Uncapped wireless vs fibre - Bandwidth comparison

[Jamie McKane]

Uncapped wireless vs fibre - Bandwidth comparison

The MyBroadband South African ISP rankings for July 2020 recently showed that Rain has replaced Telkom as the lowest-ranked Internet service provider (ISP) in the country.

Subscribers using Rain's uncapped LTE and 5G services continue to complain about unreliable network performance, saying that at certain times of day they struggle to use the Internet at all.

 

[Goosfrabba]

Jip, Rain is garbage. But fibre is on the way fortunately

 

[system32]

Makes complete technical sense, EXCEPT that Cellular (it's right there in the name), implies more than one cell.
Each cell has bandwidth.
Out of capacity in a cell - add another cell.

The problem in ZA, is that most areas aren't covered by fibre and "up to" ADSL just don't cope in Web 2.0 so we forced to use Cellular.

 

[Mekon]

Rain was ok when we first got it just before lockdown as my daughter needed to work from home and ADSL was not an option anymore. But she battles some days and her boss thinks she's buggering around. Communicating human to human with Rain is just not a thing with them so we constantly have to wait for some sort of feedback in the form of an email which does not happen. We stay on a small holding so fibre might never happen here.

 

[Jan]

Each cell has bandwidth.
Out of capacity in a cell - add another cell.

It is indeed not a straight-forward comparison. But there is no way you can densify a network enough to get to 10THz of bandwidth, and certainly no single subscriber would be given all that bandwidth even if they could.

An interesting one I left out of this comparison is ADSL. It's actually quite spectrally efficient if you look at how much throughput you can squeeze out of it compared to how much bandwidth is available on the copper cable.

 

[Geoff.D]

Uncapped wireless vs fibre - Bandwidth comparison

The MyBroadband South African ISP rankings for July 2020 recently showed that Rain has replaced Telkom as the lowest-ranked Internet service provider (ISP) in the country.

Subscribers using Rain's uncapped LTE and 5G services continue to complain about unreliable network performance, saying that at certain times of day they struggle to use the Internet at all.

There is quite a bit that is plain and simply wrong about this article that I do not know where to start.

Be that as it may -- it is the modern way of doing things these days - use the wrong words and use words inter-changeably because someone looks in a thesaurus.

The main message in the article is that wireless networks are bandwidth constrained and can never compete with wireline fibre ever. Fibre and whatever replaces fibre one day, will always be ahead in the race.

That is why using wireless, cellular, mobile, call it what you like for fixed applications is a fundamentally failed strategy. The only variable is when, how soon and how will the network fail to deliver.

When the service first launches and the network is empty, subscribers are extremely pleased with their purchase.

However, as more subscribers buy the service and climb onto the network, performance degrades. In the worst cases, the network becomes unusably slow.

The simple fact is the bandwidth available in the part of the radio spectrum available to the mobile network operators is not enough to handle the volumes of simultaneous connections/sessions that users require at the data rates they are told they will get by default.

 

[Jan]

Be that as it may -- it is the modern way of doing things these days - use the wrong words and use words inter-changeably because someone looks in a thesaurus.

Thanks for the criticism, I suppose.

Please understand that I'm trying to serve South Africa's technical community, while making technical topics from one field accessible to someone who might be working in a different even unrelated technical field.

Ideally, I'd like to be able to communicate in a way that even a non-technical person who is interested in the topic would be able to follow along.

All of this needs to be achieved without writing a page-long glossary, explainer for jargon, and list of abbreviations for every article.

At the very least you got exactly what I was trying to demonstrate with this article, so I'm on my way to win .

 

[konfab]

Uncapped wireless vs fibre - Bandwidth comparison

The MyBroadband South African ISP rankings for July 2020 recently showed that Rain has replaced Telkom as the lowest-ranked Internet service provider (ISP) in the country.

Subscribers using Rain's uncapped LTE and 5G services continue to complain about unreliable network performance, saying that at certain times of day they struggle to use the Internet at all.

My fibre line has been far more unreliable than my old LTE connection.

 

[Geoff.D]

Jes, unfortunately, we are all stuck between a hard place and a rock. The wrong terminology gets used so often that it becomes the norm, and the only thing understood by the masses.

Talking about C-, L-, and O-band is pushing the limits.

 

[Sapphiron]

Uncapped Wireless can be a lot more reliable, provided some key changes are made to retain quality

1. double to triple the price, so you don't have to oversubscribe towers in peak time to be profitable and so you can fund setting up additional towers when capacity gets low.
2. cap the throughput aggressively, and price them separately. Say 5Mbit, 10mbit and 20Mbit packages.

Most of the problems with wireless connections is it is being sold as an "up to" service, that is contrary to providing a quality service.

 

[Geoff.D]

Uncapped Wireless can be a lot more reliable, provided some key changes are made to retain quality

1. double to triple the price, so you don't have to oversubscribe towers in peak time to be profitable and so you can fund setting up additional towers when capacity gets low.
2. cap the throughput aggressively, and price them separately. Say 5Mbit, 10mbit and 20Mbit packages.

Most of the problems with wireless connections is it is being sold as an "up to" service, that is contrary to providing a quality service.

As Jan pointed out in his comment about ADSL, the rot set in there already. The "up to" terminology buggered up the concept of ADSL so badly that ADSL never really got used to its full potential.
(except for a handful of us who knew how to exploit what was available to the maximum). And that is the same group are so unhappy with the cavalier way in which OpenServe is dumping the technology without ensuring a reliable alternative is available in an area before dumping it.

A reliable, albeit limited, stable and reliable ADSL service is a far better bet than an unreliable congested wireless service any day.
But no, I do not agree with you that the way to do this was to price any service so high to discourage oversubscription, is the answer.

and BTW it is not about "towers" it is about base stations or network nodes. - Another example of using the wrong word because it is assumed the masses out there will not know what we are talking about. Just as they have learnt to associate a "tower" with a network node or base station, they could just as easily be taught the correct words.

 

[Johand]

Makes complete technical sense, EXCEPT that Cellular (it's right there in the name), implies more than one cell.
Each cell has bandwidth.
Out of capacity in a cell - add another cell.

There are limits to the number of cells you can add because cells need to not interfere with each other. Anyway at some point you are essentially adding a single cell per house with fiber running to it The future is fiber and the question just is how close to your house it will get before you switch to wireless technology (WiFi/4G/5G)!

Vumatel provides a single fiber to each house (AON) in their trenched deployments so fiber will just smash wireless in the long term.

What would have been nice in the article is to add Shannon's limit (https://en.wikipedia.org/wiki/Shannon–Hartley_theorem) which describes the maximum theoretical bandwidth (in bits/second) for a slice of bandwidth. We are already close to the theoretical limit using turbo-codes/LDPC on wireless so technology won't increase transmission speed for the same spectrum. With fiber we have MILES and MILES to go with the technology because we don't have the technology yet to encode light in the same way and efficiency as radio waves.



C = channel capacity (bits/second)
B = bandwidth
S = Signal Power
N = Noise

Increasing S means more capacity but bigger cells (less geographical sharing of spectrum) and more N (noise) from neighboring cells. Shrinking cells means reducing S and also capacity (but sharing spectrum with non-neighboring cells, or neighboring cells using directional antennas).

The best way to increase capacity is by having a larger B e.g. more spectrum.

 

[Geoff.D]

More bandwidth ...... in the form of more bands in the radio spectrum.

 

[Geoff.D]

There are limits to the number of cells you can add because cells need to not interfere with each other. Anyway at some point you are essentially adding a single cell per house with fiber running to it The future is fiber and the question just is how close to your house it will get before you switch to wireless technology (WiFi/4G/5G)!

The one and only formula that counts for anything in this game.

Nothing like a thorough understanding of the basics to hammer home the realities of network service provision.

When did Hartley and Shannon come up with this very simple concept? (You cant leave out Nyquist).

Answer: In the 1920's.

Pity all those Rain guys don't go back and read a few old textbooks before they go off half-cocked about their latest test of the basic rules hey. But then we have just had a discussion about the re-inventing of the wheel in another unrelated thread.

In other words, the only effective way to utilise radio technologies is for it to be used for that last inch between the network node (connected to the world-wide network by fibre) and the device.
And right now, plain old simple Wi-Fi is more than adequate for that task.

Take all the billions spent on complicated mobile networks and plough that into the ubiquitous delivery of fibre infrastructure and we all win.

 

[Jan]

2. cap the throughput aggressively, and price them separately. Say 5Mbit, 10mbit and 20Mbit packages.

Something interesting happening in 5G standalone is a feature they are calling "network slicing", which seems to basically come down to a kind of virtualisation of the network that can happen all the way down to the radio access network.

Provided this isn't all hype, OEMs are promising the ability for high-end clients to even buy their own dedicated carrier wave.

For regular home use, they are pushing the idea of a kind of 5G ISP that will buy a network slice from the Vodacom's, MTN's, and Rains of the world that is then specially configured for a specific use-case.

So you go to an ISP that sells a "5G for gamers" product which then guarantees a certain amount of throughput and latency, at a higher price.

This isn't really possible to do on 5G non-standalone, 4G, and earlier generation networks for the reasons @Geoff.D mentioned earlier.

So yeah... *maybe* with 5G we'll see the kind of thing you're talking about.

 

[microsoftword]

Oof sounds like South African wireless carriers are still in 2007.

 

[quovadis]

The only way for wireless to compete with fibre is densification of cells using the same spectrum allocation more efficiently or higher frequency allowing larger throughput using lower power to ensure minimal interference between neighbouring cells. It's how 5G is supposed to work anyway to deliver the promised latency and throughput. Ultimately though wireless will never prevail vs fibre as bandwidth requirements increase exponentially.

 

[Johand]

Oof sounds like South African wireless carriers are still in 2007.

Nah - overpromising and underdelivering is an international trend in mobile networks. Everything works fine in the beginning of the rollout and as adoption goes up quality craters.

 

[Jet-Fighter7700]

Quite a interesting article,
But like everything, as technology improves, service should improve.

Problem is we live in SA where mediocrity in everything is the national motto.

 

[Swa]

Makes complete technical sense, EXCEPT that Cellular (it's right there in the name), implies more than one cell.
Each cell has bandwidth.
Out of capacity in a cell - add another cell.

The problem in ZA, is that most areas aren't covered by fibre and "up to" ADSL just don't cope in Web 2.0 so we forced to use Cellular.

This here. We keep hearing about how spectrum is a problem when there are already solutions. When they eventually get the spectrum they'll complain they don't have the *base stations (*Geoff doesn't seem to like towers).

There are limits to the number of cells you can add because cells need to not interfere with each other. Anyway at some point you are essentially adding a single cell per house with fiber running to it The future is fiber and the question just is how close to your house it will get before you switch to wireless technology (WiFi/4G/5G)!

Vumatel provides a single fiber to each house (AON) in their trenched deployments so fiber will just smash wireless in the long term.

What would have been nice in the article is to add Shannon's limit (https://en.wikipedia.org/wiki/Shannon–Hartley_theorem) which describes the maximum theoretical bandwidth (in bits/second) for a slice of bandwidth. We are already close to the theoretical limit using turbo-codes/LDPC on wireless so technology won't increase transmission speed for the same spectrum. With fiber we have MILES and MILES to go with the technology because we don't have the technology yet to encode light in the same way and efficiency as radio waves.

View attachment 895358

C = channel capacity (bits/second)
B = bandwidth
S = Signal Power
N = Noise

Increasing S means more capacity but bigger cells (less geographical sharing of spectrum) and more N (noise) from neighboring cells. Shrinking cells means reducing S and also capacity (but sharing spectrum with non-neighboring cells, or neighboring cells using directional antennas).

The best way to increase capacity is by having a larger B e.g. more spectrum.

And one solution to that is to ensure two neighbouring cells don't use the same frequencies. Another is to reduce the power and range and then put cells closer together. But networks try to take shortcuts by using as much spectrum and power as possible which is the wrong approach. While this doesn't change we will keep having a spectrum shortage because it's being wasted.

Also Shannon's limit assumes that spectrum is used as a shared resource. Newer technologies take advantage of beam forming or interference patterns so two users can be using the same frequency.