American Tower operates 178.8k towers and is redefining 5G network architecture in five distinct geographical regions. The company operates towers in the United States (40.6k), India (73.6k), Latin America (40.9k), Africa (18.8k) Europe (4.8k).
Every successive technology iteration, whether it was from 2G to 3G or 3G to 4G, was accompanied by tremendous gains. This included better processing power, spectral efficiency, and antenna technology. 5G’s evolution will enable the broad usage of newly developed technologies. Specifically, 5G uses new spectrum bands, edge computing, massive MIMO, dynamic spectrum sharing, self-optimizing networks and wider spectrum allocations. All these enhancements will help enable carriers to efficiently manage the cost of their rapidly increasing network demand. American Tower will help create this new 5G network architecture. Specifically, the company will do this by building-out an edge compute business, deploying massive MIMO and facilitating dynamic spectrum sharing.
Edge Computing – Network Architecture
Currently, the cloud resides in massive data centers with thousands of server racks, spread across rows and rows of electronic equipment. These data centers use significant quantities of electrical power inside a building, which is hundreds of thousands of square feet in size. Data center providers, such as Equinix and Digital Realty, build these cloud facilities and lease-out space to third-party customers that want to use the facility. Other cloud providers are proprietary, such as Amazon (through Amazon Web Services), Microsoft (through Microsoft Azure) and Google (through Google Cloud). These cloud providers have built or leased their own dedicated data centers.
Current data center configurations result in the latency between the user pressing something on their phone and getting a response, being ~100 milliseconds, using the cloud. While 100 milliseconds latency is satisfactory currently, in a 5G environment it will not be. Services such as augmented reality & virtual reality services, have much bigger data throughput requirements. In turn, these services require significantly lower latency capabilities.
For example, in use cases like autonomous vehicles, 100 milliseconds latency is too much. The signal must be transmitted back to the cloud, get processed and then be sent back to the vehicle to instruct a movement in a much shorter timeframe. Therefore, the old-fashioned, massive data center, cloud hosting system is not adequate to support the requirements of 5G. What is known as edge compute, will become a requirement in the next five years. Specifically, edge compute will allow the cloud to operate in a way that can support 5G.
Edge Computing – American Tower’s Vision of the Network Architecture
Edge compute is the deployment of servers and storage units in server racks very far forward. This means outside the cloud or beyond the cloud. Currently, edge compute is being deployed into local areas by carriers like AT&T and Verizon, in partnership with cloud providers Amazon, Microsoft and Google. AT&T and Verizon are using their legacy central office locations to house server racks on behalf of the cloud providers.
Eventually, these edge compute locations will move to the radio access network (RAN) interface. This resides at American Tower’s portfolio of towers and on-site buildings. American Tower’s network architecture vision is to match-up 2-3 cloud providers and 2-3 carriers at the company’s tower sites. Using American Tower’s infrastructure, cloud providers will be able to interconnect with carriers, have storage and have processing. Interconnection of multiple carriers and cloud providers at a tower site will enhance the latency, throughput, and responsiveness of 5G.
Applications that are 5G-supported will Require Minimal Latency
As innovative 5G-driven applications develop, there will be a further need for edge computing. For example, edge computing will support the establishment of vehicle management and control systems for autonomous drones and vehicles. Additionally, edge compute will eventually serve the needs of interactive & immersive media delivery and cloud gaming. In these use cases, lower latency is a requirement, thus data needs to be physically closer to the user’s machine.
In a 5G and Internet of Things-based world, there will be a need to have “capillarization” of storage and processing. This will facilitate latency reduction, cost reduction and enable the caching of data closer to customers. In turn, the user experience will improve. American Tower’s portfolio of tower sites will act as convergence points for this new network architecture. Specifically, convergence will occur between wireless access networks, cloud services, the Internet of Things and enterprise networks. Furthermore, American Tower’s assets are critical given the position of the tower at the edge of today’s mobile networks.
Beyond edge computing, American Tower is able to provide edge caching. Edge caching is the storage of content in facilities either physically at or much closer to, the base of the tower.
American Tower Buys-In to Edge Computing through Colo Atl
In April 2019, American Tower took their first step in building-out an edge compute business by purchasing Colo Atl. The company is a colocation data center business in downtown Atlanta, which was purchased for $75 million. Colo Atl is not a fully-fledged data center but rather an interconnect facility. Additionally, it is situated across the street from a large data center owned by Digital Realty.
American Tower is using Colo Atl as a proof-of-concept platform to connect tower sites to. Initially, it is testing this by connecting up six of its towers in the Southeast, United States. For example, in Jacksonville, Florida, American Tower built a 20-foot-by-20-foot edge data center module. This facility provides a cloud on-ramp for small- and mid-sized wireline customers to connect to the cloud service providers. In terms of cloud service providers, American Tower has partnered with Microsoft through their Azure Edge Zones program. The Microsoft partnership exists specifically to work on these specific edge compute solutions.
American Tower’s goal is to lead in the definition of infrastructure needed for edge compute solutions. Through American Tower’s portfolio of nearly 41,000 United States towers and almost 140,000 international towers, the company has the infrastructure in-place for storage, processing equipment, electricity, security (because towers usually reside in a fenced compound), and fiber optic cable / microwave backhaul. All of these capabilities are needed for edge computing.
In the United States, on average, two to three carriers are already tenants at each of American Tower’s cell sites. Therefore, American Tower can provide a neutral-host solution. Whereby instead of having just one operator isolated at its own central office and provisioning all of its own connectivity, American Tower can develop a multi-operator, multi-cloud, micro data center at its towers.
Massive MIMO – Network Architecture
Massive MIMO (multiple-input and multiple-output) is another enhancement to 5G technology. This enhancement brings enormous antennas (as big as a refrigerator) to be placed on towers. Overall, the footprint that the a massive MIMO antenna occupies, will be similar in size to the individual antennas which it will be replacing.
Massive MIMO antennas have begun to be deployed over the past couple years, with a lot more expected to be deployed near-term. Within a massive MIMO antenna, there are 25 to 100+ smaller antennas embedded into the panel. Each one of those antennas can use different frequencies and transmit their beam to a very specific location. Overall, Massive MIMO antennas offer a dramatic increase in capacity for carriers. Additionally they allow carriers to use frequencies in a much more efficient manner.
In a 5G environment, massive MIMO antennas are particularly applicable for carriers deploying mid-band spectrum. This is because mid-band spectrum requires smaller individual antennas, on a tower. Massive MIMO antennas cluster many of those smaller individual antennas, used for mid-band deployments, into one panel. Thus, it is possible for one singular antenna panel to be placed on the tower.
Overall, the singular antenna panels used to deploy mid-band spectrum, given massive MIMO, are going to be as large, if not larger, than the antennas used for 5G low-band spectrum. A specific on the 600 MHz frequency.
Dynamic Spectrum Sharing (DSS) – Network Architecture
Dynamic spectrum sharing is another important enhancement to 5G technology. For example, Verizon is deploying dynamic spectrum sharing through software upgrades on their radios. Dynamic spectrum sharing will help to accelerate certain 5G deployments. Specifically, because it allows carriers to be more spectrally efficient with their overall 4G/LTE and 5G spectrum holdings.