The Cisco data highlights the increase in IP traffic over mobile networks, with the total in 2020 being about ten times the total in 2015. Further, almost all this traffic is associated with video, web applications, or audio streaming. Voice conversations represent a very small part of the total IP traffic on mobile networks. This means that all this traffic is passing along backbone networks to or from data centres. The video and audio files are stored in data centres, and the web applications are hosted in data centres. In other words, for every Byte of video traffic on the mobile network – the access network delivering it to the end-user – there will be a Byte of traffic on the backbone networks linking the data centres and IP routers.
Many IoT applications, such as self- driving vehicles or industrial controls, also will involve data centres. The requirements for low latency in some of these applications, however, may require local data centres as opposed to the large but remote data centres operated by some web-service companies such as Microsoft or Amazon. Future applications involving artificial intelligence, augmented reality, or virtual reality also may require low latency, necessitating more local data centres for storage and processing.
This means that as 5G applications ramp up, there will be cable demand associated with increased data centre capacity, including new local data centres. Some carriers have discussed plans to situate new local data centres in telecom switching centres – the former exchanges or central offices. There also have been proposals to provide data storage and processing capacity for local applications in mobile base stations – typically huts, vaults, or large cabinets located at the base of cell towers. Any new data centres will require cable for interconnections, probably optical cable, as well as internal cable to connect the racks of servers and processors. The data centres’ internal cabling could be copper or fibre.
The development of 5G networks also will drive demand for new cables to link the base transceiver stations to the base station controllers (backhaul) and to link the base stations to the antenna heads (fronthaul). This may account for more new cable demand than the increases in backbone traffic and the requirement for more data centres. The reason is that 5G network construction and operation will involve significantly more cells and antenna heads than previous mobile generations.
How many more? This is still being worked out. The number of 5G antennas, or the density of cell sites, will vary with the density of users. That is, urban environments will need more antennas than rural areas. This is true for previous generations. But for 5G, the dense areas will require significantly more antennas than 3G or 4G due to the frequencies being used and the capacity requirements.
The term “densification” refers to installation of more cellular antennas in a given geographic area. One example is the increasing use of small-cell technology with 4G mobile technology. The deployment of 5G networks will bring a new wave of mobile network densification, especially in urban areas. For some countries or some carriers, the total number of 4G cell sites is published. Dividing this number into the country’s square km of land area gives an indication of the average 4G cell-site radius for the entire country. For Japan, this average is about 0.8 km. For the US, it is 3.3 km, and for China it is 2.1 km. As noted, the average radius will be smaller in urban areas, and larger in rural areas.
The table shows the inter-site distance for 5G in different environments. This distance would be twice the radius. Even without knowing the distribution of urban and rural areas, the table suggests that the number of 5G sites will be a multiple of 4G sites. For urban areas, some carriers have estimated a multiple of 10, meaning that a 5G network will require 10 times more antennas than a 4G network. Looking at the range of densities in different countries, another estimate is that urban 5G networks may require 6 to 25 times more cell sites than 4G networks, depending on the country and city size.
A white paper published in late 2017 by the US Fiber Broadband Association concluded that 5G networks covering the US’s 25 largest metro areas would require 2.2 million km of optical cable, in terms of route-km. This conclusion was based on the assumption that cell-site spacing would be 750 feet (229 metres), and that this spacing is required throughout all 25 cities. The calculation also relies on assumptions about the cable architecture.
The association’s conclusion identifies the amount of cable needed to connect the cell sites, not the fibre-km. The 2.2 million km must be multiplied by an average fibre count to arrive at the fibre requirement. The average count over the entire cable plant should be at least six, and it more likely is in the range of 12 to 24, if not higher. This suggests a potential market for 24 million fibre-km, or more, to cover these 25 cities, based on the assumptions about density and cable architecture.
The 5G latency and capacity requirements suggest that fibre will be necessary for backhaul links. Backhaul speeds of 100 Gbps are likely, which are impossible with twisted pair copper. The backhaul link might be done with microwave or millimeter wave wireless systems, but this will necessitate frequency licenses and line-of-sight engineering. Most carriers have acknowledged that fibre is likely to be most effective for backhaul.
For the previous generations, most mobile network operators have used coaxial cable for fronthaul applications. In recent years, however, there have been advances in miniaturizing the digital-to-analog electronics so that some of the base station functions can be remoted to the radio head, and other functions can be hauled over fibre from further away. These developments mean that 5G cell sites may have new architectures – new approaches for optimizing the location of digital electronics and the radio heads. This in turn may mean that fibre can be used for both backhaul and fronthaul functions, along with conductors for powering the antennas. It also may mean that there will be increasing demand for hybrid cables, containing both fibres and conductors for power. Finally, the use of fibre on the towers also may lead some carriers or network-construction companies to consider bend-insensitive fibre in the cables, allowing for tighter bends and greater ease-of-handling.