the time, sms is virtually useless since your txt's are delayed by 4-5 hours. You can make a call after about trying 10 times, but because of the noise, you can never hear what the other person is saying and vice versa. So unless they are running a local relay for text's and transferring calls, their network servers are the ones that need beefing up rather than the towers themselves.
Now I'm guessing people are going to be using data more than anything, but in the past I've found 2g to be much more reliable (3g flatout did not work at the last festival, Big Day Out I was at). We found that Whatsapp became the most reliable way to communicate between our friends as your messages wouldn't get delayed for hours.
what do the AT&T vans connect to? I assume that there is no landline available for them to hook into, so how are they providing service?
my guess is satellites or communicating with other cell towers but the latter seems counterproductive as it would push a high load to a different cell tower. but if they used satellites, wouldn't there by high latency and bandwidth limits?
A few months back we were exploring a abannonded building (scheduled to be demolished), and we found there was a working cell tower ontop of the building. There was a small server room that had all the eqipment to run everything (I think it was shared by the 2 big telco's here in NZ).
From what we would could see, there were a bunch of ancient racks running 2g gear that looked at least 10 years old (all beige), and then a whole 42U rack of 12v batteries for a UPS.
Then there was a quarter-rack which had a bunch of fibre cables going into a few Huawei branded 1u boxes. LTE only got deployed a month ago so it much of been HSPA+.
I think if they had a satellite connection there would be too much latency. And the throughput wouldn't be limiting as well. They could of had one as a backup source though, or just setup as a secondary connection.
The image at the very top of the article mentions they have a microwave backhaul (see "E"), which I'm assuming goes to another ground location (i.e. a central office.)
Actually this makes better sense than what I was guessing about cell towers below. So, what I'm imagining is these temp towers will make sure that you can get connected.
I think ATT is a tier 1 network, so your connection to whatever data center will probably get bonded directly to whatever ATT datacenter which has the fastest response (ping) to the content your connecting to...love to see a traceroute for comparison.
I've actually wondered this too. I remember hearing about these temp towers at the 2013 inauguration. I understand how they allow more users to get a signal, but as you said.....what is the backhaul actually on? It could be distributed through a cell tower you may not get a signal from is my guess (like 1 that's 1mi away), or ran on a fios network (like voip).
I've seen line-of-sight microwave used for similar needs, but not in built up areas.
Edit: clicking through and actually reading the link, the picture clearly depicts Microwave backhaul. Probably to a site in the vicinity they have a nice big land-based pipe.
The article describes the microwave link as 'additional capacity', so I'm guessing that they had another (primary?) connection, but needed more throughput.
When the rugby world cup was running last year I saw a number of temporary cell sites, which had plastic ducting and fibre optics running ~50-200m to existing telecoms cabinets or cell sites.
I imagine that At&t is doing something similar, hooking into an existing cable at the nearest roadside cabinet or traditional cell site.
Does anyone know if they do any local web caching? There's nothing mentioned, but you'd think that a small-ish transparent web cache would save them a lot of bandwidth.
