Wireless Waffle - A whole spectrum of radio related rubbish
Maths: Not Ofcom's strong suit?signal strength
Friday 9 May, 2014, 08:02 - Spectrum Management, Satellites
Posted by Administrator
esoa duncesIt seems that following the ESOA submission to Ofcom concerning the apparent errors in the RealWireless study on spectrum demand for mobile data reported by Wireless Waffle on 15 Febuary, the offending report has now been re-issued (note the publication date is now 11 April 2014) with the axis on Figure 44 which shows data traffic density re-labelled from 'PB/month/km˛' (PetaBytes) to 'TB/month/km˛' (TeraBytes), thereby reducing the calculated data traffic by a factor of 1000 and now making the document internally consistent. Well done Ofcom and RealWireless, though they could have publicly admitted the apparent error, instead of quietly re-issuing the document with no fanfare. Presumably this now makes ESOA look rather silly.

But... even a 10th grade student could complete the sum that is behind the ITU data forecasts and realise that the axis should have read 'PB' all along (and therefore that the internal inconsistencies are not fixed and that the data in the ITU and RealWireless models is still hundreds of times too large). Here, for you to try, are the values - taken from the ITU's 'Speculator' model - and the maths you need to apply. The values are for 'SC12 SE2' which represents people using 'high multimedia' services in urban offices and is with the ITU model in its 'low market' market setting (it has a higher one too).

User density:120,975 users per km˛
Session arrival rate per user:3.3 arrivals per hour per user
Mean service bit rate:13.83 Mbps
Average session duration:81 seconds per session

Now for the maths...
  • First, multiply the first two numbers to get 'sessions per hour per km˛'. (120,975 × 3.3 = 399,217.5)
  • Then multiply this by the average session duration to get 'seconds of traffic per hour per km˛'. (399,217.5 × 81 = 32,336,617.5)
  • Then multiply by the mean bit rate to get 'Megabits of traffic per hour per km˛'. (32,336,617.5 × 13.83 = 447,215,420)
  • To make the numbers more managable, divide by 8 to get from bits to bytes, then by 1,000,000 to get from Megabytes to Terabytes (447,215,420 ÷ 8,000,000 = 55.9)
So the traffic assumed by the ITU model for people using 'high multimedia' services in urban offices is 55.9 Terabytes per hour per square km. But the figure in the graph in the RealWireless report is per month, so we need to scale this up from hours to months. We now have the thorny question of 'how many hours are there in a day', which for mobile data traffic is not necessarily 24 as you might expect. If the above figures are meant to represent the busy hour (the busiest hour of the day), it would not be right to multiply the value by 24 to get daily traffic, as this would assume every hour to be as busy as the busiest. As a conservative measure, let's assume that the daily traffic is 10 times that of the busiest hour. So daily traffic per square km would be 559 TeraBytes (55.9 × 10 just in case you couldn't work this out in your head).

The number of days in a month is relatively easy to work out, it's 30.4 on average (365.25 ÷ 12). So monthly traffic per square km would be 559 × 30.4 = 16,994 TeraBytes per month per km˛.

ofcom maths skillsThis is the monthly data for just one urban traffic type in the ITU model, there are 19 others. Ignoring the others completely, Figure 44 of the RealWireless report should show monthly traffic in urban areas for the ITU model being 17,000 TeraBytes per month per square km, include the other activities that urban office workers undertake and the value should be much higher still. But it now shows as being just over 100 TB/month/square km for the ITU and less for the RealWireless prediction, 100 or more times too low. Oh dear!

RW report page 085

whos stupid nowSo having corrected the figure in the RealWireless report, it is now wrong. It was correct before. And it still does not tally with the total data forecast for the UK that is in the same report.

Surely there are people at Ofcom who own a calculator, have a GCSE in maths, and possess a modicum of professionalism such that they would want to check the facts before blithely allowing their suppliers to fob them off with an 'oops, we mis-labelled an axis' argument. Presumably they thought that it was ESOA who couldn't handle a calculator properly.

Now who looks silly?
add comment ( 235 views )   |  0 trackbacks   |  permalink   |   ( 3 / 602 )

The Porridge thickens...signal strength
Monday 21 April, 2014, 19:34 - Spectrum Management
Posted by Administrator
Following the recent Wireless Waffle piece on Valles Marineris sized chasm in the values used by the ITU in predicting the demand for IMT spectrum in 2020 spotted by the European Satellite Operators Association in their response to Ofcom's mobile data consultation, others have noted similar gulfs.

tim farrarTelecoms analyst Tim Farrar (pictured right) published an article in GigaOm entitled 'Note to the telecom industry: beware of false models'. In it he takes a different view to ESOA. The ESOA response tries to use the values in the ITU's 'Speculator' model to define the data traffic that the UK would experience in 2020 and discover that applying the values in the ITU model yields results that far exceed forecasts. The GigaOm article instead looks directly at the values found in the ITU model and concludes that they are up to 1000 times too high which generally concurs with findings of the ESOA analysis.

ebu logo 144x144Next the European Broadcasting Union (EBU) have chipped in. Their document, 'Crystal balls, tea leaves or mathematics' critically examines the ITU's model and similar to the others concludes that there are a 'number of erroneous elements'.

Wireless Waffle has been able to get hold of a copy of the 'Speculator' and so exclusively for you, here are some of the values that are causing people such as ESOA, Mr Farrar and the EBU such consternation:

ParameterCurrent ValueNotes
Spectrum EfficiencyFor GSM/UMTS/LTE: 2 to 4 bits/second/Hz/cell.
For LTE-Advanced: 4.5 to 7.3 bits/second/Hz/cell
These look like highly aspirational values!
Call Blocking Rate1%This represents the chance of not being able to make a call (i.e that there is a 99% chance of success).
Population DensityMaximum of 222,333 per sq kmThis occurs in 'SE2, SC12' which equates to interactive high multimedia use in offices in dense urban areas.
Mean Service Bit RateSC6 (streaming super high multimedia): Up to 1 Gbps
SC11 (interactive super high multimedia): Up to 1 Gbps
Really? 1 Gbps on average!

office workerThe population density figure for urban offices using 'interactive high multimedia' is brain achingly odd. For other uses in urban offices, the population densities are significantly lower, so it is not clear why the use of these interactive high multimedia would be so prevalent in offices compared to other applications. Have the ITU assumed that all office workers do all day is play games and watch videos?

A mean (average) service bit rate of 1 Gbps seems excessively excessive. If this was the peak service rate then, maybe, just maybe, this would be possible (and only possible on LTE-Advanced networks, not on the others). But to assume that it is an average seems just crazy.

Of course the big question is, what would the 'Speculator' say, if the values input to it were more realistic? To try and answer this question requires some kind of estimation of what realistic actually means. Whilst we make no claims for the realism of any of the values proposed below, here are some alternative values...

ParameterNew ValueNotes
Spectrum EfficiencyFor GSM/UMTS/LTE: 0.55 to 1.5 bits/second/Hz/cell. For LTE-Advanced: 1.1 to 3 bits/second/Hz/cellThe values for LTE-Advanced are taken from the ITU's own Report M.2134. Those for GSM/UMTS/LTE are half the LTE-Advanced values (roughly in line with the original ratios).
Call Blocking Rate2%A value that more operators would recognise.
Population DensityReduced so that the weighted average values are the same as those in the ESOA report for the UK (e.g. ~11000 per sq km in Urban areas).This should mean that running the ESOA calculations would at least yield the correct population for the UK.
Mean Service Bit RateCapped at 100 Mbps.Seems a little more reasonable based on the technologies likely to be in use by 2020.

The big question is obviously therefore, what does this do to spectrum demand? The original and revised figures are shown in the table below.

Low440 MHz580 MHz900 MHz480 MHz1340 MHz1060 MHz
High540 MHz660 MHz1420 MHz600 MHz1960 MHz1260 MHz

What does this tell us? Oddly, in both cases, the demand for GSM/UMTS/LTE spectrum has increased. This is probably due to the lower spectrum efficiency that these technologies have been assumed to achieve. Conversely, the total spectrum demand has dropped significantly and all of this reduction has come from spectrum for LTE-Advanced.

But what is most striking about these calculations is not necessarily the differences in the results, but the simplicity with which it is possible to present alternative values and find a different outcome. For example, no effort has been made in the above analysis to check the way in which the ITU model apportions traffic between the 2G/3G networks and the LTE-Advanced network. Could, for example, it be argued that by 2020 major carriers in advanced markets (e.g. USA) will have moved all of their data traffic to LTE-Advanced and that only 2G will remain for legacy voice services. itu outlook gloomyThis would almost certainly serve to vastly reduce the amount of 2G/3G spectrum that would be needed, whilst providing only a modest increase in the amount of spectrum that would be needed for LTE-Advanced, given the technology's improved spectrum efficiency. In this case, the total requirement would probably fall further. Or could it be that we will all be living in a virtual environment, with Google glasses projecting us a view of the world in full HD as we stroll around the office - requiring umpteen times more data than the ITU model predicts.

The fact is that any model of this kind, no matter how many brains were employed in developing it, can never be more than a 'best guess', especially when looking 7 to 10 years into the future. Weather forecasters struggle to predict the level of precipitation 7 to 10 days into the future and no-one in their right mind would decide if they needed to carry an umbrella a week next Tuesday based on their forecast. Nor should the vast wireless community take decisions based on this one forecast, it would be irresponsible of them to do so and if the weather changes, they may end up getting soaked!
add comment ( 448 views )   |  0 trackbacks   |  permalink   |   ( 3 / 666 )

H comes after E, E comes after G...signal strength
Friday 4 April, 2014, 09:57 - Much Ado About Nothing
Posted by Administrator
ww explains it allWhen using your mobile phone, smart phone or tablet have you ever noticed that next to the signal strength bars (usually found at the top of the screen), there is often a letter (or two) that seem to change almost at random as you move around, and even sometimes when you aren't moving at all? Have you ever wondered what these letters are there for and what the implications of them changing from one letter to another are? Well wonder no more because the answers are about to follow, as Wireless Waffle explains it all...

Letter(s)MeaningTypical Connection SpeedExplanation
G or GPRSGPRS10-20 kbpssignal bars gThis is a 2G service and is the slowest connection you can get. It's often achingly bad.
E or EDGEEDGE50-60 kbpsThis is also 2G and the second slowest connection - theoretically up to 384 kbps but almost never this fast. Think 'dial-up' internet (if you can remember back that far).
3G3G80-100 kbpsThis is the original 3G mobile system and is good (compared to GSM) but still not brilliant.
HHSPA0.5-2 MbpsA truly broadband wireless connection with good real-life connection speeds.
H+Evolved HSPA (or HSPA+)2-8 Mbpssignal bars h+An even faster connection, might even be termed 'zippy'.
4GLTE5-20 MbpsThe fastest connections available today.
RRoaming-Beware - this means you are connected to a network outside your home country and data costs could be astronomical! The R is sometimes shown in a triangle.
XNo signal-On some phones, an X appears above the signal bars if there is no signal at all.

Note that the typical connection speeds given above are those that are generally achieved in real-life. Though in theory the technologies used can offer faster connections, much depends on how many users are in a cell and what they are doing, how close to the centre of the cell you are, whether you are stationary or on the move, and a whole host of other factors.

Arrows (sometimes coloured, and sometimes integrated into the signal bars) pointing up and down are also illuminated. This just shows whether you are downloading (the downward arrow) or uploading (the upward arrow) data to the mobile network.

youtube bufferingIn addition, the number of bars shown on your signal meter will also affect how good your connection is. So a '3G' connection with all signal bars lit might be better than a 'H+' connection with only one bar lit. However a full strength signal may not necessarily mean a fast connection as most phones show the 'strength' of a signal and not the 'quality' It's quite possible to have a full strength signal that's suffering lots of interference and thus is bad quality.

What does any of this matter? It doesn't really, but if you are wanting to view a YouTube video and your phone is showing 'G' or 'E', the chances of you getting a fast enough connection are virtually nil.
2 comments ( 401 views )   |  0 trackbacks   |  permalink   |   ( 3 / 777 )

Marconiji and Elastic-Magneto Wavessignal strength
Tuesday 1 April, 2014, 08:04 - Much Ado About Nothing
Posted by Administrator
gujarat science 4 bookIt has been widely reported with great delight across India and in the world-wide media that the Level 8 Social Science school text books being given to children in the Indian state of Gujarat contain many inaccuracies. Examples of these heinous inaccuracies include 'factlets' such as these gems:
  • It was Japan that dropped a nuclear bomb on America during World War II (and not vice versa).
  • Carbon Trioxide (CO3) has increased due to the cutting of trees (CO3 is highly unstable but can be made by blowing ozone at dry ice).
  • That after the partition of India in 1947, a new nation called 'Islamic Islamabad' was formed whose capital was Khyber Ghat (and not that Pakistan was formed with a capital Islamabad).
Wireless Waffle has managed, through a contact living in the state, to get hold of a copy of the Level 4 Physics book used in classrooms across Gujurat. Like its social science bretheren, it is also chock-full of errors and mistakes. For example, it claims that:
  • marconijiIndian inventor Guglielmo Marconiji (pictured right) discovered the elastic-magneto wave and his brother, Luigi, is the source of the character in the SuperMario game.
  • Elastic-magneto waves travel at the speed of steam, which is generally taken to be 100 km per hour.
  • The first radio station in India went on-air in 1748 and was called 'Bhapa Stesana Bharati' (Indian Steam Radio), as it used steam engines to generate the elastic-magneto waves. The signals from this station will not reach the moon until a week next Tuesday, if the timetable doesn't change.
  • Phones work by transferring ear molecules between those speaking. Conference calls are a modern invention because of the need to clone ear molecules to be heard in several places at once.
Laughable as these 'factlets' may seem, there are many, much worse, part-truths that are espoused by those who ought to know better... Here's looking at you, ITU!
add comment ( 265 views )   |  0 trackbacks   |  permalink   |   ( 3 / 713 )

<<First <Back | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | Next> Last>>