Telkom LTE and Gaming Performance

[sTrideR|JP]

To demonstrate the gaming experience with Telkom Mobile LTE, I've recorder this short clip of Battlefield 1 with the network performance overlay enabled:

https://youtu.be/HmOraZr-Vbc

This is a representative example of the latency performance between approximately 18:00-22:00 on almost every weekday, sometimes going on until as late as 00:00, which is coincidentally the only times during the day that I am actually able to spend time on gaming.

As the clip shows, the latency averages around 450 ms - compared to around 210 ms during daytime - which is completely unplayable and characteristic of garbage service, whereas the 200 ms range is manageble, as most South Africans that play on European servers would probably agree. Also note this isn't exclusive to BF1, the same issues are present in BF3/4, WoW, Secret World; the latency becomes garbage as does the throughput of the connection. Local latency at the specified times are in the range of 180-250 ms, which is atrocious for servers that are less than 30 km from my home.

 

[myNeo]

To demonstrate the gaming experience with Telkom Mobile LTE, I've recorder this short clip of Battlefield 1 with the network performance overlay enabled:

https://youtu.be/HmOraZr-Vbc

This is a representative example of the latency performance between approximately 18:00-22:00 on almost every weekday, sometimes going on until as late as 00:00, which is coincidentally the only times during the day that I am actually able to spend time on gaming.

As the clip shows, the latency averages around 450 ms - compared to around 210 ms during daytime - which is completely unplayable and characteristic of garbage service, whereas the 200 ms range is manageble, as most South Africans that play on European servers would probably agree. Also note this isn't exclusive to BF1, the same issues are present in BF3/4, WoW, Secret World; the latency becomes garbage as does the throughput of the connection. Local latency at the specified times are in the range of 180-250 ms, which is atrocious for servers that are less than 30 km from my home.

Few things:

1. Which band are you using rather use lower dnd and upload band 1800mhz as it has much lower latency due to the nature of the frequency. This has been covered here but somewhere in the millions of posts. Do not use the 2300mhz band which is what most people thing needs to be used. Games require low latency not high DND and UPL speeds. Basic requirements should be taken into consideration i.e. as long as you dont have a up and down of like 0.8 up and 1mbps down.

2. ANY wireless by nature is not designed to be most efficient for gaming or "real time" Services as well as reliable. Watch your signal when you have a storm etc you will see how it fluctuates as well as during peak times. All these influence the latency and other requirements.

3. By nature LTE/3G Mobile networks are designed for high contention ratios due to the amount of devices, congestion will always be an issue.

4. The nature of Mobile connections it will drop the connection quickly reestablish it. Games hate this esp FPS they actually actively monitor it to protect from "lag switches" this then causes the protection mechanism to kick you out of the lobby or you to have a bad experience.

You can game with it but nothing beats fixed line. Just some notes on experience with this and gaming.

 

[sTrideR|JP]

Few things:

1. Which band are you using rather use lower dnd and upload band 1800mhz as it has much lower latency due to the nature of the frequency. This has been covered here but somewhere in the millions of posts. Do not use the 2300mhz band which is what most people thing needs to be used. Games require low latency not high DND and UPL speeds. Basic requirements should be taken into consideration i.e. as long as you dont have a up and down of like 0.8 up and 1mbps down..

I actually haven't tried playing with the LTE bands, I'll try that tonight and see whether it makes a relative difference, thanks.

2. ANY wireless by nature is not designed to be most efficient for gaming or "real time" Services as well as reliable. Watch your signal when you have a storm etc you will see how it fluctuates as well as during peak times. All these influence the latency and other requirements..

The signal is more or less constant in the -90 dBm range, it's raining here at the moment and sitting at -92 dBm, compared to the recording where it was a clear-skied evening. I know that wireless isn't optimal for gaming, but the fact that it performs incredibly well during the daytime defeats that argument.

3. By nature LTE/3G Mobile networks are designed for high contention ratios due to the amount of devices, congestion will always be an issue..

That's true, and to my understanding that's why Telkom imposed a limitation on the number of uncapped LTE consumers per base station - which either has not been implemented or it was done really poorly. Having a local speedtest drop from 40 Mbps / 5 Mbps down to barely 1 Mbps / 0.1 Mbps is disgusting.

4. The nature of Mobile connections it will drop the connection quickly reestablish it. Games hate this esp FPS they actually actively monitor it to protect from "lag switches" this then causes the protection mechanism to kick you out of the lobby or you to have a bad experience..

Again, the fact that this connection works perfectly well during daytime and turns into garbage in the evening defeats that argument. If it were true, I would experience massive latency 24/7, which is not at all the case.

 

[myNeo]

I actually haven't tried playing with the LTE bands, I'll try that tonight and see whether it makes a relative difference, thanks.



The signal is more or less constant in the -90 dBm range, it's raining here at the moment and sitting at -92 dBm, compared to the recording where it was a clear-skied evening. I know that wireless isn't optimal for gaming, but the fact that it performs incredibly well during the daytime defeats that argument.



That's true, and to my understanding that's why Telkom imposed a limitation on the number of uncapped LTE consumers per base station - which either has not been implemented or it was done really poorly. Having a local speedtest drop from 40 Mbps / 5 Mbps down to barely 1 Mbps / 0.1 Mbps is disgusting.



Again, the fact that this connection works perfectly well during daytime and turns into garbage in the evening defeats that argument. If it were true, I would experience massive latency 24/7, which is not at all the case.

I urge you not to use a once off application of latency test with the wrong method of a single metric because your signal is x now it should be better.

1. You don't know how your tower shares or transfers load or bandwidth. Does it pass or use a fiber backbone. You using a vs b in different times needs to be done more than once. In fact this is one of the four metrics at a single instance in time. My own opinion I would run it over a period of a week and look at the peeks and drops you could have brilliant RSNQ but your SRIN is low and the combination of these create a good signal. Many threads covering this here.

2. The fact that it performs sometimes and others doesn't actually proves my point of: congestion, signal strength, and band impedance. I am not here to validate points these are facts. I have it working well even for gaming, merely trying to point you in the right direction. so you can enjoy it to.

3. Go read up about 4G specifically the band Telkom uses and the frequencies the documentation talks about temperament and environmental factors one being heat = worse cool = great 3G doesn't have this. 5G doesn't have this. LTE-A introduced async to.try and aid in this.

4. Its working well the limitation infact to well and incorrectly at times on users that haven't exceeded. They limit the users they don't drop the users people still stay connected.

Hope this helps

 

[sTrideR|JP]

I urge you not to use a once off application of latency test with the wrong method of a single metric because your signal is x now it should be better.

1. You don't know how your tower shares or transfers load or bandwidth. Does it pass or use a fiber backbone. You using a vs b in different times needs to be done more than once. In fact this is one of the four metrics at a single instance in time. My own opinion I would run it over a period of a week and look at the peeks and drops you could have brilliant RSNQ but your SRIN is low and the combination of these create a good signal. Many threads covering this here.

Do you mean RSRP/RSRQ?

I'm not arguing that signal strength is the only factor affecting latency, of course not, but the fact that the connection turns into garbage during the evening points towards a congestion issue that I don't know how to address on my side. And this isn't new, I've had this connection since early February of this year and the latency issue has become progressively worse - from experiencing it 1-2 evenings per week to 6-7 evenings per week with results similar to the posted video, BF4 also has a network performance overlay that I used to check a few months ago. So it's certainly not a single data point.

2. The fact that it performs sometimes and others doesn't actually proves my point of: congestion, signal strength, and band impedance. I am not here to validate points these are facts. I have it working well even for gaming, merely trying to point you in the right direction. so you can enjoy it to.

I know you're trying to point me in the right direction, that's why I posted in the first place. But arguing that the connection is bad because it's bad isn't really helping to figure out what exactly the issue is and how I can fix it - but like I said I do agree that there are several things affecting the quality, I just don't understand why it should be bordering unusable for 4-6 hours per day, when I am at home and most active on the internet.

3. Go read up about 4G specifically the band Telkom uses and the frequencies the documentation talks about temperament and environmental factors one being heat = worse cool = great 3G doesn't have this. 5G doesn't have this. LTE-A introduced async to.try and aid in this.

4. Its working well the limitation infact to well and incorrectly at times on users that haven't exceeded. They limit the users they don't drop the users people still stay connected.
Hope this helps

That seems to be the problem: Telkom's dealing with congestion is limiting to the point of it being barely usable.

 

[myNeo]

There is nothing you can do. From the symptoms you describe its pure congestion. That's the answer - congestion gets fixed by more infrastructure to handle the load. The new roll outs are on "pause" while the whole huawai / telkom backbone migration takes place.

The additional users are not only LTE data users but cell phone users too they are the same towers. Telkoms mobile market share has significantly grown in the last four months. You might have a tower site down for upgrade close to you and every tom dick and Piet is connecting to your tower at night.

The last well informed stat I saw is that only 15-20% of the user base gets capped the rest actually give data back at the end of the month. So you competing with 85% of normal users that want the same or more speed as you.

Not only the R's actually the most important for latency and jitter is the SNiR or whatever the right spelling is (on mobile right now). You could be living on the tower and your R's are great but your SRIN is bad and you will have a bad experience for gaming. Won't effect large downloads though.

Look at my previous posts on the forum I use to get 102mbps with 16ms ping that's better than fiber over the last three months its gone to on 1800mhz best 10mbps and 15up with 14ms ping and 30mbps down with 15up on 2300mhz.

I even had the head of Huawai and Telkom mobile tech division here and we worked on numerous things that arrived at congestion, its their number one enemy right now.

Just a interesting point most of the other mobile providers have the same. They work on a similar principle of overbooking like airlines.

 

[sTrideR|JP]

There is nothing you can do. From the symptoms you describe its pure congestion. That's the answer - congestion gets fixed by more infrastructure to handle the load. The new roll outs are on "pause" while the whole huawai / telkom backbone migration takes place.

The additional users are not only LTE data users but cell phone users too they are the same towers. Telkoms mobile market share has significantly grown in the last four months. You might have a tower site down for upgrade close to you and every tom dick and Piet is connecting to your tower at night.

The last well informed stat I saw is that only 15-20% of the user base gets capped the rest actually give data back at the end of the month. So you competing with 85% of normal users that want the same or more speed as you.

Not only the R's actually the most important for latency and jitter is the SNiR or whatever the right spelling is (on mobile right now). You could be living on the tower and your R's are great but your SRIN is bad and you will have a bad experience for gaming. Won't effect large downloads though.

Look at my previous posts on the forum I use to get 102mbps with 16ms ping that's better than fiber over the last three months its gone to on 1800mhz best 10mbps and 15up with 14ms ping and 30mbps down with 15up on 2300mhz.

I even had the head of Huawai and Telkom mobile tech division here and we worked on numerous things that arrived at congestion, its their number one enemy right now.

Just a interesting point most of the other mobile providers have the same. They work on a similar principle of overbooking like airlines.

Well that sucks then. I played around with the LTE bands but it didn't seem to make a noticeable difference, but it'll take some time to figure out which one is working better.

Strangely enough here's a speedtest from just now - and it's storming rain at the moment - showing a really decent ping with RSRP = -98 dBm and SINR = -5 dB; both which confirm that the RF conditions are pretty awful:

 

[myNeo]

Well that sucks then. I played around with the LTE bands but it didn't seem to make a noticeable difference, but it'll take some time to figure out which one is working better.

Strangely enough here's a speedtest from just now - and it's storming rain at the moment - showing a really decent ping with RSRP = -98 dBm and SINR = -5 dB; both which confirm that the RF conditions are pretty awful: View attachment 400489 View attachment 400495

Without having all your data I can see one glaring problem a SINR of -5 is terrible! So that means you have a goodish physical strength to the tower but you have an enormous amount of noise / jitter / latency. Rather use the Siax speed test or any other that you might get to test the Jitter on the line whilst ping is one indicator you need to factor the "noise" on the signal with the Jitter.

I also noticed you doing a speed test to a local server in Pretoria, its a false test which many people dont realize. You should be doing a speed test to either a European (UK, Germany) or USA server. Or whilst on BF1 figure out which DC you connecting to. This will give you the correct speed. Local test is always going to have a great ping, less physical travel and most the test servers sit on backbones so it will be quick.

If you playing on a console i.e. Xbox One settings>Network>Detailed Network will give you the true ping rate to the server, packet loss and up and down speeds to the actual server. Don't know or bother with PS4

If you on a PC check the ToolBox tool suite there are numerous threads here the tool is LTE Watch that gives you real time signal, jitter, dropped packets etc. If you can find the IP range of the BF1 servers (easy to get on the internet) you can Ping them directly and the tool gives you the relevant info i.e. Packet Loss, Ping Rate, Jitter etc

This was a invaluable tool for me to trouble shoot as its constant and not point in time. Takes one factor to drop the signal significantly for the game to think you manipulating networking and penalize you in which ever way it does. Some games kick you out of the lobby, some games penalize you by applying damage ref which just makes your experience cr@p and everyone else just pot shots you. It depends on the game.

The unfortunate part is a wireless signal is never stable. Its just about making it stable enough for the bounds of the application - in this case a game.

On the bands it could simply mean the towers in your area are not 2300mhz compatiable so even though you forcing a band its defaulting or its reaching a tower far from you.

For gaming 2300mhz is cr@p dont use it even if the dwnld is 2000000000000000mbps. The latency due to the nature of the frequency being longer weaker wave lengths will never live up to the stronger shorter and more pentratble 1800mhz.

HTH

 

[sTrideR|JP]

Thanks I'll definitely give LTEwatch a go and see if I can figure something out there. I've been using the E5186 toolbox to change the frequencies, setting it to 2100 MHz flat out disconnects me from the network, but forcing it to 1800 MHz or LTE-A seems to be fine.

After the speedtests done earlier I tested in game and averaged around 195-205 ms latency, with the RSRP = -98 dBm and SINR = -5 dB, during storming rain, sick.

From my experience, during said bad times the local speedtest ping was usually over 200 ms, which is garbage, which is why I posted the reference tests during horrible RF conditions. So logically, while RF conditions obviously play a role, some or other bull**** is going on that is causing the drops in performance.

 

[myNeo]

Thanks I'll definitely give LTEwatch a go and see if I can figure something out there. I've been using the E5186 toolbox to change the frequencies, setting it to 2100 MHz flat out disconnects me from the network, but forcing it to 1800 MHz or LTE-A seems to be fine.

After the speedtests done earlier I tested in game and averaged around 195-205 ms latency, with the RSRP = -98 dBm and SINR = -5 dB, during storming rain, sick.

From my experience, during said bad times the local speedtest ping was usually over 200 ms, which is garbage, which is why I posted the reference tests during horrible RF conditions. So logically, while RF conditions obviously play a role, some or other bull**** is going on that is causing the drops in performance.

That is not bad ping rates to be honest. The fastest you can physically send light from ZA to a london data center is 150ms. Thats best case scenario as crow flies min hops best conditions with fiber. Consider you using Wireless on a disjointed backbone that then has to still link to the main backbone then follow the route of all the other ADSL/Fiber data and out the country an additional 50ms is not bad.

Search the forum how to set 2300mhz using the netmode in the API on the Toolbox. 2100 != 2300 infact its worse than 1800.

 

[myNeo]

also go download (I paid for a copy you get 14 days free) ping plotter, it will show you exactly where the slow down is on which hop. Might help you isolate the issue. Just be prepared to do some debugging.

 

[RedhotNeo]

Are you using the LTE antenna. that will definetly boost your signal strength..

Those pings/latency are terrible.. i usually get that when using a vpn connection but normal is under 20ms even in bad weather conditions and im pretty far from the closest tower

 

[sTrideR|JP]

That is not bad ping rates to be honest. The fastest you can physically send light from ZA to a london data center is 150ms. Thats best case scenario as crow flies min hops best conditions with fiber. Consider you using Wireless on a disjointed backbone that then has to still link to the main backbone then follow the route of all the other ADSL/Fiber data and out the country an additional 50ms is not bad.

Search the forum how to set 2300mhz using the netmode in the API on the Toolbox. 2100 != 2300 infact its worse than 1800.

How did you get to 150 ms for light to travel between SA and London?

I'll check out Ping Plotter, thanks.

@RedHotNeo I'm not using the large mounted antenna no, but I'm planning to install it after the Telkom techies were here and wanted to mount it on a tree. I don't think that'll do much for my latency though, although it will surely improve the maximum throughput that I would be able to get. Like I said, for 2-3 nights during the storms here I was able to get BF1 pings in the range of 190-205 ms, even though the throughput was around 1.5-2 Mbps as a result of the really, really poor RF conditions.

 

[myNeo]

How did you get to 150 ms for light to travel between SA and London?

I'll check out Ping Plotter, thanks.

@RedHotNeo I'm not using the large mounted antenna no, but I'm planning to install it after the Telkom techies were here and wanted to mount it on a tree. I don't think that'll do much for my latency though, although it will surely improve the maximum throughput that I would be able to get. Like I said, for 2-3 nights during the storms here I was able to get BF1 pings in the range of 190-205 ms, even though the throughput was around 1.5-2 Mbps as a result of the really, really poor RF conditions.

When they were testing the best possible route to the DC's in Europe they did the least amount of hops first to get the lowest ping rate. The engineers then calculated a theoretical travel speed given data transfer over light and got to a figure of 150ms as the fastest.

 

[sTrideR|JP]

When they were testing the best possible route to the DC's in Europe they did the least amount of hops first to get the lowest ping rate. The engineers then calculated a theoretical travel speed given data transfer over light and got to a figure of 150ms as the fastest.

Oh so it's the physical travel time (~40 ms) plus additional delay introduced by subsequent hops, wasn't too clear.

 

[My name is Husain]

I actually haven't tried playing with the LTE bands, I'll try that tonight and see whether it makes a relative difference, thanks.



The signal is more or less constant in the -90 dBm range, it's raining here at the moment and sitting at -92 dBm, compared to the recording where it was a clear-skied evening. I know that wireless isn't optimal for gaming, but the fact that it performs incredibly well during the daytime defeats that argument.



That's true, and to my understanding that's why Telkom imposed a limitation on the number of uncapped LTE consumers per base station - which either has not been implemented or it was done really poorly. Having a local speedtest drop from 40 Mbps / 5 Mbps down to barely 1 Mbps / 0.1 Mbps is disgusting.



Again, the fact that this connection works perfectly well during daytime and turns into garbage in the evening defeats that argument. If it were true, I would experience massive latency 24/7, which is not at all the case.

My telkom lte works like crap from 7 pm to 10pm as well thought I was the only one with this problem. Now I see that I am not alone.

 

[Snyper564]

My telkom lte works like crap from 7 pm to 10pm as well thought I was the only one with this problem. Now I see that I am not alone.

this post is 6 years old....

Log a ticket with Telkom they will resolve it