Confused on LTE upload speed

[Yuhfhrh]

I see a lot of the time references saying sprint's 5X5 LTE has a theoretical 37mbps down and 10+(?) up. And also Sprint promises 6-8mbps down/2-3mbps up. Why is the upload always less than the download? The few days I've had LTE here in the Olathe, KS area the upload is almost always just as high if not higher than the download. Typically 7mbps down/up to 12mbps down/up and more recently 7mbps down and 9mbps up. Why is it like this? Is Sprint going to limit upload speeds in the future or something?

Edited July 16, 2012 by yuhfhrh

[pii100]

Upload is typically slower in wireless systems due to the fact your phone has a weaker transmitter than the tower.

[Yuhfhrh]

Upload is typically slower in wireless systems due to the fact your phone has a weaker transmitter than the tower.

 

Thanks! That makes a lot of sense.

[pyroscott]

Also, there are 2 receive antennas and only 1 transmit. You can have MIMO on your downlink but not on your uplink.

[WiWavelength]

Also, there are 2 receive antennas and only 1 transmit. You can have MIMO on your downlink but not on your uplink.

 

Well, technically, the uplink, too, can use MIMO. WiMAX can utilize uplink MIMO. But, as I mentioned the other day, it was just too soon, the technology too immature. However, uplink MIMO will return with LTE Advanced.

 

AJ

[WiWavelength]

On a related note, I have observed several times what appears to be uplink transfer rate shaping. Note the early burst, then significant drop in the uplink transfer rate vs time graphs.

 

 

 

Honestly, if true, uplink rate shaping would make sense. Most uploads are relatively small -- no more than 1-2 MB. So, provide high uplink throughput for the first 1-2 seconds. That will complete most uploads, give the appearance of faster speeds, and get the uploader off the network more quickly.

 

Any other observations of this phenomenon?

 

AJ

[S4GRU]

On a related note' date=' I have observed several times what appears to be uplink transfer rate shaping. Note the early burst, then significant drop in the uplink transfer rate vs time graphs.

 

Honestly, if true, uplink rate shaping would make sense. Most uploads are relatively small -- no more than 1-2 MB. So, provide high uplink throughput for the first 1-2 seconds. That will complete most uploads, give the appearance of faster speeds, and get the uploader off the network more quickly.

 

Any other observations of this phenomenon?

 

AJ[/quote']

 

I have noticed this in several posts from other members where you can see the whole upload test trail. Interesting.

 

Robert via CM9 Kindle Fire using Forum Runner

[4ginnc]

On a related note, I have observed several times what appears to be uplink transfer rate shaping. Note the early burst, then

Honestly, if true, uplink rate shaping would make sense. Most uploads are relatively small -- no more than 1-2 MB. So, provide high uplink throughput for the first 1-2 seconds. That will complete most uploads, give the appearance of faster speeds, and get the uploader off the network more quickly.

 

Any other observations of this phenomenon?

 

AJ

 

I've noticed something similar even on my legacy 3G connection at home using Sprint's speed test app and the speedtest.net app. The speed will spike at the very beginning, usually to something like 256kb/s, then sit there for a few seconds until it starts a more realistic rate of traffic (usually < 100kb/s for me).

 

It doesn't sound like traffic shaping on the network side to me, but more like the IP stack trying to figure out the optimal flow on the network. I've seen similar situations with ftp traffic in the past when we were debugging some network issues at work, but the slowdown was caused by packet loss by some faulty intrusion prevention devices. So when the network layer realized things weren't as fast as it should have been, it slowed down the send rate.

 

Just a guess, but that's what I've been seeing.

[iansltx]

Good old TCP slow start/fast recovery...