Results of tests in Washington, D.C., are in line with the kinds of data speed claims made by two operators – Verizon and Clearwire. Stay tuned for future tests that include AT&T and T-Mobile in the mix.
Afew weeks ago, I was in Washington, D.C., and decided to measure data speeds for the two networks for which I have equipment. I measured Verizon’s LTE and Clear’s WiMAX networks every day, several times a day from my hotel room. I was not set up to measure the speed of the T-Mobile USA or AT&T networks but when I am back in D.C. in mid-March, I should be able to do so. As I travel this year, I will be making measurements on as many networks as I can from as many locations as I can.
Since the term 4G has become so widely used for WiMAX, LTE and HSPA+ networks, the buying public is confused and does not know exactly what 4G is except that it is supposed to be faster than 3G. I have been asked many times in the past few months how someone with a T-Mobile or AT&T account now has access to 4G when they signed up for service and it was only 3G. I explain to them that both AT&T and T-Mobile have enhanced their 3G networks and upgraded them and that now they are using technology which is accepted by the 3GPP standards body as 4G, and they should notice a speed increase on these networks.
For this trip, I decided to run all of the tests from my hotel room, using the same laptop and exchanging modems between the Verizon LTE network and the Clear network. In addition, I chose Speakeasy.net/speedtest because it offers both downlink and uplink testing and testing from servers across the United States. For this series of tests, I used the Washington, D.C., server because that would give me the shortest use of the Internet. The Internet itself is a variable that needs to be taken into account when running tests.
I have run extensive speed tests from my office using servers in Los Angeles, San Francisco, Washington, D.C., and New York. I have run tests every day in the morning and in the evening from my office location and have charted the differences in data rates, which have been reported using these servers. I have determined that the data speeds reported using Speakeasy’s speed test vary a great deal between the West Coast and the East Coast and also based on the time of day. In general, the tests made using the more local servers show faster speeds than those on the East Coast, and sometimes there is as much as several Megabits per second difference in the speeds, which I believe has to do with the loading on the Internet. Therefore, for these tests, I used the closest server to my location.
The caveat for all of these tests is that I have no way of knowing how much other traffic is on the particular cell sectors from which I was operating in. But since the tests were run for four days and at different times, I believe that this is a good indication of the broadband speeds which were available to me for my tests.
As you can see from the chart, Verizon’s LTE is a little more than twice as fast as the Clear WiMAX system, and if you take into account that the Verizon system is running in 20 MHz of spectrum (10 MHz for uplink and 10 MHz for down link) and that the Clear system, as it has been described to me, makes use of 10 MHz of spectrum for each of its cell sectors by using Time Division Duplex (TDD) rather than Frequency Division Duplex (FDD), these tests are as close to being equal in the amount of spectrum being used.
Next time I am in Washington, I will be taking my tests on the road as it were, making them in taxis as I traverse the city and then in several coffee shops close to my destinations. I also hope to be able to test AT&T and TMobile’s systems at the same time and will report on these results as well.
As I continue to travel across the country, I will be testing as many wide area broadband services as I can. These test results are based on a fixed location, not mobile, but are, to me, a good indication of the data speeds available across two of the services. Again, this is within a single cell sector, and the total number of users within that cell sector was an unknown during my tests.
I believe that Verizon’s LTE network is set up as a standard Category 3 network, meaning that it is making use of all three download speeds (fastest closer to the center of the cell, medium in the middle and slowest speed at the cell edge), and two data speeds for the uplink. This is a fairly standard set of network settings for LTE and, of course, there are many ways of changing the settings and optimizing the network and I am sure that over time, Verizon will be tweaking the network – perhaps on a cell sector by sector basis in order to achieve the maximum amount of throughput for each area.
The results I obtained are in line with both network operators’ claims for data speeds, which is a good sign that neither operator is misrepresenting their data speeds, as has been the case in the past by others. Again, the caveat is that these speeds will vary depending upon the network loading in any given cell sector. For example, if at the same hotel, facing the same direction, there were many users, some of whom were streaming video down to their laptops, the results would have been lower on both networks. The issue for network operators is broadband services are often times consumed from fixed locations, and the architecture of a cell system is based on customer movement from sector to sector.
This is one reason that some operators have run into a loading problem, and it is important to keep in mind that the problem for wireless is sector-based and not systemwide. In a city such as Washington, D.C., it is possible to have a large number of users within a single sector, making the network seem slower while other users a block or two away will be experiencing much faster data access. As network operators turn to indoor cells (femto) and handing off to Wi-Fi systems, the demand for services will be spread out not only across the wide area network but also off the network, especially inside homes and office buildings. This will help network operators manage the data traffic across their networks.
However, it is important to remember that the network operators do need to carefully manage their networks so that they can provide the best possible service to as many of their customers as is possible. Within LTE there are many more ways to manage data capacity from a network perspective than in any other air interface, but pricing will remain as a major way to manage data usage going forward. The more data a customer uses, the more it is going to cost. This is no different than the way we purchase our electric power or water today. The more we use in a given month, the more it costs us for Kilowatt hour of usage or cubic meters of water we use.
One final point: Both LTE and WiMAX perform differently when using different amounts of spectrum. For example, the proposed public safety nationwide interoperability broadband network, as of today, only has 10 MHz of spectrum available (5 X 5 MHz), and if we use the Verizon LTE numbers and divide by two, then the public safety network will have an average download speed of 7.369 and an uplink speed of 2.048. It should be clear to all of us that this is not enough spectrum for public safety since most of their incidents occur within a single cell sector. Using these numbers, those on the scene trying to send video – say from a sniper scope or from a portable camera – using a data rate of 1.2 Mbps would swamp the network with only two cameras running and no one else would have access.
LTE is a great new broadband offering, but it needs to be matched with enough spectrum to make it usable. The LTE specifications say that LTE can be used in 1.4, 3, 5, 10, 15 or 20 MHz portion of spectrum. However, when used in a 5 X 5 MHz configuration, the capacity per cell sector is half of what it is in a system making use of 10 X 10, as the Verizon system is doing. It is not just about the wireless technology for a broadband network but a combination of the technology and the amount of spectrum which is available.
I am looking forward to updating my test results and adding both AT&T and T-Mobile to the mix. There will be many organizations providing test results measured using vans and a lot of very expensive test equipment, which also will take into account the loading on any given cell sector. My tests are not designed to replace or compete with those done by the experts; rather, I view mine as real world tests with results that will be fairly typical for the network customers of a particular network.
Seybold heads Andrew Seybold Inc., which provides consulting, educational and publishing services. For more information, visit http://www.andrewseybold.com.