Showing posts with label Data Speeds. Show all posts
Showing posts with label Data Speeds. Show all posts

Monday 23 May 2022

5G Reality Check - Data Rates

One of the common questions that we encounter is why are 5G speeds so low as we were promised 5G downlink speeds of 20 Gbps. Most people do not understand how the 5G speeds are calculated and what do they depend on. In many cases, the network won’t be capable of delivering higher speeds due to some or the other limitation. 

In a new presentation, I try to explain the theoretical speeds and compare them with real world 5G data rates and even try to map it to why these speeds are what they are. Hopefully people won't mind me adding some humour as I go along.

Video and Slides embedded below

Embedded below is the Twitter thread on Speedtests ðŸ˜‚

Related Posts

Wednesday 19 August 2020

Would 5G NSA undergo Sunset? When?


I have been thinking about the long term evolution of 5G and have now reached the conclusion that it would make sense in the long run to switch off non-standalone 5G. This would of course be only after 5G core has been tested and used extensively. Instead of writing my reasoning, here is a 10 minute video and the corresponding slides.





Let me know what you think in the comments below. If you agree, when do you think is the best time for 5G NSA Sunset?


Related Posts:

Friday 23 August 2019

The Politics of Standalone vs Non-Standalone 5G & 4G Speeds


A short video (and slides) discussing the operator dilemma of standalone (SA) vs non-standalone (NSA) 5G deployment, frequency refarming and why 4G speeds will start reducing once SA 5G starts to be deployed.

Video




Slides



Related Posts:

Thursday 18 July 2019

5G SpeedTests and Theoretical Max Speeds Calculations


Right now, Speed Tests are being described as 5G killer apps.



A good point by Benedict Evans



Everyone is excited and want to see how fast 5G networks can go. If you use Twitter, you will notice loads and loads of speed tests being done on 5G. An example can be seen above.


I recently heard Phil Sheppard, Director of Strategy & Architecture, '3 UK' speak about their 5G launch that is coming up soon. Phil clearly mentioned that because they have a lot more spectrum (see Operator Watch blog post here and here) in Capacity Layer, their 5G network would be faster than the other UK operators. He also provided rough real world Peak Speeds for Three and other operators as can be seen above. Of course the real world speeds greatly depend on what else is going on in the network and in the cell so this is just a guideline rather than actual advertised speeds.


I have explained multiple times that all 5G networks being rolled out today are Non-Stand Alone (NSA) 5G networks. If you don't know what SA and NSA 5G networks are, check this out. As you can see, the 5G NSA networks are actually 4G Carrier Aggregated Networks + 5G Carrier Aggregated Networks. Not all 4G spectrum will be usable in 5G networks but let's assume it is.

To calculate the theoretical maximum speed of 5G NSA networks, we can calculate the theoretical maximum 4G Network speeds + theoretical maximum 5G Network speeds.

I have looked at theoretical calculation of max LTE Carrier Aggregated Speeds here. Won't do calculation here but assuming 3CA for any network is quite possible.

I also looked at theoretical calculation of 5G FDD New Radio here but then found a website that helps with 5G NR calculation here.

If we calculate just the 5G part, looking at the picture from Three, we can see that they list BT/EE & O2 speeds as 0.61 Gbps or 610 Mbps, just for the 5G part.

Looking at the calculation, if we Input Theoretical max values in this equation:

Calculating just for DL

J - number of aggregated component carriers,
maximum number (3GPP 38.802): 16
input value: 1

v(j)Layers - maximum number of MIMO layers ,
3GPP 38.802: maximum 8 in DL, maximum 4 in UL
input value: 8

Q(j)m modulation order (3GPP 38.804)
For UL and DL Q(j)m is same (QPSK-2, 16QAM-4, 64QAM-6, 256QAM-8)
input value: 8 (256QAM)

f(j) Scaling factor (3GPP 38.306)
input value: 1

FR(j) Frequency Range 3GPP 38.104:
FR1 (450 MHz – 6000 MHz) и FR2 (24250 MHz – 52600 MHz)
input value: FR1

µ(j) -value of carrier configuration (3GPP 38.211)
For DL and UL µ(j) is same (µ(0)=15kHz, µ(1)=30kHz, µ(2)=60kHz, µ(3)=120kHz)
input value: 0 (15kHz)

BW(j)- band Bandwidth, MHz (3GPP 38.104),
should be selected with Frequency Range and µ(i) configuration:
input value: BW:40MHz FR1 µ:15kHz:

Enter a PRB value (if other)
default: 0

Rmax (if you don't know what is it, don't change)
Value depends on the type of coding from 3GPP 38.212
(For LDPC code maximum number is 948/1024 = 0.92578125)
default: 0.92578125

*** Only for TDD ***
Part of the Slots allocated for DL in TDD mode,
where 1 = 100% of Slots (3GPP 38.213, taking into account Flexible slots).
Calculated as: the number of time Slots for DL divided by 14
default value: 0.857142

Part of the Slots allocated for UL in TDD mode,
where 1 = 100% of Slots (3GPP 38.213, taking into account Flexible slots).
Calculated as: 1 minus number of Slots for DL
default value: 0.14285800000000004

Calculated 5G NR Throughput, Mbps: 1584


As you may have noticed, BTE/EE has 40 MHz spectrum while Vodafone in UK have 50 MHz of spectrum.

Changing
BW(j)- band Bandwidth, MHz (3GPP 38.104),
should be selected with Frequency Range and µ(i) configuration:
input value: BW:50MHz FR1 µ:15kHz:

Calculated 5G NR Throughput, Mbps: 1982

Now Three UK has 100 MHz, immediately available for use. So changing

µ(j) -value of carrier configuration (3GPP 38.211)
For DL and UL µ(j) is same (µ(0)=15kHz, µ(1)=30kHz, µ(2)=60kHz, µ(3)=120kHz)
input value: 1 (30kHz)

BW(j)- band Bandwidth, MHz (3GPP 38.104),
should be selected with Frequency Range and µ(i) configuration:
BW:100MHz FR1 µ:30kHz:


Calculated 5G NR Throughput, Mbps: 4006

In theory, a lot of speed is possible with the 100 MHz bandwidth that Three will be able to use. We will have to wait and see who can do a theoretical max SpeedTest. In the meantime remember that a 1Gbps speed test will use over 1 GB of data.



Related Posts:

Sunday 23 June 2019

Finland: A country with only Unlimited Data Plans


I was listening to Elisa couple of weeks back, at 5G World Summit. One of the things that surprised me was that Elisa offered unlimited data plans but the price varied based on the maximum speeds possible. The same approach was going to continue with 5G. When 5G data speeds would improve, new packages will be added with the improved speeds.


Tefficient has pointed out multiple times that even though all operators in Finland offer unlimited data plans, their ARPU has increased in 2018. This is in contrast to the other mature markets, even though they may not be offering unlimited data plans.


Same thing was pointed out by Rewheel research that highlighted in their May 2019 report that, "Finnish operators that executed ‘unlimited everything’ strategies were the undisputed champions of the 4G era"

A mashable article pointed out that "5G will be crazy fast, but it'll be worthless without unlimited data". This is very true.


Mobile operators should start thinking about how they can offer unlimited data plans, especially if they keep touting applications that are going to use loads of data. As you can see from the tweet above, a 1 hour 8K video streaming would roughly use between 7 - 10 GB of data.

Let me know your thoughts.

Sunday 10 February 2019

Theoretical Throughput Calculation of FDD 5G New Radio (NR)


A nice video by Peter Clarke on 5G NR throughput calculation for FDD. Right now it's only in the video form but will hopefully be available as a tool on his excellent website here. A tool for 4G throughput calculation is available here.




Related Links:

Sunday 11 June 2017

Theoretical calculation of EE's announcement for 429Mbps throughput


The CEO of UK mobile network operator EE recently announced on twitter that they have achieved 429 Mbps in live network. The following is from their press release:

EE, the UK’s largest mobile network operator and part of the BT Group, has switched on the next generation of its 4G+ network and demonstrated live download speeds of 429Mbps in Cardiff city centre using Sony’s Xperia XZ Premium, which launched on Friday 2 June. 
The state of the art network capability has been switched on in Cardiff and the Tech City area of London today. Birmingham, Manchester and Edinburgh city centres will have sites upgraded during 2017, and the capability will be built across central London. Peak speeds can be above 400Mbps with the right device, and customers connected to these sites should be able to consistently experience speeds above 50Mbps. 
Sony’s Xperia XZ Premium is the UK’s first ‘Cat 16’ smartphone optimised for the EE network, and EE is the only mobile network upgrading its sites to be able to support the new device’s unique upload and download capabilities. All devices on the EE network will benefit from the additional capacity and technology that EE is building into its network. 
... 
The sites that are capable of delivering these maximum speeds are equipped with 30MHz of 1800MHz spectrum, and 35MHz of 2.6GHz spectrum. The 1800MHz carriers are delivered using 4x4 MIMO, which sends and receives four signals instead of just two, making the spectrum up to twice as efficient. The sites also broadcast 4G using 256QAM, or Quadrature Amplitude Modulation, which increases the efficiency of the spectrum.

Before proceeding further you may want to check out my posts 'Gigabit LTE?' and 'New LTE UE Categories (Downlink & Uplink) in Release-13'

If you read the press release carefully, EE are now using 65MHz of spectrum for 4G. I wanted to provide a calculation for whats possible in theory with this much bandwidth.

Going back to basics (detailed calculation for basics in slideshare below), in LTE/LTE-A, the maximum bandwidth possible is 20MHz. Any more bandwidth can be used with Carrier Aggregation. So as per the EE announcement, its 20 + 10 MHz in 1800 band and 20 + 15 MHz in 2600 band

So for 1800 MHz band:

50 resource blocks (RBs) per 10MHZ, 150 for 30MHz.
Each RB has 12x7x2=168 symbols per millisecond in case of normal modulation support cyclic prefix (CP).
For 150 RBs, 150 x 168 = 25200 symbols per ms or 25,200,000 symbols per second. This can also be written as 25.2 Msps (Mega symbols per second)
256 QAM means 8 bits per symbol. So the calculation changes to 25.2 x 8 = 201.6 Mbps. Using 4 x 4 MIMO, 201.6 x 4 = 806.4Mbps
Removing 25% overhead which is used for signalling, this gives 604.80 Mbps


Repeating the same exercise for 35MHz of 2600 MHz band, with 2x2 MIMO and 256 QAM:

175 x 168 = 29400 symbols per ms or 29,400,000 symbols per second. This can be written as 29.4 Msps
29.4 x 8 = 235.2 Mbps
Using 2x2 MIMO, 235.2 x 2 = 470.4 Mbps
Removing 25% overhead which is used for signalling, this gives 352.80 Mbps

The combined theoretical throughput for above is 957.60 Mbps

For those interested in revisiting the basic LTE calculations, here is an interesting document:




Further reading:

Monday 9 November 2015

5G and Evolution of the Inter-connected Network


While there are many parameters to consider when designing the next generation network, speed is the simplest one to understand and sell to the end user.

Last week, I did a keynote at the International Telecom Sync Forum (ITSF) 2015. As an analyst keynote, I looked at how the networks are evolving and getting more complex, full of interesting options and features available for the operator to decide which ones to select.

There wont just be multiple generations of technologies existing at the same time but there will also be small cells based networks, macro networks, drones and balloons based networks and satellite based networks.

My presentation is embedded below. For any reason, if you want to download it, please fill the form at the bottom of this page and download.



Just after my keynote, I came across this news in Guardian about 'Alphabet and Facebook develop rival secret drone plans'; its an interesting read. As you may be aware Google is actively working with Sri Lanka and Indonesia for providing seamless internet access nationally.


It was nice to hear EE provide the second keynote which focused on 5G. I especially liked this slide which summarised their key 5G research areas. Their presentation is embedded below and available to download from slideshare.




The panel discussion was interesting as well. As the conference focused on timing and synchronisation, the questions were on those topics too. I have some of them below, interested to hear your thoughts:

  • Who cares about syncing the core? - Everything has moved to packets, the only reason for sync is to coordinate access points in wireless for higher level services. We have multiple options to sync the edge, why bother to sync the core at all?
  • We need synchronisation to improve the user’s experience right? - Given the ever improving quality of the time-bases embedded within equipment, what exactly would happen to the user experience if synchronisation collapsed… or is good sync all about operators experience?
  • IoT… and the impact on synchronisation- can we afford it? - M2M divisions of network operators make a very small fraction of the operator’s revenue, is that going to change and will it allow the required investment in sync technology that it might require?

Saturday 23 May 2015

The path from 4.5G to 5G

In the WiFi Global Congress last week, I heard this interesting talk from an ex-colleague who now works with Huawei. While there were a few interesting things, the one I want to highlight is 4.5G. The readers of this blog will remember that I introduced 4.5G back in June last year and followed it with another post in October when everyone else started using that term and making it complicated.

According to this presentation, 3GPP is looking to create a new brand from Release-13 that will supersede LTE-Advanced (LTE-A). Some of you may remember that the vendor/operator community tried this in the past by introducing LTE-B, LTE-C, etc. for the upcoming releases but they were slapped down by 3GPP. Huawei is calling this Release-13 as 4.5G but it would be re-branded based on what 3GPP comes up with.


Another interesting point are the data rates achieved in the labs, probably more than others. 10.32Gbps in sub-6GHz in a 200MHz bandwidth and 115.20Gbps using a 9.6GHz bandwidth in above 6GHz spectrum. The complete presentation as follows:



Another Huawei presentation that merits inclusion is the one from the last Cambridge Wireless Small Cells SIG event back in February by Egon Schulz. The presentation is embedded below but I want to highlight the different waveforms that being being looked at for 5G. In fact if someone has a list of the waveforms, please feel free to add it in comments


The above tweet from a recent IEEE event in Bangalore is another example of showing the research challenges in 5G, including the waveforms. The ones that I can obviously see from above is: FBMC, UFMC, GFDM, NOMA, SCMA, OFDM-opt, f-OFDM.

The presentation as follows:




Saturday 23 November 2013

Bandwidth is not the answer – it’s stationarity


Martin Geddes did an interesting presentation in Future of Broadband workshop. The ITU has the following write-up on that workshop

Eye-opening, evangelical and extremely well attended: this afternoon’s Future of Broadband workshop was all about exploding established concepts on how telcos should go about improving both customer experience and their bottom line.
Ranking broadband in terms of speed is the standard approach, but speed is not the only thing that matters in this business, according to Martin Geddes of Geddes Consulting, running the workshop in conjunction with Neill Davies of Predictable Network Solutions.  He illustrated his point with a series of examples drawn from customers accessing broadband at different speeds – but with unexpectedly different experiences.
Slower broadband, whether over cable, satellite or fibre, in many cases offered a better quality of customer experience than the faster variant. Why? Variability, or rather lack of variability, is the key. A stable service, even it is slower, enables POTS-quality VoiP, whereas a highly-variable, faster service delivers a less satisfactory customer experience – and, by definition, an unhappier customer.
“The hidden secret of networking is that the network delivers loss and delay between packets,” said Geddes, “There is more to broadband than speed or capacity: with many customers wanting lots of different things at once, we also need an absence of variability, and that is what we call stationarity.”
Looked at from the network operator side, there are two key areas to consider: what is driving the cost of broadband and pushing capex sky high, and how to retain and increase your customer base to bring in the revenue. The answers, it seems, are not immediately obvious.
To start with, the knee-jerk telco reaction of pouring capex into infrastructure upgrades and increased capacity is simply not the way to ensure good quality of service and happy customers.  Demand for broadband is highly elastic, expanding to consume whatever supply is on offer and creating a “jack-hammer effect” – which produces variability. Paradoxically, increased investment in bandwidth may be behind that very poor service which leads to customer churn and the panicked assumption that another upgrade is necessary – an “investment cycle of doom.”
This is a deep systemic problem in the industry investment machine. Rushing to premature upgrades masks the real core issue, that of quality of service.  The presenters demonstrated this in heaven-hell model, where full network capacity and happy customers is telco heaven – and the converse, unhappy customers and underused network, is of course telco hell.  Getting the balance is not easy, as increasing local networks pushes down the quality of experience for applications with strong stationarity requirements – exactly what the customer is after.
For Martin, there is a tiny root cause of this: all current packet-based infrastructure relies on it being idle and keeping queues empty to ensure good quality. So your assets must stay idle to keep your customer. The solution lies in thinking about how to reframe both this problem, and the exact nature of the resource the operators are selling.
“Don’t make packets move for their own sake, but focus on customer experience. Change the resource model,” urged Martin. “Throw away the bandwidth model and thought process.” Efficiently allocating resources to customers is more important than bandwidth. Increase capacity, but only in a very targeted way.  In other words, meet heterogeneous  demand with a differentiated product.
This, then, is how to ensure a future of broadband heaven: understand that quality of experience is a function of loss and delay. Characterize your supply requirements properly. Work out what customers are after, certifying fitness of purpose for a particular, actual customer demand rather than a generalised one-size-fits-all concept. And, in the words of the workshop presenters: “Don’t sell bandwidth – sell differential experiences.”

His presentation is embedded as follows:



Thursday 5 September 2013

Throughput Comparison for different wireless technologies

Merged various slides from the recent 4G Americas presentation to get a complete picture of data throughput speeds for various technologies.