Fast Forward» How the speed of the internet will develop between now and Commissioned by: NLkabel & Cable Europe. Project: PDF

Fast Forward» How the speed of the internet will develop between now and 2020 Commissioned by: NLkabel & Cable Europe Project: Publication number: Published: Utrecht, June 2014 Authors:

Please download to get full document.

View again

of 59
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.

Computers & Electronics

Publish on:

Views: 28 | Pages: 59

Extension: PDF | Download: 0

Fast Forward» How the speed of the internet will develop between now and 2020 Commissioned by: NLkabel & Cable Europe Project: Publication number: Published: Utrecht, June 2014 Authors: Ir. Tommy van der Vorst Ir. ing. Reg. Brennenraedts MBA Ir. David van Kerkhof Dr. ir. ing. Rudi Bekkers Management summary This study provides insight in the future upload and download bandwidth demand for residential broadband connections, by answering the following research question: How will upload and download bandwidth change between now and 2020? Dutch consultancy Dialogic and the Eindhoven University of Technology were commissioned by NLkabel and Cable Europe to carry out this study. Dialogic is an independent research consultancy, focused on innovation and specializing in telecommunications. The Eindhoven University of Technology (TU/e) is a large research and educational institute, and has already collaborated with Dialogic on a variety of projects. It is important to recognize that residential broadband speed demand has a very diverse character. Households vary considerably in their intensity of use, the type of applications they use their connection for, and the amount of traffic these applications generate. In this study, we address this diversity by differentiating between different categories of users, and our overall outcomes are averages over these categories. While the development of the demand for upload speed is related to that of download speeds, their developments are not necessarily identical. Among other things, download speeds are sometimes more critical to users as they are often waiting to consume content or use an application, while they are typically less pressed when uploading data. However, down- and upload speeds are closely entwined. In this study, a prospective quantitative model of bandwidth demand was developed in order to answer the research question. Figure 1 shows the estimated rise in demand over the coming seven years, expressed as the average speed of a sufficient subscription. This means that this subscription satisfies the demand of the average user. In 2013, the average sufficient provisioned speeds are estimated as 15.3 Mbit/s (downstream) and around 1.6 Mbit/s (upstream). Note that this is an average: very large differences exist between different user groups. The demand for bandwidth is expected to grow exponentially over the next seven years. The CAGR (compound annual growth rate) is 40% for downstream and 44% for upstream traffic demand. In 2020, sufficient subscription speeds for the average user are forecast to be approximately 165 Mbit/s (downstream) and 20 Mbit/s (upstream). Dialogic innovatie interactie 3 Forecasted development of the average sufficient provisioned speed Provisioned speed (Mbit/s) Year Upstream Downstream Figure 1. Estimated growth in traffic volume demand relative to 2013 according to our model The subscription speeds presented in Figure 1 were computed by estimating the total demand for traffic volume per day, and calculating the relative urgency of that traffic (the time in which it is to be transmitted). Traffic volume was estimated per user group and per service category. The following pictures show which services drive the need for more bandwidth Estimated upstream demand for services in the period Upstream demand (Mbyte/day) Other services Consultative web browsing Online video Remote backup Remote workplace Online music Peer-to-peer file sharing Overhead Future revolutionary services Figure 2. Projected average daily volume of upstream traffic per residential subscription for the years Dialogic innovatie interactie Estimated downstream demand for services in the period Downstream demand (Mbyte/day) Other services Consultative web browsing Online video Remote workplace Online music File downloads Peer-to-peer file sharing Personal cloud storage Social media / Web 2.0 Overhead Future revolutionary services Figure 3. Projected average daily volume of downstream traffic per residential subscription for the years The first figure above shows the projected volume of upstream traffic demand by service for the period We estimate that in addition to future revolutionary services, overhead traffic and peer-to-peer file sharing will continue to comprise the majority of the upload traffic. Since overhead traffic partly consists of acknowledgement traffic, this demand is mainly driven by high download demand. Peer-to-peer traffic is modeled to be supply-driven rather than demand-driven, and is therefore subject to the growth of the upstream demand. Other demanding services in 2020 are remote backup services, online video and future revolutionary services. The future revolutionary services are modeled by means of a probability distribution of the impact of revolutions and their expected occurrence frequency. These services could for instance be comprised by a surge in the number of connected devices and accompanying services in a household. The estimated demand for daily upstream traffic in 2020 will average at just over 3,000 Mbyte per day. Figure 3 presents the estimated volume of downstream traffic demand for the same period. We found that the majority of downstream demand is from online video. Traffic for consultative, overhead and peer-to-peer file sharing are expected to require significant demand. Again, the future revolutionary services also comprise a large part of the traffic in The total downstream demand for 2020 is estimated at almost 8,000 Mbyte per day. For our estimates, we distinguished different user groups: power users, innovators, mainstream users and laggards. Take-up of the traffic demand for services is not only caused by the increased intensity of services, but also because more users will start using them. Dialogic innovatie interactie 5 Table 1. Estimated development of average sufficient provisioned speeds (in Mbit/s) for different user groups Power users* Up Down * Note power users: The estimations for the sufficient provisioned speeds for power users are based on a different method in which traffic for peer-to-peer is modeled to be supply-driven rather than demand-driven. This means that the power users will always maximally utilize the provisioned bandwidth. Innovators Mainstream users Laggards All users Up Down Up Down Up Down Up Down Table 1 shows the estimated development of subscription speeds that will be sufficient for the various user groups. We found that power users have an extreme usage pattern compared to the other groups. The power users, who account for 2 percent of the users have an estimated demand of 1,155 Mbit/s downstream and 315 Mbit/s upstream in Laggards exhibit a considerable different usage pattern: they have an estimated downstream demand of 6.6 Mbit/s and an upstream demand of 0.8 Mbit/s in A schematic overview of the methodology used in this study is presented in Figure 4. Firstly, we estimated the current aggregate traffic demand and translated this into the current aggregate bandwidth demand. In addition to this translation, we divided the current traffic demand into demand by user group and demand by service. The growth leading to future demand was estimated for each category. We separately estimated the demand for revolutionary services services that do not yet exist but are expected to in the future. Demand by user group Growth by user group Future demand by user group Current aggregate traffic demand Current aggregate bandwidth demand Demand for revolutionary services Future aggregate traffic demand Future aggregate bandwidth demand Demand by service Growth by service Future demand by service Figure 4. Schematic overview of the demand model Based on future demand by user group, by service and for revolutionary services, we estimated future aggregate traffic demand. This demand was subsequently translated into 6 Dialogic innovatie interactie bandwidth demand, analogous to the translation of current traffic demand into bandwidth demand. The estimates of current demand are based on quantitative data from several sources, mainly Sandvine s Global Internet Phenomena, Cisco s Visual Networking Index (VNI) and three Dutch network operators recent measurements of cable networks. These measurements were conducted both on Hybrid Fiber-Coaxial networks (HFC) as well as FttH networks. The divisions by service and by user group are based on the literature review and expert interviews, as are the estimates for growth in demand. Dialogic innovatie interactie 7 Table of contents Management summary Introduction Research questions Status of the researchers Reading guide Methodology Introduction to methodology Task 1. Measurement of traffic on networks Task 2. Interviews with experts Task 3. Literature research Task 4. Demand modelling Task 5. Answering the research questions Model composition Scope of the model Modeling approach Modeling aggregate demand Modeling demand by service Modeling differences between user groups Estimating bandwidth demand for traffic Modeling demand growth Current demand Aggregate demand Demand by service Demand for bandwidth Differences between user groups Demand growth Growth by service Growth in capacity and speed demand Growth from future revolutionary services Differences between user groups Future demand Aggregate demand Demand by service Demand for capacity and speed Differences between user groups Discussion & conclusions Answering the research questions Limitations Conclusion Appendix A. Consulted experts References Dialogic innovatie interactie 9 1 Introduction Ever since the introduction of the internet, the demand for bandwidth has continued to grow. Even though home access connections provide generous amounts of bandwidth nowadays, records are still being broken. A recent example is the network traffic record reached when Apple released its latest mobile operating system [13]. The end of growth in bandwidth demand seems to be nowhere in sight. In this study, we predict the future demand for traffic and bandwidth. We mainly focus on the upstream bandwidth in that respect. However, since the upstream bandwidth is inextricably linked with the downstream bandwidth, the research will also research the latter. 1.1 Research questions The central question in this study is How will upload and download bandwidth demand have developed by 2020? In order to provide the answer, we first need to address the following set of sub-questions: 1. To what extent do currently available applications contribute to upstream traffic? 1 2. To what extent has the need for upstream traffic of currently available applications changed in recent years? 3. Which business applications require high upstream traffic? 4. To what extent will consumers use more business applications by 2020? 5. To what extent will consumers use other applications with a high demand for upstream traffic by 2020? 6. Which upstream and downstream speeds will be sufficient for future demand? Our research focuses mainly on consumers in the Netherlands and other West European countries with highly developed broadband markets. Questions 1, 2 and 3 focus on the current situation, while the final three questions deal with the future development of consumer demand for internet traffic. It is essential to decide on a time horizon in order to formulate conclusions relevant for defining policy as well as strategy. As the broadband market is highly dynamic, we have chosen a time horizon of seven years (until 2020). 1.2 Status of the researchers This study was conducted by Dialogic innovatie & interactie and the Eindhoven University of Technology (TU/e). Dialogic is a research consultancy in the Netherlands, focused on innovation and specializing in telecommunications. In the past fifteen years Dialogic has conducted studies for many clients in the public and private domain, in both the Netherlands and internationally. Examples are ACM (the Dutch national regulatory 1 With respect to both upstream and downstream traffic, we distinguish between volume and (peak) speeds/capacity. Throughout this document we use upstream or downstream traffic to indicate volume. Dialogic innovatie interactie 11 authority for telecommunications), European Commission, Dutch Ministry of Economic Affairs, most Dutch provinces, et cetera. Dialogic maintains close connections with the world-wide academic community, and has already collaborated with Eindhoven University of Technology on various telecommunications-related projects. The Eindhoven University of Technology (TU/e) is a large research and educational institute based in the south of the Netherlands, and has already collaborated with Dialogic on a variety of projects. 1.3 Reading guide The report is structured as follows: Chapter 2 outlines our methodology. In order to find answers to the research questions, we conducted interviews with experts, desk research/literature study and used measurement data on upstream and downstream traffic. In chapter 3 we elaborate on the composition of the bandwidth demand estimation model. In the subsequent three chapters, we look at current demand, estimating firstly the growth factors involved and finally future demand. Chapter 7 ends with our conclusions, discussion and limitations. 12 Dialogic innovatie interactie 2 Methodology This chapter describes the methodology we applied to answer the research questions presented in chapter 1. We will elaborate on the building blocks used to create the bandwidth demand estimation model. 2.1 Introduction to methodology As you can see from the diagram in Figure 5, we used three different sources to build the model and estimate its parameters (calibration): measurement of traffic on networks, interviews with experts and literature research. Task 1. Measurement of traffic on networks Task 2. Interviews with experts Task 3. Literature research Task 4. Demand modelling Task 5. Estimation of future demand Figure 5. Overview of the research methodology In the following paragraphs, we will elaborate on these building blocks and how they are interrelated. 2.2 Task 1. Measurement of traffic on networks We used (raw) data from measurements of networks that deliver fixed broadband services to residential end users. First of all, we gathered network measurements carried out by several Dutch Internet Service Providers (ISPs), who chose to remain anonymous 2. Throughout this report, we will use the following names to indicate the various ISPs: ISP A: a large, Dutch ISP providing a variety of primarily cable-based subscriptions. These subscriptions have asymmetric advertised speed limits. Traffic measurements were conducted on random days from November 5-12 in 2013 on a randomly selected subset of 1000 subscribers. 2 Note that the ISPs data was obtained from network management systems. No data on the content of transmissions was recorded or analyzed, nor could measurements be traced to individual subscribers. Dialogic innovatie interactie 13 ISP B: a smaller, local Dutch ISP that serves approximately 5,000 subscribers, of whom a small number is a (small or home) business. Connections are either over cable or fiber, but always have symmetric speed limits. Measurements were made on November 23, 2013 and cover all subscriptions. ISP C: a large, Dutch ISP providing a variety of primarily cable-based subscriptions. These have asymmetric advertised speed limits. Besides this private ISP data, we used public data sources of the following parties on internet traffic: Sandvine [2][3]: Sandvine is a manufacturer of internet traffic monitoring and shaping equipment that presents actual usage data of traffic on its networks every quarter. In the report Global Internet Phenomena, they present download and upload traffic in Europe based on actual usage data. Cisco VNI [7]: Hardware manufacturer Cisco publishes the Visual Networking Index online, a tool that estimates IP traffic growth until It bases its estimates on number of users, application adoption, minutes of use and bitrates and speeds. 2.3 Task 2. Interviews with experts We conducted ten interviews with international domain experts to confirm and validate expected trends regarding demand and usage, for both upload and download speeds and traffic. The ten experts fall into three groups: service providers, researchers / academics and network operators. Interviewees were selected by the project team based on their expected expertise and represent a balanced combination of the above-mentioned expert types. Half of the interviewees are Dutch; the other experts are mostly West European, and one American. In some cases a relevant company was approached and the company selected the most suitable staff member; in other cases the project team deemed a person suitable, based on published research or the project team s previous experience. A list of the interviewees can be found in Appendix A. The interviews were semi-structured and lasted on average 1.5 hours. 2.4 Task 3. Literature research Literature research provided supplementary insight. We used academic and non-academic sources to provide input for the model estimation, including the following types of literature: Several research papers providing detailed insight in the usage pattern of specific services, both quantitative and qualitative. Research on innovation studies, adoption models and growth modelling. Additional reports on traffic measurement. Some literature includes types of growth modelling, such as the methodology description of Cisco s VNI [9]. 2.5 Task 4. Demand modelling The first three tasks aimed to provide input for the demand model. The sources were used both to create as well as calibrate the model. The composition of the model, which will be explained in more detail in the following chapter, was based on assumptions extracted from the expert interviews and the literature review. The model was then calibrated with 14 Dialogic innovatie interactie estimates taken from all the building blocks. Significant additional sources were the actual measurements of network traffic by the ISPs, Sandvine and Cisco. 2.6 Task 5. Answering the research questions Finally, the outcomes of the demand model were used to answer the research questions presented in chapter 1. Dialogic innovatie interactie 15 3 Model composition Having explained how our bandwidth demand estimation model was created, we will now describe the composition of the model. Firstly, we present a conceptual model, briefly introducing all its components. Later on we will further elaborate on these components and our assumptions. Moreover, we will indicate how the various research sources were used to calibrate the model (the calibration is presented in chapter 5). 3.1 Scope of the model For the purposes of this study, we made several key choices regarding the model s scope and boundaries. The aim was to estimate bandwidth demand by the year 2020 and the development between 2013 and then. The unit of analysis was household connections. Although we did not model individual households, we aimed to distinguish different groups as well as indicate the distribution of demand over households. Note that we thus only model the use of business applications for connections with a consumer subscription; also small offices or home offices (SOHO) are excluded. Geographically speaking, we were interested in those West-European countries where residential broadband connections are commonplace. We therefore assumed that the exogenic growth would be negligible, meaning that we did not expect a significant impact on the demand resulting from first-time internet users. We have modelled demand for bandwidth rather than availability of bandwidth, i.e. bandwidth traffic consumption is driven by demand rather than supply. We assumed that the available bandwidth is secondary to the demand: this means that operators will use the expected demand as a guideline for dimensioning their network Traffic and bandwidth In this report,
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks