DigiTAG seeks to inform members and non-members
about important developments in the digital terrestrial television market.
Each month, DigiTAG distributes its web letter with news updates and
further exploration of one topic.
To subscribe to the DigiTAG Web Letter, send
an email to the DigiTAG Project
Evaluating spectrum requirements for mobile telecom services
A key topic for discussion at the next World Radiocommunications Conference (WRC) in November 2015 will be the allocation of spectrum to mobile telecom operators. Under the agenda item 1.1, national administrations will consider additional spectrum allocations to mobile telecom services on a primary basis as well as identify additional frequency bands that could also be used by such services. In preparation, the ITU Working Party 5D has carried out studies to determine the spectrum requirements for mobile telecom services. The results of these studies have been submitted as to the Joint Task Group 4-5-6-7 that is preparing an input to the WRC-15.
Already, the mobile telecom industry has presented its data calling for a significant increase in access to spectrum to satisfy its forecasted growth in demand for wireless broadband services. However, the data presented has faced intense scrutiny and its validity questioned.
Spectrum requirements according to ITU-R M.2290
In January, the ITU issued Report ITU-R M.2290 which quantifies the spectrum needs for the mobile telecom industry to the period up to 2020. According to this report, between 1340 MHz (in low-density areas) and 1960 MHz (in high-density areas) will need to be allocated to mobile telecom services to satisfy the forecasted demand for services. This is significantly higher than earlier estimates.
These figures have been calculated using a spreadsheet-based methodology, known as the ITU Speculator, which is outlined in ITU-R M.1768-1. This methodology takes into consideration 20 types of service categories representing different data rates and applications as well as 6 types of service environments (i.e. dense urban, rural, etc).
In Report M.2290, the data for the input parameters have been updated with the intention of reflecting current market developments as well as to consider such issues as traffic off-loading. According to the telecom industry, which supplied the data for the report, many market forecasts made prior to 2007 had underestimated actual market developments and needed to be updated. It purports to have used data based on studies made by different organisations including the UMTS Forum.
However, it is these input parameters that have been criticized in two studies recently published.
Critique of Report M.2290 from satellite industry
A study recently submitted to the ITU has found significant inconsistencies with the parameters used to estimate mobile traffic. According to this study, issued by a group of organisations mainly from the satellite industry, the use of more realistic values would lead to a significant reduction in the total spectrum requirements for mobile services.
In the model used in Report M.2290 to determine spectrum requirements, unrealistic figures have been used for both the user density (number of people using each application) as well as the amount of traffic consumed per user. This conclusion has been reached by comparing the traffic density figures for urban and suburban areas used in Report M.2290 with the data in the UMTS Forum Report 44 which provides mobile traffic forecasts until 2020. The study does not seek to validate the data in the UMTS report but rather to use it as a benchmark against Report M.2290.
Specifically, the traffic density forecast in urban and suburban areas by the UMTS Forum report is calculated by estimating the proportion of overall wireless traffic generated in the area and dividing by the total urban area within the country in square kilometers. Consideration is given to the fact that urban dwellers consume more traffic than those living in rural areas as well as to the fact that the number of urban dwellers will increase and urban areas will become increasingly dense. However, the traffic density assumed for suburban areas in Report M.2290 is higher than the calculated urban average in the UMTS Forum data. More significantly, the figures in Report M.2290 are overestimated by a factor of 100 or more. According to one specialist in a Gigaom article, the data from the UMTS Forum estimates that by 2020, the average demand for data per square kilometre per month in an urban area will be five terabytes. However, this is less than 1/1200 of the figure assumed in Report M.2290 for suburban areas.
While it can be expected that spectrum requirements should include areas of high demand, including times of high demand (peak demand), this demand should not be used to represent demand for a country as a whole.
Critique in EBU Technical Review
Similarly, a study in the EBU Technical Review has undertaken a critical examination of the mobile spectrum requirements in Report M.2290 and has identified a number of erroneous elements. As a result, it has concluded that the spectrum requirements for mobile telecom services have been greatly overestimated.
The study provides a critical examination of the input assumptions made in Report M.2290. Spectrum requirements will vary depending on such factors as geography and network topology. In terms of geography, Report M.2290 does not take into consideration geographical variations in the demand for mobile services where, for example, growth may be higher in some regions of the world compared to others. Instead, global mobile data traffic forecasts are used to estimate spectrum requirements. Report M.2290 also does not take into consideration various network topologies that determine its overall capacity. While increasing spectrum will increase network capacity, it is also possible to increase capacity by adjusting the network configuration or making a transmission system more efficient.
According to the Technical Review study, several inaccurate assumptions regarding data rates, user densities, and mobility have been made. Among the various types of mobile services that must be considered, the mobile industry has included “super-high multimedia” for high data rate applications with peak data rate of 30 Mbit/s to more than 100 Mbit/s. In the calculations of spectrum requirements data rates exceeding 400 Mbit/s have been assumed. However, it is unclear what type of applications would require such high data rates given that a full HDTV signal requires a data rate of less than 10 Mbit/s using the MPEG-4 AVC standard (even less with the HEVC standard) and Ultra HDTV services are unlikely to need more than 30 Mbit/s per programme. In addition, it is most probable that for such high-data applications, viewers will not use a cellular network but instead a WiFi or fixed broadband network.
Report M.2290 also fails to consider economic issues facing the telecom industry in some parts of the world. Investments will be necessary to finance the required increases to network capacity without which the forecasted traffic growth will not be possible. However, mobile telecom operators face increased financial pressure and it is unlikely that average revenue per mobile user will increase substantially in the coming years. Currently, investment in network capacity outpaces additional revenue from mobile users.
Finally, Report M.2290 does not address recent developments in mobile technologies that will help increase network capacity. Already, current mobile networks carry a substantially higher amount of data within the available spectrum than was considered feasible in the past. The new compression standard HEVC will also help to reduce capacity requirements of data.
In the second part, the Technical Review study outlines the inadequacy of the forecasting methodology used for determining future mobile traffic data. The traffic predictions made in Report M.2290 estimate a growth factor range between 44 and 80. However, according to the EBU study, errors have been made in the time series analysis. Specifically, no evidence is given about the assumed dynamic structure for traffic evolution and the data used for the mathematical functions are based on earlier forecasts thereby reducing its statistical validity.
To illustrate how the extrapolation approach used in Report M.2290 to forecast mobile traffic is misleading, the study applies this approach to forecast a time series of the Dow Jones Index between May 1993 and December 1999 and the NASDAQ Index between February 1996 and February 2000. In both cases, the approach forecast continued growth and did not consider any decline of growth as has actually occurred.
The spectrum requirements of the mobile telecom industry as estimated in Report M.2290 are substantially higher than initially estimated. The implications for such requirements are enormous if the WRC-15 decides to satisfy them, not only for the current spectrum users but also for those that rely on their services. While it is not clear which frequency bands will be used to fulfill the spectrum requirements of the mobile telecom industry, it will be necessary for the current users to be displaced given that telecom networks are unable to share spectrum. The costs, both social and economic, of such a displacement will be high. As a result, it will be essential to ensure that the demands for more spectrum by the mobile telecom industry is based on information that is both valid and realistic.
Source: Natalie Mouyal, on behalf of the DigiTAG Project Office
DigiTAG is an association of stakeholders in the digital terrestrial TV industry and has members from broadcasting, network operators, regulatory, and professional equipment and consumer electronics manufacturing organisations throughout the world. DigiTAG has recently re-launched with new Statutes and the mission of defending and promoting digital terrestrial television on a worldwide basis, and, notably, will work tirelessly to protect spectrum for broadcasting, regardless of the technical standard used on the DTT platform.
Please contact the DigiTAG office if you have any news that you would
like to include in the next DigiTAG Web Letter.
To unsubscribe from the DigiTAG Web Letter, please click
here and put 'UNSUBSCRIBE' in the subject title of your email.