Whilst the potential use cases of 5G remain widely spoken about, less airtime has been given to how underlying infrastructure will need to adapt to meet the demands of 5G networks. In this white paper, the FTTX Council Africa examine the role for fibre in 5G and how this will complement other forms of infrastructure.
Welcome to the connected world where the pace of technological advancement is quickly accelerating, and a new revolution is dawning.
Over the past two decades Fixed Network Operators (FNO’s) have been building fibre optic networks that deliver gigabit and tens of gigabit speeds to homes and businesses around the world at approximately 1000 times faster speeds than we had in the year 20001,2.In addition, mobile communications have gone from luxury to necessity in just a few decades. In 1990, there were over 5 million mobile users the U.S.3, and in 2016 there were about 310 million connected devices4. Last year, the average smartphone used 4,432 MBs of data per month, 51.6 times more than non-smartphone handsets5. Smart everything, from cities to medicine, transportation, and commerce, will make our world safer, healthier, and better. All these benefits will be enabled by ultra-reliable technology connected to everything through millions of mobile devices operating at blinding speeds. By 2021, it is estimated that there will be 27.1 billion networked devices and connections, up 58.47% from 17.1 billion today6. Most of these devices will connect over wireless architecture over a short distance to a dense fibre connected network of cells that don’t exist today, through fibre that is largely not yet in place. The technology that will connect users to this is 5G. Since the 1990s the world has seen four generations or “G”s of wireless technology move consumers from “bag phones” to LTE, and 5G is next. This technology will enable an entirely new world driven by innovation that is about to exponentially increase. For instance, in just the past few years our fitness and health can be monitored by a wristband, we can watch seemingly limitless video content, and answer the front door through a smartphone app from almost anywhere in the world. These, and many other innovations, were first enabled and lately restricted by 4G LTE, which can’t keep pace with new application and bandwidth demands. 5G is not just a faster version of 4G LTE, but a mission-critical platform that will launch the world into living truly smart.
Figure one: Fibre enabled 5G applications and associated network requirements
The needs for the next stage of mobility define 5G, with the promise of gigabit speeds to each mobile user, sub one-millisecond latency, connecting 1 million devices per square kilometre7, 5G is a prerequisite for widespread deployment of other technologies. This includes autonomous vehicles, drone networks, and augmented/virtual reality (AR/VR), which can bring enormous societal benefits. In comparison to where we are today, 5G will enable exponentially higher peak data rates and will require a massive increase in the number of fibre-fed cell sites. These enhanced current applications and future applications will require a new generation of wireless network, but to understand how that network must be built requires understanding the history of wireless network evolution.
In general, most of the world’s wireless networks were originally designed to carry voice calls, with “macrocell” sites roughly 1-40 kilometers apart8. However, delivered bandwidth drops off as users move away from the cell site. That means that a user’s experience worsens as the user moves away from the cell, especially with high bandwidth applications such as video. For that reason, the cellular industry has developed a range of smaller cells often called micro, pico, or femto cells. These small cells are used to “fill in” spots in the network to boost bandwidth in low throughput areas.Since 5G uses higher frequencies to reach higher data rates, its range is about 10 to 15 times shorter than 4G LTE. For 5G to meet its promise, small cells need to be only 200 to 1,000 feet apart. Also, to deliver at least 1 Gbps peak speed to each user, the minimum downlink speed required to each cell is 20 Gbps, and uplink peak data rate is 10 Gbps9. For example, one kilometre that is supported by one cell site today for 4G LTE may require approximately 30 cells for 5G. Moreover, in order to achieve 10 to 20 Gbps rates, each cell will have to be connected to fibre as the backhaul architecture. Feeding all of these cells with fibre will require up to 4 km of fibre optic cable in that single square kilometer.The quantity of incremental new fibre and cells that could be required to support 5G is substantial. Let’s consider a US example. Based on data gathered in the 2010 U.S. census, the top 25 urbanised areas are 33,181 square miles10. (Urbanised area extents are defined by the Census Bureau based on population density and other characteristics of the built environment11.) If those top 25 urban areas were fully served with 5G service as assumed in the graphic above, approximately 265,448 miles of fibre cable would be required to fully support those systems. While the fibre cable requirements for each deployment will be different based on geography, this example shows how deep fibre must be deployed to fully enable 5G in an urban [city] environment.
Figure 2: Fibre fed small cells propagate 5G
Figure 3: 5G cell density
Recently, Deloitte Consulting LLP estimated that the U.S. requires $130 to $150 billion of fibre investment in the next five to seven years for broadband fibre deployment12. An undertaking that significant would require thousands of new manufacturing and installation jobs around the country. If this endeavor is taken seriously and the government lowers barriers to incentivise new fibre and small cell construction, our nation will benefit from this new technology, and it would stimulate entirely new applications and industries. By one estimate, one extra wireless job creates six and a half additional jobs13. A dense fibre network can enable this future economy, but we are still years away at our current pace of fibre optic deployment.
What is 5G and why is it important?
Whilst the final definition of 5G is still debatable, we can accept that the following components will form part:
- 1-10Gbps connections to end points in the field (i.e. not theoretical maximum)
- 1 millisecond end-to-end round-trip delay (latency)
- 1000x bandwidth per unit area
- 10-100x number of connected devices
- (Perception of) 99.999% availability
- (Perception of) 100% coverage
- 90% reduction in network energy usage
- Up to 10-year battery life for low power, machine-type devices14 5G is the next generation of wireless network, and to better understand 5G requires us to consider firstly, what benefits it is expected to deliver and secondly, how they will be delivered.
The mass adoption of mobile communications has surpassed expectation and it keeps growing. This mass adoption has consequences for the fixed and mobile networks on which it depends. According to the Cisco Visual Networking Index Forecast, mobile data traffic has grown 4,000-fold over the past ten years and almost 400 million-fold over the past 15 years. In 2015 alone, 563 million mobile devices and connections were added to networks15. This explosion in data traveling over networks, and devices connecting to them also drive, and is driven by, emerging applications and services. We require high bandwidth, low latency connections to stream the newest shows, livestream via social media networks, and video chat with loved ones. Latency is the time it takes for a data packet to travel across a network from one point on the network to another. High latencies may affect the perceived quality of some interactive services such as phone calls over the internet, video chat, or online multiplayer games16. To explore augmented and virtual reality, wire our entire homes with sensors, automate factories and farms, and make a fully autonomous transportation system, the architecture of our current wireless network will have to transform. Fibre and cell densification will be required to feed more cells that will be much closer to users than they are today. Today’s wireless networks were not built to handle the necessary bandwidth or low latency needed to enable these applications.
So how is 5G different? 5G promises the delivery of gigabit wireless speeds, but it is also more. With peak speeds over 1 Gbps to the individual user and latency less than 1 ms17, 5G becomes a stepping stone towards broader acceptance of technologies such as autonomous vehicles and augmented/virtual reality. To understand the power of this functionality, it helps to understand that watching TV shows or movies on Netflix uses about 1 GB of data per hour for each stream of standard definition (SD) video, and up to 3 GB per hour for each stream of high definition (HD) video18, and it takes 100-400 ms to blink your eye19.
Figure 4: Densification from 3G to 4G requires 25x more fibre; to go from 4G to 5G at least 16x more fibre is required
Autonomous Vehicles
In addition to the above mentioned technologies, the benefits of 5G to society will be immense. For the purpose of autonomous vehicles, creating a reliable system will require minimising the vehicle response time (latency) to quickly changing traffic conditions. At the time of this writing, it is not clear whether autonomous vehicles will primarily use 5G or dedicated short range communications (DSRC), or some combination of the two. However, for autonomous vehicles to become a reality, low network latency is imperative.
A car or truck traveling 112km/h is traveling 31m/s, and even small amounts of latency in the network, when combined with the time required other computations in a vehicle, could mean the difference between a safe ride and an accident. Also, a single autonomous vehicle with always-on sensors and cameras is expected to produce massive amounts of data. According to Intel CEO Brian Krzanich, around 40 terabytes (40,000 GB) of data can be generated for eight hours of driving20, which will likely need to be transmitted somewhere. This is like watching Netflix HD Video 24 hours a day for about 541 days21. Those transmissions will travel over a very short distance wirelessly to a small cell, then over a fibre optic network to a control centre. When the fibre network is constructed and available to enable autonomous vehicles, one can imagine the world with fewer stoplights and reduced traffic congestion. Disabled people will have much more freedom to travel. Finally, and most importantly, it’s expected that our digital drivers supported by a reliable fibre optic network will be much safer than human drivers, greatly reducing auto accidents and fatalities.
Virtual reality and augmented reality
Fully realised networked virtual reality (VR) and augmented reality (AR) are additional technologies that require ultra-low latency and ultra-high bandwidth. With VR and AR, it becomes possible to have a front row seat at any concert, sporting event, or another event in the world. Beyond entertainment, VR holds great promise as a training tool for a wide variety of industries, enabling simulation of complex tasks. Also, remote surgery and other telemedicine procedures can become more available than they are today. However, without ultra-low latency roughly less than 20 milliseconds22, users are more likely to experience motion sickness, and without bandwidth, the virtual world doesn’t replicate the real world. Estimates of the bandwidth requirements of fully networked VR range from 500 Mbps to 4 Gbps per user23.All of the gadgets connected to the internet and the applications just referenced fall into a massive bucket referred to the Internet of Things (IoT). Worldwide, the IoT will increase revenue more than four times from $892bn in 2015 to $4tnin 202524. By 2022, there will be 1.5bn IoT devices with cellular connections, up from 400mn in 2016 worldwide[25], assuming new FTTH and 5G networks are in place.
5G - limited by physics; built upon dense fibre connections
Many people’s natural inclination is to think of fibre and wireless as competing technologies. In reality, the two technologies are highly complementary and co-dependent.
At the most basic level, wireless technologies free people from the constraints of place and enable them to lead fuller lives, but they are limited in range and bandwidth.
Fibre technologies offer virtually unlimited bandwidth over very long distances to fixed locations. Billions of wireless devices generate performance demands on networks. These demands can only be satisfied with the prodigious amounts of bandwidth made available by fibre-based networks.
Whilst the promise of 5G is incredibly exciting, this evolution will place enormous demands on fixed-wireline networks. 5G requires additional small cells to be closer together versus existing systems. This, in turn, increases the need for fibre to these small cells. The reason why comes down to basic physics.
Figure 5: The small cell value proposition
5G is expected to expand into new spectrum areas beyond what is used for LTE and LTE-A. Although some unlicensed spectrum may become available, available spectrum below 5 GHz will remain somewhat limited. This situation has focused the industry on looking at the available spectrum in the 30 and 300 GHz bands, which are also known as millimeter band frequencies indicating the very short wavelength of the signal26. These bands provide much more bandwidth compared to LTE (hundreds of MHz verses tens of MHz). Unfortunately, this gain in bandwidth, required for both increased speeds and lower latency, comes at the expense of much more limited radio wave propagation properties. At millimeter band frequencies, signals need line-of-sight pathways from the radio-to-end-user device, and these signals are easily attenuated. Ultra-high data rates require a very high signal-to-noise ratio for the data to be properly received and interpreted, and any significant attenuation reduces the delivered bandwidth. Even simple obstacles like foliage, raindrops, and building materials have dramatic impacts on the signal range with ultra-high bandwidth. This is why 5G cells could be limited to a radii of 210 metres or less. This is very different from the cell radii that we are familiar with for 2G, 3G, and 4G. Original cellular networks were designed to carry voice calls with “macrocell” sites spaced from 5 to 16 km or farther apart. The rollout of 4G was accompanied by the concept of a small cell to fill in gaps in specific areas of high traffic density, under the umbrella of a macro cell. With the introduction of 5G, an entirely new paradigm regarding density is being introduced. The move to drastically smaller cell radii than required for 4G creates the need for a significantly larger number of cell sites.
For example, let’s assume a 230m cell radius requires roughly 60 cells to cover a square mile. Connecting these 60 cells per square km together requires approximately 13km of cable. Extrapolating these numbers across the areas where we live, work, and drive will require millions of new cell sites and millions of miles of additional cable. This is just for one service provider, assuming no duplication of infrastructure.
For these speeds and distances, small cell backhaul with copper for 5G will not work due to its limited bandwidth. Microwave can help in some areas where running a cable is not as practical, but ultimately this is not scalable in the vast scope of meeting the 5G promise. It becomes apparent that fibre is the only obvious choice to serve the increased number of wireless serving points at the required transport bandwidth.
Figure 6: Connect the dots
Of course, not all areas will be connected at once. Likely, areas with very heavy consumer demand based on population and high data usage demographics will be brought online first, then other areas will be filled in, including major traffic corridors. The overall build will take a long time, and without fibre infrastructure, some areas may never see 5G service, further increasing the digital divide that exists today between urban/suburban and rural areas.
There’s an important additional wrinkle that will happen with the 5G rollout that will require an abundance of fibre. Traditional networks have had computing and processing power co-located at the cell site with the wireless antennae and transmission equipment, such that the data from the cell site to the network is compressed, and this is known as backhaul. That will change with the 5G rollout. The 5G rollout will be accompanied by a different network architecture called “Cloud (or centralised) Radio Access Networks,” otherwise known as C-RAN.
The C-RAN architecture places the main processing power away from the local cell site deeper in the network and aggregates the processing of many cells into one location. The benefit of C-RAN is the extremely complex radio signals required for 5G can be much more efficiently (and inexpensively) processed in a central location instead of at each cell site.
The concept can be loosely compared to cloud computing that has revolutionised so many activities. Cloud based applications accounted for 86% of mobile data traffic at the end of 2016[27]. 5G will have relatively dumb cell transmission equipment at the edge of the network, but very high speed and low latency connections will be needed to connect that equipment to the processing power at the centralised locations. This is another reason much more fibre will be required for 5G to meet its promise. The capacity required for this function is called fronthaul. Compared to the backhaul connections for 4G LTE, fronthaul to the 5G cell site will require up to 10 times higher data rates.
From the CRAN processing location, all of that information will then need to be transmitted to its final location, through the core of the network. The term used to describe the network capacity needed on the backside from the processing hardware through the core of the system is called backhaul capacity. Both fronthaul and backhaul will need to be increased dramatically to handle the projected demand of 5G networks.
Convergence as a springboard to 5G
Converged wireless/wireline operators today typically run three separate networks (residential, business, and mobile), primarily overbuilding themselves. When the number of mobile and business endpoints were relatively small, this was a viable solution since the endpoints were often in different locations. However, 5G densification requires 10 to 100 times more cell sites than exist today. Building and operating three separate networks is not extremely efficient and is not practical at scale. When possible, the value of a converged physical infrastructure increases with the number of connected endpoints.
A view on how these different endpoints could coexist and be connected with either passive optical networks (PON) or point-to-point (P2P) is conceptually shown below in Figure 6. (red: business, blue: 5G, green: FTTH residential)
The way fibre fronthaul or backhaul networks for 5G are being constructed may significantly vary between operators, target topology, or geographical region. It is certain that wireless operators will try to leverage existing fibre assets as much as possible, but they will also need to deploy additional fibre.
As we see in today’s 4G environment, there will be a multitude of different design approaches that are being used to provide connectivity to the ever-increasing number of cell sites. Fibre based approaches may include dark fibre, P2P fibre connections, coarse wavelength division multiplexing (CWDM), dense wavelength division multiplexing (DWDM), or PON connections. Other methods in 4G include microwave or copper based techniques. As discussed earlier in the paper, it is widely expected that aside from limited microwave backhaul, the main connectivity method will be fibre for 5G. This is necessary to satisfy the stringent bandwidth and latency requirements.
In some areas where PON networks for residential services have already been constructed, the possibility exists to leverage the fibre already in the ground and mix existing residential and small business rollouts with 5G feeders in the same network. Some PON technologies have been developed and standardised over the last 15 years. Speeds have evolved from 155 Mbps APON in 2000 to 100 Gbps PON, currently in the standardisation process with IEEE. All of the “flavours” of PON use the same outside plant architecture that has been deployed in mass since the early 2000s, and multiple newer generations of PON can be operated across a single fibre, courtesy of the forethought of standardisations committees. Where fibre exists, this can enable fibre networks to carry 5G traffic simultaneously with existing traffic of other types.
The backhaul for 4G may have been for “slow” connections like GPON, but 5G will need up to ten times higher data rates for support fronthaul or mid-haul. Possible available technology choices today are XGS PON (10 Gbps up and downstream) followed by NGPON2 (40 Gbps upstream and downstream, with potentially 80 Gbps on the near-term horizon), WDM PON, and even higher rate PON solutions in the future. Although the newer protocols have had limited scale rollouts in the U.S., it is widely expected when 5G standards are finalised in 2020, a service provider will have a choice of various protocols to provide fronthaul and backhaul capacity for 5G networks around the world, in conjunction with their existing and newly deployed fibre network architecture.
Regardless of the type of protocols used over fibre, service providers and community stakeholders of all types should begin planning now to include extra capacity in their networks for 5G. Having fibre in an area is likely one of the best ways to preposition a community to be early on the list for the rollout. Conversely, the more barriers in place to deploying fibre, the later a community is likely to be served.
What will the future be? FTTH or 5GTTH
Everyone agrees that fibre is a prerequisite for 5G. However, the question “Will 5G To The Home replace all fibre networks (FTTH)?” is often raised. That’s somewhat to be expected, given that there is much discussion about the benefits that 5G can bring, but less discussion about how to implement it. These types of discussion has led to the misunderstandings in the industry and the age old “either/or” mentality prevails amongst operators.
In order for 5G to be realised, Mobile Network Operators have to change their thinking. They can no longer view fibre as a threat to their business, but instead need to accept it as an enabler and an absolute necessity to realise 5G. The traditional architectures of wireless networks are forever gone. In fact, in the FTTX Council there is a new saying: “There certainly is a lot of wires in wireless communications”. It is the dawn of the infraco – an era where the symbiotic relationship between fixed and mobile is truly realised.
From an investors point of view the corollary to the above question is, “If I invest in fibre, will it be become a stranded investment when 5G comes?” By now, it should be apparent that the answer is “No, in order to realise 5G fibre is a necessity!”
In fact, substantially more fibre will be required than what is currently available. It is possibly one of the biggest investment opportunities that investors has seen to date. The FTTX Council Africa further suggest that there will be mass consolidation in the markets as MNOs will aim to buy FNO’s and FNO’s will venture into the wireless space attempting to land grab 5G cells along their fibre routes. 5G opens up new market opportunities for FNO’s and it is a relative simple market entry since they already own the backhaul.
It also provides major opportunity for both towercos to expand their assets. According to TowerXchange more than 30 tower transactions of scale have been completed in the sub-Saharan African market and independent towercos now own 56,240 (or 38%) of the region’s estimated 146,947 towers. The vast majority of the region’s towerco owned towers are owned by four players, American Tower, Eaton Towers, Helios Towers and IHS Towers although a number of build to suit players continue to show steady organic growth with a handful demonstrating an appetite for tower portfolios that their larger competitors have shied away from. Many towercos are now venturing into the fixed line space, reasoning that it is all just ínfrastructure.
Industry will more than likely see further consolidation as towercos enter the fixed line market and absorb FNO’s and visa versa.
For this reason the FTTX Council agrees that there may be scenarios where 5G may provide the only connection to homes, this may only happen where the economics of reaching a small percentage of existing homes with fibre is prohibitive. The Council does not believe fixed 5G to the home as the only connection to the home will become the norm, simply because the cost of data is prohibitive and MNOs are finding it very difficult to compete withthe  fixed line market at a packet level. In fact, we have adopted as fact that even with very large investments in fronthaul and backhaul network capacity and small cells, bandwidth demand will continue to outstrip supply as it has continually done in the past. Mobile traffic will represent 20% of all IP traffic by 2021 compared to 8% in 201628. However, even after 5G is deployed, home network connections, which will need to support multiple 4K and soon 8K video streams, hundreds of in-home internet connected things, and multiple VR and AR users, will require much higher bandwidths than can be delivered by 5G. In addition, unlike 4G LTE, there are barriers to high data rate 5G reaching inside of homes due to its inability to penetrate exterior walls, so it is likely that FTTH networks deployed to reach an optical network terminal (ONT) located at an interior point deep within the home will continue to predominate as is the norm today, with Wi-Fi providing the final connections to devices inside the home. Ongoing investments in FTTH are increasing. Based on research conducted by Africa Analysis (figure 7) indicates healthy growth rates in the fibre market.
Figure 7: March 2018
Whilst we are satisfied with the increase in not only deployment but the take rate as well, there are still many more homes that need fibre, and as highlighted earlier in the paper, if 5G is going to be a reality for consumers, it needs to be much closer to residential more fibre is needed to enable 5G.
Conclusion
5G holds great promise. 3G and 4G also held great promise, but only for a decade or two. The original hope for 4G was to provide 100 Mbps to the end user, per the ITU-R M.1645 recommendation29. Although it has been a great service and vast improvement over 3G, the reality is that the original promise of 4G LTE technology has not been completely fulfilled30, mainly because it is costly to build out networks to supply the bandwidth needed, and user demand is insatiable. Other options such as small cell backhaul with copper for 5G and microwave will not work. 4G LTE-A will be a bridge to 5G. 5G will provide more capacity and lower latency as previously discussed, but demand for capacity will continue to rise. Emerging standards indicate 25 Gbps and 100 Gbps systems will improve 5G services over time and will bridge to future generations of wireless technology. The explosion of the Internet of Things will start a snowball effect that will lead to a constant need for bandwidth. The quicker the roadblocks to deployment are removed, the sooner the world will change. One thing remains constant, the need for fibre optic deployment deep into the network to enable the high-tech economy of the future is an absolute requirement. Finally the two architectures have merged to form a unified network.
The road to 5G is paved with fibre, and the time is now to ramp up deployment.
References
1 http://money.cnn.com/2000/11/01/technology/fcc_dsl/
4 https://www.ctia.org/industry-data/ctia-annual-wireless-industry-survey
5 http://www.cisco.com/c/dam/assets/sol/sp/vni/forecast_highlights_mobile/index.html#~Country
7 https://www.itu.int/md/R15-SG05-C-0040/en
8 http://www.pitt.edu/~dtipper/2720/2720_Slides5.pdf
9 https://www.itu.int/md/R15-SG05-C-0040/en
10 https://www.census.gov/dataviz/visualizations/026/508.php
11 https://www.census.gov/dataviz/visualizations/026/
13 https://www.ctia.org/industry-data/wireless-quick-facts
17 http://ieeexplore.ieee.org/document/7169508/all-figures
18 https://help.netflix.com/en/node/87
19 http://bionumbers.hms.harvard.edu/search.aspx?task=searchbypop&rpp=100
21 https://help.netflix.com/en/node/87
22 https://www.qualcomm.com/documents/whitepaper-making-immersive-virtual-reality-possible-mobile
24 Machina Research, Forecasting the Totality of the IoT Revenue Opportunity, April 28, 2016; available at: https://machinaresearch.com/report/forecasting-the-totality-of-the-iot-revenue-opportunity/
25 https://www.ctia.org/industry-data/facts
26 https://www.youtube.com/watch?v=GEx_d0SjvS0
27 http://www.cisco.com/c/dam/assets/sol/sp/vni/forecast_highlights_mobile/index.html
29 http://www.itu.int/dms_pubrec/itu-r/rec/m/R-REC-M.1645-0-200306-I!!PDF-E.pdf
30 http://ieeexplore.ieee.org/document/7169508/all-figures
