Mobile network quality: What is the truth?

[jmak]

The reality is that people do use their phones while on the move - more than 5% of calls that you estimate, as when travelling by car, bus, taxi or train to work and back is a good time to catch up on calls that cannot be made at work. By the way a call made in a car could be made on a hands free device by the driver or by passengers so your logic does not apply.

Apology for not extending my reference from handset to include car-kits but I am seeing fewer and fewer of those and felt they are not really relevant anymore. Most people go for bluetooth headsets or wired headsets that comes with the phones. But you are right, other hands free kits exist. I will also concede that I did not include all alternative transport mediums.

I don't know anyone that actually keep their personal calls for when they are in the car driving around, but maybe I know the w

Moving tests are an important measure of hand over between towers. Seems you have a vested interest as you know how bad the handover failure rates are.

I agree, mobile tests are important for testing handovers. But for the purpose of checking network quality on average for all consumers you can't check the quality of a small geographic area and for a use-case that is covering such a small percentage of usage. This is like doing a HIV sample test on homeless people in a specific township and using that as a number statistic for South Africa.

It is not a vested interest, just looking at the bigger picture and maybe I have access to a bit more information than the average Joe

Its not that straight forward. While you may be sitting in your office and making a call, there are many others who make calls while driving. The scenario where you are not moving is pretty simple in terms of how the NW works, basically you will be attached to one tower and once the call is setup and connected there should be no way in which the call can drop. So what a stationary call tests is actually the CSSR and not the rest of the important KPIs.

Some of the reasons why you might drop a call while stationary(which I can quickly think of):
1. Fading
2. Cell breathing (and resulting HO to 2G if coverage and capacity available)
3. RF attenuation due to change in cellphone orientation and hand-position (think polarisation and attenuation caused by you hand and head, as well as people walking by)
4. Handover (better condition, capacity)

Drive tests on the other hand access the overall quality of the NW, the accessability and retainability over many sites. It also allows the operators/vendors to pick up any discrepancies in the setup/configuration of the site. It is overall a much better test of the NW.

As mentioned above, it is definitely an important test, and as you say it will help identify possible configuration issues. But it remains a much smaller percentage of traffic on a network and is not a good measure to pin a networks performance on since it measures a small portion of network traffic. So I would still disagree on driving test being a better test for a network wide KPI.

 

[Mike Hoxbig]

The truth is that every 2nd or 3rd call of mine on MTN is either dropped midway through conversation, or it fails to connect in the first place.

Vodacom is even worse, because I just seem to get poor coverage with them. Maybe others are more fortunate with them.

 

[Electron1]

Thanks @jmak, nice reply!

I do agree that a much wider sample be taken with better defined testing procedure and it should be countrywide - perhaps even showing major metropolitan areas and rural areas separately.
I do still believe that moving tests should be part of the criteria - I am in my car on average about 2 hours a day travelling to clients and this is when I catch up on calls (on a hands free!) and experience dropped calls regularly.

 

[BrandOfRiva]

Wonder if any of these networks are using the stats they have to see where/when the dropped calls happen? A very basic one would be repeat call backs one after the other between two numbers in succession. Not hard at all and if run over time can show a few useful trends.

 

[SteveO]

Some of the reasons why you might drop a call while stationary(which I can quickly think of):
1. Fading
2. Cell breathing (and resulting HO to 2G if coverage and capacity available)
3. RF attenuation due to change in cellphone orientation and hand-position (think polarisation and attenuation caused by you hand and head, as well as people walking by)
4. Handover (better condition, capacity)



As mentioned above, it is definitely an important test, and as you say it will help identify possible configuration issues. But it remains a much smaller percentage of traffic on a network and is not a good measure to pin a networks performance on since it measures a small portion of network traffic. So I would still disagree on driving test being a better test for a network wide KPI.

All your drop reasons shouldnt really be effecting the call, Fading for instance is only caused by moving, the RAKE receiver in the UE will counter any Fading due to multiple signal paths.Cell Breathing will be countered by power management (if the UE follows the 3GPP standards for power leveling, which the iPhone struggles to do) Attenuation due to orientation is not really an issue, body loss is already 3dB and any resultant loss should be less than that. Handover will only occur if the signal is bad enough to be in the HO parameters set by the operator or vendor, the call will not setup on poor signal, it will instead drop to 2G before call setup.

I do agree that testing a small portion of NW traffic is not the way to go, however 1 drive test should cover around 16 sites, this should be repeated to cover a spread of sites.

In my experience, working for a vendor, we generally test every site in the NW once they go live, then we use performance parameters to continue to monitor these sites and the overall network. This doesnt give you an overview of the user experience however and this is where drive tests are a better measurement as you as the test engineer become the user and can thus pick up some of the issues the user will face.

 

[Bern]

ICASA needs to have a more clear testing methodology here imo which includes multiple area's, and considerably more calls.

Either way - the sample size is far too small.

Agreed, state the testing methodology, the equipment, etc and then make sure the sample has enough calls in multiple areas at stated time windows to get a proper statistic.

Well, even if ICASA test 10 000 calls nation wide and the operators fail, what are they gonna do about it?

Yes, that is actually a bit more of a problem - what exactly happens if VC or MTN just say no, ignore it or try smother it in litigation?

I cannot remember when last I had a dropped call, then again, I'm never on a call while mobile. If you are in a fixed place like outside or in a building, dropped calls should not happen. If you are moving around, especially in a building the signal strength can be effected by many things out of the control of the operator. So, rule of thumb, want to make or take a call, relax, go take a seat, have a cuppa. Life’s too short to be rushing from one place to the other. If driving, turn onto silent, put some music on, calls can wait till you are back in the office. No dropped calls and not frustration.

I think you completely miss the whole point of mobile telephony - why don't you just stick with a Telkom line then? The whole point of cellular technology is to allow MOBILE (as in on the move) calls. Operators claim that they achieve this, but in actual fact don't which is the issue. If there is a technological restraint then why did I not get so many call failures and dropped calls on GSM (same technology) 5 years ago while mobile?

I get PLENTY of dropped and failed calls now when I am stationary in many different locations, so it is not specifically a problem related to being mobile.

So in your opinion the 1-5% of calls made while driving is much more important for a network than the other 95%-99% of calls made while sitting and walking? Or using your words, 95%-99% of consumers are useless compared to the people driving around of which probably 75% are anyway illegal since they are not using a headset?

The acid test for a cellular network is its performance of someone who is mobile, that is what they are designed to allow (as in changing signal and handovers from tower to tower), so a mobile test is the best way to measure if the network is performing properly. Otherwise VC/MTN/etc could just go set up a stationary test with a nice yagi antenna and get brilliant results....

 

[jmak]

All your drop reasons shouldnt really be effecting the call, Fading for instance is only caused by moving, the RAKE receiver in the UE will counter any Fading due to multiple signal paths.Cell Breathing will be countered by power management (if the UE follows the 3GPP standards for power leveling, which the iPhone struggles to do) Attenuation due to orientation is not really an issue, body loss is already 3dB and any resultant loss should be less than that. Handover will only occur if the signal is bad enough to be in the HO parameters set by the operator or vendor, the call will not setup on poor signal, it will instead drop to 2G before call setup.

I do agree that testing a small portion of NW traffic is not the way to go, however 1 drive test should cover around 16 sites, this should be repeated to cover a spread of sites.

In my experience, working for a vendor, we generally test every site in the NW once they go live, then we use performance parameters to continue to monitor these sites and the overall network. This doesnt give you an overview of the user experience however and this is where drive tests are a better measurement as you as the test engineer become the user and can thus pick up some of the issues the user will face.

Seems like I should not have used stationary as antonym to drive-test, we are starting to go past each other Should rather have used "low mobility" in place of the more common word "stationary"; then we might have understood each other better...

For the rest I say lets leave it where it is. This will go into nitpicking about technical specs which will help no-one and bring us nowhere since it will end up in opinions. It seems we have similar opinions in similar environments but still have different methodologies in approaching and solving problems, probably due to different positions in the same sector. But hey, 2 different approaches can only have a better outcome than a single approach, can't it?!.

 

[Bosvark@]

A quick cell connection on Vodacom is almost impossible unless you sitting below the tower...