Forum Discussion
My tmobile cell tower is overloading my router settings to make it worse
I just got this tmobile home internet last monday and for two days it was amazing and then someone started messing with the settings to make it worse by alot.
One thing they did was change the TCP so any new connection starts out with a very slow download and then it ramps up to better speeds, i can only get full speeds on a download service like steam now, webpages use next to no download speed and if you have auto quality enabled or you can’t choose a video quality for a video, all videos on computer and TV start out at like 160p and then go up to 720p or more after a while.
Next they keep overloading my router settings to force me on to other bands and to slow down my speeds the speed test pings are in the 180s, and they over amplify the sinr it was normaly 12-15 for days and they overload it to 40 to make it slower i was getting 168 mbps downloads maxed with a 70 ping, now its around 100mbps but the pings and the TCP ramp up change really degrade quality of the internet
the cell tower on cellmapper.net is eNB ID 102529 - LTE
i have included a image of the signal quality changed by someone before and after a reboot
I am curious as to how you determined “they” changed the way TCP is functioning. Are you taking packet captures and doing in depth packet analysis of the sessions or what? The normal operations with TCP when there is congestion is to use a sliding window dynamically, making it smaller or larger when possible. The two hosts in the session communicate back and forth to optimize the flow based upon how much data can be received before the acknowledgement. It is a function of the TCP/IP stack operation based upon IEEE standards. The two end devices determine the window size. That is normal.
“One thing they did was change the TCP so any new connection starts out with a very slow download and then it ramps up to better speeds” (the two devices in the session)=they
I would guess that there was some congestion and the routers and the hosts in the session were compensating with a dynamic window size to address the congestion. Analysis of the packets in a capture should provide some profile of the traffic. I am just going to make a guess you are in an urban area and not a rural location. Please correct me if I am wrong. What I have observed from community conversations that users on the n41 band in a urban area where tower density is greater tend to see more congestion due to more users and I personally think the cells are loaded heavy. I could be wrong as I have no way to know how many subscribers are loaded on a given n41 cell source but I have seen more people in urban areas complaining about congestion. I live in a rural area with n41 and my downloads can be 250-400 MBps and varies depending upon the time of day/night and load.
The cellular signal metrics you have posted on the middle capture are pretty awesome. The RSRP, reference signal receive power for both the 4G LTE and the 5G frequencies is in the good & excellent range and the SINR is excellent as well. The RSRQ, reference signal quality for the LTE is good and the RSRQ for the 5G is excellent.
The signal quality in all three captures is good to excellent and the SINR though down on the middle capture after the gateway was restarted is still in the excellent range. So the metrics changed but the reason appears to be maybe not what is speculated. More data is needed to reveal a better picture. IF you notice on the first and last the 4G LTE band is a B66 frequency. The middle capture reflects B71 on the 4G LTE. So, I would expect the upload traffic to be a bit lower. It is not possible to tell with just the cellular metrics for the signal power, quality, and signal to noise ratio IF the gateway acquired a lock on a different cell or not as there are NO PCI values for the specific 4G LTE signals nor the 5G signals.
I can only speculate that there might be another tower or cells that the gateway gets a lock on and makes a transition from one cell to another on both the 4G LTE and the 5G NR frequencies. If you see the significant cellular metrics change AND the bands change look for the PCI, physical cell identifier, and see if you can locate both cell sources. Goto CellMapper.net and confirm both 4G and 5G cells. It might be that your location is between the cell sources and makes a transition in part due to the RSRP change and congestion. I am not sure how far your location is from the n41 source but 168 MBps down with n41 is not that impressive. Clearly the b71 and b66 are two different 4G LTE cells. Confirm the n41 5G cells as those will do the heavy lifting for the download traffic. A little more digging suggested.
The signal quality as you put it is not changed by someone. It is changed by environmental factors and cell source lock more than likely. Just like mobile handsets can jump from one cellular source to another the gateway might be making the same signal handoff or transition.
- iTinkeralotBandwidth Buff
I am curious as to how you determined “they” changed the way TCP is functioning. Are you taking packet captures and doing in depth packet analysis of the sessions or what? The normal operations with TCP when there is congestion is to use a sliding window dynamically, making it smaller or larger when possible. The two hosts in the session communicate back and forth to optimize the flow based upon how much data can be received before the acknowledgement. It is a function of the TCP/IP stack operation based upon IEEE standards. The two end devices determine the window size. That is normal.
“One thing they did was change the TCP so any new connection starts out with a very slow download and then it ramps up to better speeds” (the two devices in the session)=they
I would guess that there was some congestion and the routers and the hosts in the session were compensating with a dynamic window size to address the congestion. Analysis of the packets in a capture should provide some profile of the traffic. I am just going to make a guess you are in an urban area and not a rural location. Please correct me if I am wrong. What I have observed from community conversations that users on the n41 band in a urban area where tower density is greater tend to see more congestion due to more users and I personally think the cells are loaded heavy. I could be wrong as I have no way to know how many subscribers are loaded on a given n41 cell source but I have seen more people in urban areas complaining about congestion. I live in a rural area with n41 and my downloads can be 250-400 MBps and varies depending upon the time of day/night and load.
The cellular signal metrics you have posted on the middle capture are pretty awesome. The RSRP, reference signal receive power for both the 4G LTE and the 5G frequencies is in the good & excellent range and the SINR is excellent as well. The RSRQ, reference signal quality for the LTE is good and the RSRQ for the 5G is excellent.
The signal quality in all three captures is good to excellent and the SINR though down on the middle capture after the gateway was restarted is still in the excellent range. So the metrics changed but the reason appears to be maybe not what is speculated. More data is needed to reveal a better picture. IF you notice on the first and last the 4G LTE band is a B66 frequency. The middle capture reflects B71 on the 4G LTE. So, I would expect the upload traffic to be a bit lower. It is not possible to tell with just the cellular metrics for the signal power, quality, and signal to noise ratio IF the gateway acquired a lock on a different cell or not as there are NO PCI values for the specific 4G LTE signals nor the 5G signals.
I can only speculate that there might be another tower or cells that the gateway gets a lock on and makes a transition from one cell to another on both the 4G LTE and the 5G NR frequencies. If you see the significant cellular metrics change AND the bands change look for the PCI, physical cell identifier, and see if you can locate both cell sources. Goto CellMapper.net and confirm both 4G and 5G cells. It might be that your location is between the cell sources and makes a transition in part due to the RSRP change and congestion. I am not sure how far your location is from the n41 source but 168 MBps down with n41 is not that impressive. Clearly the b71 and b66 are two different 4G LTE cells. Confirm the n41 5G cells as those will do the heavy lifting for the download traffic. A little more digging suggested.
The signal quality as you put it is not changed by someone. It is changed by environmental factors and cell source lock more than likely. Just like mobile handsets can jump from one cellular source to another the gateway might be making the same signal handoff or transition.
- iTinkeralotBandwidth Buff
- iTinkeralotBandwidth Buff
I agree the http or https websites should load and render pretty quickly, but i am not convinced the theory about the router being over amplified as the problem. Sorry, I am not trying to argue with you it is just a difference of opinion. My logic is pretty simple. The cellular metrics represent the radio signal strength, quality, signal to noise ratio and the bands for the transmissions. In your original post the 4G LTE signal went from b66 to b71 and back to b66. The more recent one also reflects b66 as the signal so the frequency from that cell should be as before, assuming it is the same LTE cell source. Without the PCI you cannot tell. The original metrics you posted reflected the gateway had received 5G on the n41 frequency which is the midband millimeter frequency. The current metrics reflect the 5G signal is from a different cell that is n71 so that is a lower gigahertz frequency. So, we are not always comparing apples to apples here. Another factor to consider is the distance to the cell source and external influences that can result in the signal being degraded in flight. The reference signal receive power for the b66 is a bit poor and is a cell edge strength so I would expect that to be slower. The reference signal receive power for the n71 at -87 is good and the reference signal receive quality at -11 is also good. The signal to noise ratio for both signals are also good so there is nothing about those recorded values that suggest the gateway is being over amplified.
Just for comparison:
My 4G LTE is a bit stronger signal than yours and it is a B2 signal. (different frequency)
The 5G I receive is a n41 and it is NOW -10 dBm less than the n71 I received for over a year.
The signal to noise ratio went up recently, and signal quality went up. Performance went up too after T-Mobile turned on an n41 cellular source which appears to have replaced the n71 we had.
This test was ran just before 5 pm EST and that is with n41. With the prior n71 5G the best I would receive for download testing would be ~185 Mbps. This us using the same reference server every time. I set the preferred server to test against and I run the test from my MacBook Pro on a wired Ethernet connection over the Nokia gateway I have had since January 2021.
You might have a gateway that is just not working very well. You don’t report which model it is but given you are taking the cellular metrics from an Android phone with the mobile application I can only speculate that it is either the Arcadyan or the newer Sagemcon gateway. Maybe the gateway is a lemon? It does happen. Maybe the cells the gateway connects with are having heavy traffic loading and as a result more bandwidth throttling which results in the lag and delays. I tend to believe the problem is possibly in the cellular tower or the backhaul and traffic prioritization. It could be the gateway is not behaving well. Sorry I just don’t buy the over amplified theory. Nothing in the data I have seen supports such a theory. Any cellular operator has to follow strict guidelines when transmitting their frequencies. The reference signal receive power RSRP values do not suggest anything out of the ordinary.
- greg21111Roaming Rookie
for anyone who wants to test actual loading of websites and not just some speedtest muti connection test of bandwidth you never get use go to https://speed.cloudflare.com/ and it will show you how fast your internet really is for small files which is the exact problem i was having but i couldn’t show it, i knew i would figure it out.
I have been doing screen recordings of when it gets slow along with a active speed meter on screen and then i would show the file size and directly do a speedtest right after
here are my awful results i have been dealing with after day 3 (oct 20th) of home internet, you can see for 100k and 1MB downloads how bad they are and the M for the results at bottom is in Mbit
- iTinkeralotBandwidth Buff
I went to CellMapper and looked up the eNB ID you listed and the gNB ID 2161772 - NR tower is the only n41 I can see recorded in the area which appears to be down east of WA state community college and west of 177. That sort of makes sense given T-Mobile tends to deploy the n41 cells closer to colleges. Well, here that is the case. That isn’t to say that is the only n41 cell in the area but that seems to be the only one that is recorded in their database. There appear to be a fair number of n71 cells deployed about the area so T-Mobile was clearly pushing for coverage when they initially deployed the 5G in the area. They could probably use a few more n41 cells in the area. That would probably account for more congestion. Insufficient bandwidth and too many users on the n41. Keep in mind when there are phone connections on the cell they are prioritized better than the home internet gateway connections so yes it seems congestion could well be related.
If you can get T-Mobile support to provide you with the location of the cell your gateway connects to for the 5G I would suggest to find out. Do watch the PCI values as CellMapper.net may not have the information in the database but if you have an Android phone you can download the CellMapper mobile application and then with an account you can upload the information to the CellMapper servers. IF more people would upload the 5G cell information into the CellMapper database it would be helpful. The unfortunate thing is there is no iPhone compatible CellMapper application. I believe this is due to how Apple makes access to the cellular info more restricted.
- greg21111Roaming Rookie
below are settings the router auto applied two hours later last night to show how much the router is being over amplified and this morning with these same settings i tested 191 mbps and 70 ping on speed test and then later in the day i was having enormous processing buffering lag like i type in gmail.com or google.com and there was a 7-10 seconds for the request to go through, and those are text pages that should load in a fraction of a second, the same thing happened for all apps that have web components later in the day steam, adobe, apple store, ubisoft, i click a link and there is a good over 5 seconds before anything happened to the pages so much so i thought it wasn’t online but it finally loaded and i did a speed test right after of 140 and 80 ping, but those apps still didnt load correctly and now its back to 190 ping and like 90Mbps but it doesnt feel like it, which most like means its on over amplified band again.
and i do have data speed trackers that live in task bar, or top menu bar or by them selfs where i see next to no data as a page is loaded then it slowly ramps up over maybe 20seconds or as much as 1min to hit full speed if a page allows it, also i have a program that shows graphs so u see line charts of how the data is loading in and normally it would start really high at say 3-5MB and then go 8MB and then 15MB or max connection over maybe 3-5 seconds not slowly trickle up over a minute
- greg21111Roaming Rookie
i have the 3rd gen router Sagemcon, who would i inform if the n41 was being pushed to hard to my gateway because 40 is insane and it was getting maybe 11 the day before when it would connect to n41, and also when i saw the 5g being pushed to 40 is the first time i ever saw the 4g bands not being close to 20 SNR let alone being negative numbers
the delay for website loading is crazy i can look at website with alot of images and it wont break 500k downloading scrolling down an endless page it finally ramps up after time but its very brief it feels like shady stuff is going on behind the scenes but it wont report on a speed test because it does trigger the full download buffer or whatever it is to use more data
n71 which is what i prefer because of ping times and it seems better for my location, but i dont know how to lock it if i can
- iTinkeralotBandwidth Buff
So you recorded the SINR = 40 value reported from two of the first cellular metric recordings. I think one thing to consider or question to ask is, how accurate is the recording/reporting device or application that is being used? If the values reported are truly accurate or not we cannot be sure. It may depend upon the software and the interpretation of the available information. One thing I learned long ago is to always question the data and not take it at face value. I find the mobile application to be buggy and often will hang up and fail on my iPhone 12 Pro to the point that I have had to reboot my phone because the app crashed and hung everything up. Do I trust that application. No I do not.
Below according to XYXEL Networks
SINR
SINR (Signal Interference + Noise Ratio) also called CINR (Carrier to Interference + Noise Ratio) is the ratio of the signal level to the noise level (or simply the signal-to-noise ratio). The SINR value is measured in dB (dB). It's simple: the higher the value, the better the signal quality. At SINR values below 0, the connection speed will be very low, since This means that there is more noise in the received signal than the useful part, and the probability of losing a connection also exists.
Your most recent record of the n71 cellular metrics reflects the SINR = 14 which is good but not excellent. Still a pretty clean signal at 14. I have to question the n41 reporting a value of 40 for accuracy as i don’t trust the TMO mobile application nor am I very impressed with the Arcadyan or Sagemcon gateways. The Nokia also has a few warts but I have grown fond of mine after reading various conversations about the warts of the other two gateways.
WE, you, me or any other subscriber does not have control to get a lock on a given cell. You can contact support and IF you get with a really good support engineer that listens to what you are asking and also will take the request up the chain to a higher level support engineer then you might be able to get them to move the preference or lock to one of the n71 towers. They can see where your gateway gets a signal source and can preference the signal source IF they will. I have seen conversations where users have had success with this. I think, in part, it is holding a respectful, polite conversation with the engineer and requesting the connection preference/lock where you have seen it provide a respectable level of service. Please and thank you are always appreciated.
I worked as an enterprise network escalation support engineer for 22 years. Our team dealt with what we referred to as the ugly 5%. I know how important it is to take care with the conversation and work for a mutual outcome. If you are respectful it is usually more helpful. Sometimes you have to ask to talk to someone else. A support engineer that is treated with mutual respect is more likely to work harder for you. I have dealt with many very difficult, demanding customers but also some that were just elated I had flown in to work out the problems. I always appreciated the forgiving customers even though they could still be demanding. They had to be as they had superiors demanding answers from them. In the end when the problem is resolved it is a good feeling of achievement.
Until the n41 cells are properly vetted the n71 may be a more comfortable place. From January 2021 to March-April of 2022 the n71 cell we can use was unstable only twice about 3-4 days each time. When T-Mobile recently moved us onto an n41 cell I was instantly apprehensive as I had seen numerous community conversations about problems when on the n41 cell source. So far in this location the n41 has been just as solid as the n71 and shows no symptoms of congestion. I really suspect that what you were seeing was just that. It could be the prioritization of the traffic on the n41 cells in the area or some other configurations need to be improved to make it sing. You will not get as much potential for blazing speeds out of the n71 but it may be more stable and provide a solid average for speeds. I would take stable over flashy fast when the flashy fast is unreliable. Could the problem be buffer bloat? Maybe.
If the community is hungry for lower cost internet and users are flocking to T-Mobile for the home internet solution in the area then they may be selling more subscriptions than the current infrastructure can properly handle. The 5G internet solutions are still maturing and cellular delivery is not a simple venture. As you stand up cells you also have to make sure the back haul can handle the end user demands. If the upstream pipes cannot carry the volume of course you have issues. Then more investment must be made and well timelines and plans are all good but not always properly executed.
- greg21111Roaming Rookie
i ran a pinging app to see if that is why my webbrowsing loading speeds are so bad, and the packet loss is really high in the 80-90% lost.
does this have something to do with how there is such a slow ramp up for downloads on everything? i read people say its TCP bufferbloat or tcp slow ramp up, which they do to reduce bandwidth, but im not an networking expert and it feels illegal and a scam, you get sold a certain amount of speed a website should use that amount if its able to especially if you are on in the early hours of the morning when everyone is asleep and no one on network.
- iTinkeralotBandwidth Buff
The excessive latency of ICMP packets and loss appears to be due to T-Mobile throttling them. It is probably not related to the slow ramp up of the sessions. The throttle of ICMP also breaks trace routing which is also a nuisance as both are important for troubleshooting. They are exercising excessive controls and it is tarnishing the solution.
mid you run a Wireshark packet capture while executing a download or other activities and then run the expert analysis it should provide some clarity on the behavior. If you have never run a packet capture with Wireshark it is a little bit of a learning curve. Understanding packet structure and protocols and how TCP/IP functions is also rather important but that can be learned. Research on the web can provide a great deal of information and tutorials. It is a huge time sink. Been there and done that. I had my CCNP certification for years and had to test for recertification every 2-3 years. It was a huge pain but kept us current and on top of our game.
Related Content
- 11 months ago
- 3 months ago
- 2 years ago
- 4 years ago