x
4G/3G/WiFi

Road-Testing Verizon's 4G-LTE

I had the chance to carry out a real-world road test of Verizon Wireless's 4G-LTE service recently. On a week's trip to New York for Light Reading's annual Mobile Backhaul & Core event, and then on to Silicon Valley, I was able to put the world's preeminent LTE network through its paces.

My conclusion: It's better than any mobile network I've ever used. It delivers substantially better speeds than advanced HSPA networks and sets a new performance baseline for mobile broadband.

To run the tests, I used a Novatel 4G-LTE MiFi device, paired with a couple of smartphones and a laptop, and used the service as the typical business traveler would – in that I needed access at hotels, in conference rooms, in meetings, at the airport, traveling in cars, and at restaurants and bars. In other words, this is a real-world review of mostly typical usage scenarios.

Certainly the service felt fast and reliable, but to get a more objective view I ran a series of tests using the Speedtest.net app on the new Motorola Atrix smartphone I also happened to have on review. The 50-odd results are by no means a definitive measure of network performance, but do reflect real-world locations, usage and coverage. The results are shown in the chart below:



Bear in mind that most of these results were indoors or from moving vehicles – both of which are factors that normally reduce throughput – and not generated where I had line-of-sight to the antenna or was particularly close to the base station. In fact, I never especially looked for an LTE antenna during the week of testing.

To summarize, my results were as follows:

  • Downlink Performance: Speeds of 4-5 Mbit/s were common on the downlink and virtually always above 2 Mbit/s. Interestingly, I never got beyond 5 Mbit/s. I'm not sure if that is because the network is configured to generate that type of per-user throughput, or of if didn't burst higher just because I was never in good enough signal conditions to achieve high peak speeds. For reference, if you consistently get 1 Mbit/s downlink performance on an HSPA network in London you're doing well these days.

  • Uplink Performance: This was a revelation and consistently delivered 1 Mbit/s and often bursting to 3 Mbit/s. This compares favorably with to the 200-300 kbit/s I'm used on to on HSPA and makes applications such as video calling viable. It's interesting to note that TeliaSonera has also highlighted uplink speeds on its first LTE networks came as pleasant surprise.

  • Latency: This is harder to call. At around 100ms roundtrip times were, without question, good. But they weren't absolutely stunning either, and as often as latency would drop down to 93ms or 94ms at best, it would rise to 110ms to 120ms (although hardly ever beyond that). This is by no means poor performance for a mobile broadband network and two years ago would have been incredible, but after a period of intense optimization, latency in the most advanced HSPA networks is now close to par with these numbers. It's possible the latency figures reflect the larger cell sizes of 700MHz, or perhaps the underlying IP and EPC network engineering? In any case, it is an area where ongoing optimization will improve performance over time. This is still a very new network, after all.


Beware of reading too much into these speed-test figures, which I think underplay just how good the network is from an end-user perspective. My anecdotal experience was that the service is nothing short of superb. Being able to pull a solid 4 Mbit/s while in a vehicle moving at 70 mph is incredible performance. Maintaining a broadband connection in a taxi though the urban canyons of New York City is a true feat of radio network design. And being able to make Skype calls at 50-60 mph on the Bayshore Freeway is, well, kind of fun (for nerds, anyway).

Note, too, that these are all examples that highlight the value of 700MHz spectrum and how it enables the combination of speed and coverage to make a service compelling to users. No doubt 800MHz in Europe will be similar.

The Novatel MiFi unit was easy to use (it has only one button), hangs on to the LTE signal incredibly well – as the vehicle tests demonstrate – and kicks out a decent WiFi signal. Under heavy use (streaming video, or from a moving vehicle) the battery would drain within a few hours and the MiFi would get warm – hot, even. But that's to be expected: Heavy data usage consumes mobile device battery capacity and there's no way around that. My 3G personal hot spot is the same.

For "snacky" smartphone use, the MiFi drops to dormant mode to save power, and with a laptop it powers off USB, so for practical purposes it worked out pretty well. In a hotel room it could be plugged in and forgotten about – it felt like using always-on, premium WiFi.

— Gabriel Brown, Senior Analyst, Heavy Reading

Page 1 / 2   >   >>
sandnetguy 12/5/2012 | 4:59:50 PM
re: Road-Testing Verizon's 4G-LTE

instead of using a mifi, try this test again with a usb dongle.  the numbers i get with my pantech in la, orange county, san fran, san jose, nyc, oakland, hayward is downlink between 12-35 mbps depending on if i'm indoors or out. 


uplink is 4-18 mbps


latency sub 40-50 ms


my thunderbolt has slightly different numbs and 2x the latency.


acting as a hotspot, my hp cant do better than 1 mbps down and 3 up...but my wife's macbook air gets 12-15 mbps...i think there is a glitch on the wifi b/c i've heard/read similar issues...certain notebooks get good wifi connections and others get crap...havent cared enough to figure out why since i use my dongle.


 


basically, i'm saying interesting test.  but i know from frequent testing and usage in the cities i mention that the LTE performance is significantly better than the stats you presented.

sandnetguy 12/5/2012 | 4:59:49 PM
re: Road-Testing Verizon's 4G-LTE

ps...i think the latency you are experiencing is being introduced by the mifi.  do the test again with the dongle vs. mifi and you'll see a big difference.  will get cut at least in half.  95-100 ms is what i get when i use thunderbolt as hotspot.

Gabriel Brown 12/5/2012 | 4:59:46 PM
re: Road-Testing Verizon's 4G-LTE

 

Hi sandnetguy,

Thanks for your comments. Looks like you are getting terrific performance.

I had considered the device would impact the results (since this is obviously a factor), but didn’t have a way of testing against, for example, a USB modem. This was a pure on-the-fly review as I went about my daily business.

My experience with 3G MiFi’s and 3G dongles is that performance is very similar. I don’t think the MiFi made much difference to latency in these tests, although your experience perhaps suggests otherwise. I’d be interested to see data of a USB and MiFi side by side at the same time and location. 

One issue with USB modems is you don’t normally use them in the same way as a MiFi/smartphone and possibly the laptop scenario is more favorable to higher performance. 

Recent Novarum speeds tests in this article show average downlink of 6.4 Mbit/s and 128 ms latency using a USB modem over a large number of tests, which is in the same ball park as I observed.

In any case, to the extent the actual performance is substantially better than I experienced underlines the thrust of the review: that it is a excellent network and a step beyond what we’re used to 3G.

Interested to hear any other experiences you have with LTE.


 

sandnetguy 12/5/2012 | 4:59:45 PM
re: Road-Testing Verizon's 4G-LTE

lets try this link instead...


https://picasaweb.google.com/117109672477919912516/LTE?authkey=Gv1sRgCLfV25_2opuXWQ

sandnetguy 12/5/2012 | 4:59:45 PM
re: Road-Testing Verizon's 4G-LTE

is there a way to upload pictures?  maybe i'll put a link up on a photoshare site.


so, i've had the LTE dongle since jan 7 and travel a fair bit and i have a few thots that are a bit against the grain.


a) silicon valley/san fran has the best performance.  in back of cab, going 70 is where i got 35 mbps down and 16 up with 40s ms ping.  in various hotels, speeds werre more like 12 x 16, pings in the 40s.


b) la, depending on street by street, it can be 12x9 or 20 x 12 or 16 x 4.  in my house, i consistently get 12 x 5 with 30s ms pings.  compare that to my cable modem 16 burst down, 1.7 up and 45+ ms pings...LTE.


c) oh, and so far the "wait till networks get loaded" argument has been bunk.  pure bunk.  if anything the network has gotten faster since jan in almost every area.


and in case anyone is wondering.  no, i'm not a verizon employee.  no, i'm not a verizon fan boy.  no, i'm not a paid blogger.  i just find the entire subject of LTE and its potential to be very transformative, thus interesting.


a big part of the "transformative" is the latency.  fat pipe, eh, whatever.  fat pipe, better than landline latency?  ok...combine both and its a real big deal.


ps...the worst ping times i consistently get is in nyc.  usually in the 60s-90s...i usually find the terrestrial handoff to be in PA on comcast.


lets see if this works:


https://picasaweb.google.com/lh/webUpload?uname=117109672477919912516&aid=5628104916539256081


just a smattering of pics from past 6 months.  notice my notebook issues with hotspot.

gtchavan 12/5/2012 | 4:59:40 PM
re: Road-Testing Verizon's 4G-LTE

I spent a month with Verizon's 4G plan and I have to say conclusively, Verizon 4GLTE is a scam, their data is almost always delivered in 3G.  Even in their store they could not deliver 4G consistently.  I received 4G LTE data rates in one or two occasions past midnight and not even in the same area.   Even worse this same mifi device that you supposedly had great experience with, every day at least once was signalled down by the base station so I had to reset by removing the battery (what a pain).  They blamed it on defective device every time I went to the store and I exchanged their mifi device 3 times.  All 3 mifi devices had the same problems so it was not a defective device issue.  Don't fall for the defective device trick this is how they keep you arround past 30 days so you have to pay contract cancellation fees.


My conlcusion is this that Verizon has not upgraded their infrastructure front or back end to deliver 4G data and they offer 4G by policy and not for all.  I used to have a 3G modem that had unliminted data and it could consistently deliver 2mb/s of data and they switched me to 4G and I couldn't even go to my old plan.  I therefore had to cancell my data plan with Verizon.  It costed me $200. This should be illegal. 


Didn't you have to take out the battery even once to reset the device?  Did you ever observe a green LED on your MIFI device, if so that is the only time you had 4G LTE.  In one or two occasions that I had green LED I was seeing 16MEG upload and down load speeds did you ever see this?  That was the best data rate I ever observed but since they manage delivery of LTE by policy you can't get the same data rate in the same spot (ever).  I tried it in Palo Alto, Mountain View and San Jose and same results.


My advise is stay with your 3G plan, until Verizon comes clean and tells us when they will truely have 4GLTE implemented.   

Gabriel Brown 12/5/2012 | 4:59:40 PM
re: Road-Testing Verizon's 4G-LTE

@sandnetguy -- the best location for performance in my trip was San Franciso airport. 

Gabriel Brown 12/5/2012 | 4:59:40 PM
re: Road-Testing Verizon's 4G-LTE

Hi chuckj -- I can't remember if its the green or purple light, but whichever it was I'm pretty sure it was virtually always on LTE. You can see when it dropped to 3G by the latency spike in the chart. 


VZW has, presumably, invested heavily in coverage in New York and Silicon Valley. If you have only a weak signal (or none at all), then perhaps that explains your problems?


I didn't have to reset it or remove the battery over the three or four days I used it.


 

gtchavan 12/5/2012 | 4:59:37 PM
re: Road-Testing Verizon's 4G-LTE

With all due respect Gabriel, you should have read the owners manual on your modem.  It clearly states that the only time you have 4G is when the LED is green and believe me when it is green it is data hog heaven but you will practically never see green LED even in their own stores.   I seriously doubt that you did a serious study of their data rates if you did you would have read the owner manual of the module.   


As for not having to reset your modem once a day, I simply do not believe it.  I had 3 modems that all had to be resetted once a day.  It would have been nice if the device had a hard reset so you would not have to remove the battery to reset the device. While I had my 3 different modems, it was very obvious that the base stations simply force the data hogs down, especially in the afternoons.  Again with all due respects, your article is so positive as if you were paid to write it.  We all know where you advertisement dollars come from.  So I am not suprised.


Here is the link to the modem's manual:


http://www.novatelwireless.com/content/pdf/Support/Verizon/90026384%20R1%20UG%20MiFi4510L%20VZW%20NVTL%2015Apr2011.pdf


 


 

Gabriel Brown 12/5/2012 | 4:59:36 PM
re: Road-Testing Verizon's 4G-LTE

It was definetly on LTE. It only has one button, no need for a manual.


I did switch it off at least once day though, mostly to save battery, which drains quickly, as noted. That would be equivalent to a reset I suppose.


True, I am pre-disposed to a positive view of the network and technology. I'm interested in getting a sense of where the technolgy will take us in a few years and trying to understand the potential of LTE for new, better services.


It was the same when 3G first launched -- it wasn't perfect, obvioulsy, and there was a lot of bellyaching around the industry, but still I was mostly thinking: "Wow, Internet in a car. Internet wherever you go in London. Mobile Internet, that's awesome."


 

Page 1 / 2   >   >>
HOME
Sign In
SEARCH
CLOSE
MORE
CLOSE