Thread: General STM always enforced?
View Single Post
Old 17-10-2013, 22:22   #194
qasdfdsaq
cf.mega poster
 
Join Date: Aug 2004
Posts: 11,207
qasdfdsaq is cast in bronzeqasdfdsaq is cast in bronzeqasdfdsaq is cast in bronzeqasdfdsaq is cast in bronze
qasdfdsaq is cast in bronzeqasdfdsaq is cast in bronzeqasdfdsaq is cast in bronzeqasdfdsaq is cast in bronzeqasdfdsaq is cast in bronzeqasdfdsaq is cast in bronzeqasdfdsaq is cast in bronzeqasdfdsaq is cast in bronzeqasdfdsaq is cast in bronzeqasdfdsaq is cast in bronzeqasdfdsaq is cast in bronzeqasdfdsaq is cast in bronzeqasdfdsaq is cast in bronzeqasdfdsaq is cast in bronze
Re: STM always enforced?

Quote:
Originally Posted by Kushan View Post
I thought you said that only Cat6-8 were LTE-A? The Snapdragon 800 is only CAT-4, so is there more to it than that?
Cat6-8 are LTE-A only, support for them requires a device to be LTE-A. LTE-A did not remove the previous categories though so an LTE-A device does not have to be cat 6/7/8, it can also be cat 4. In this case we have a device that supports LTE-A capabilities (i.e. carrier aggregation) but not the increased speeds it brings, and only supports previously existing LTE speeds. The main benefit at the moment is to allow carrier aggregation so operators that do not have fully contiguous allocations of the maximum available spectrum to achieve standard LTE speeds by combining multiple, smaller blocks. For example 3, which has allocations of 2x15Mhz and 2x5Mhz can combine the two into a single 2x20Mhz to achieve 150Mbps with LTE-A, while with LTE (standard) they could never achieve full Cat 4 speed. It doesn't benefit anyone else really, since EE and Vodafone already have multiple contiguous blocks of maximum size, which is a very rare asset globally.


Quote:
This is true, but how much bandwidth is there to go around, really? DC-HSDPA+ is meant to be 42Mbit but I rarely ever see it break double digits.
That depends on your cell size and how well you manage/optimize your network. Things change quickly - the average I get from DC-HSPA now is 11Mbps, compared to 2Mbps a year ago - though O2 still does 12Mbps without DC-HSPA in my neck of the woods. Plus, dual-cell is implemented really really badly in the UK by all networks IMO. And then there's the fact the app most people use (Speedtest.net) being next to useless for anything above 10Mbps on mobile.

For LTE(A) EE and Sodafone both have 3-4x the allocated spectrum than they had for 3G, and with LTE MIMO is basically universal thereby doubling capacity again under certain conditions. That's why EE advertise "average" speeds on LTE as around 5x that of 3G, which they conservatively put as 10-15Mbps vs. 2-3Mbps

Quote:
Not only are you dealing with the perils of a wireless technology, you're also competing with sometimes hundreds of other people. I would like to know how much capacity the various spectrum blocks can truly have.
Well then you're looking into matters of cell density and spectral efficiency, but take a typical example of a 3-sector macrosite in a city, which would deliver a max of 3x150Mbps (450Mbps) usable capacity using a single 20Mhz carrier, EE and Vodafone both have enough spectrum for three of these, putting them at a total capacity of 1350Mbps per site. Cell density can be dozens to hundreds per square mile in cities, and total 15,000+ across the UK per network. To be fair capacity will more likely be limited by the gigabit backhauls VM are putting in for these sites, if not by poor signal or interference than total spectrum capacity.

To put it in perspective, the amount spectrum each of the larger networks have dedicated to 4G is roughly equivalent to the amount VM have dedicated to "fibre broadband" - except with equal down and up spectrum. So instead of 16 down, 4 up, you've got the equivalent of something that's more like 15 down, 7.5 up on mobile (7.5 down x2-way MIMO)

---------- Post added at 22:22 ---------- Previous post was at 22:21 ----------

Quote:
Originally Posted by kwikbreaks View Post
I did think I'd seen STM over the weekend but it may just have been congestion as my SamKnows reports show that there is plenty of that but using multiple streams still allows downloads using a download manager or TBB speedtests to hit the headline rate.
That does sound distinctly like congestion. I usually use a UDP speedtest to rule that out as that completely eliminates the effect of "multiple streams".

Quote:
No I was wrong. STM has now kicked in but about 30 minutes after the downloads completed. That's quite a joke really if speeds are restricted way after the time when it actually matters.
Joke? VM? Blasphemy!
qasdfdsaq is offline   Reply With Quote