The difference between 5G and previous generation networks is enormous, not only in terms of bandwidth and latency but also in terms of network planning. The use of high-frequency, short range millimeter waves makes the 5G network planning process a complex task.
One of the most significant new challenges comes with the different approach required for network infrastructure. Let's talk about what exactly 5G infrastructure is and how to approach it in terms of 5G network planning.
Network infrastructure means all of the software and hardware elements that allow a network to deliver Internet connections according to the set requirements. The newest generation comes with an entirely new set of challenges about infrastructure.
Before 5G, the process of cellular network planning was based on either of the two approaches:
1. Coverage planning was based on the range of antennas. In this case, crucial aspects included terrain shape and architecture.
2. Capacity planning was based on the number of users and their requirements. For this model, planners had to estimate how many people may use the network at the same time, estimate bandwidth per user, and avoid potential interferences.
After initial deployments, planners would analyze the performance, add new stations for load management, and disassemble those that turned out to be unnecessary in the large scheme of things.
The most significant aspect was planning positions of cell towers that would cover large areas with a range up to a few kilometers on flat terrain. Of course, mobile operators had to estimate the potential number of connected devices (population) and take into account terrain shape, buildings, and vegetation, but only to a limited degree. Compared to the new generation, that task was relatively straightforward.
In the case of 5G NR networks, the issue is far more complicated because they use much higher frequency spectrum bands, up to even 52,6 GHz millimeter waves (mmWave). Besides the ability to provide incredible network performance, the critical characteristic of the high-frequency spectrum is a much shorter range and increased susceptibility of received power due to enhanced environmental influence on mmWaves. Compared to the mid-band spectrum, mmWaves are more likely to be influenced by obstacles and factors such as heavy rain, snow, road traffic, vegetation, and many more.
Another key aspect that makes the issue of 5G radio infrastructure planning more challenging is the required set of parameters for a given network. They may include minimum requirements for bandwidth, latency, reliability, and availability. Before new generation wireless networks are planned, business leaders must define what exactly it is supposed to provide.
For example, if a cellular network must be sufficient for critical communications, autonomous driving, or even remote surgeries, it must reliably deliver a connection with under 1 ms latency. To ensure this, the network must meet certain requirements in terms of the radio standard and hardware. There’s absolutely no room for mistakes, sudden connection speed drops, or latency spikes in those cases. There are three defined deployment models that each have different sets of parameters depending on the purpose of the particular network:
eMBB (enhanced Mobile Broadband) allows a network to deliver AR/VR live media and 8K Ultra HD or 360-degree streaming. At this point of 5G evolution, this is the only model deployed on a larger scale, because eMBB networks can be deployed in the NSA (Non-Standalone) model, which means they can utilize existing LTE infrastructure.
URLLC (Ultra Reliable Low Latency Communications) is required for applications such as the aforementioned autonomous driving. This model requires SA (Standalone) deployment, which means the network must rely entirely on 5G infrastructure. So far, there are very few implementations in this model.
mMTC (Massive Machine Type Communications) means the network has to be compatible with massive IoT implementations. In such cases, low-energy consumption is highly prioritized.
Each of these deployment models has different requirements in terms of the hardware necessary for the network, base station density, and all other elements of 5G infrastructure. Effectively, whatever the designated purpose of the network is directly influences what kind of cell infrastructure it will need. The service- and usage-oriented approach for network planning makes service providers and planners rethink the entire strategy for deployment.
In terms of radio hardware, the essential network elements are base stations (gNodeB), which could be small cell transmitters called femtocells or picocells.
gNodeB base stations are fundamental for standalone 5G NR Massive-MIMO deployments. They operate in Sub-6GHz bands and further up to 52.6 GHz millimeter waves.
Small cells (femtocells and picocells) are smaller transmitters that operate on a very high frequency spectrum in small ranges. They are supposed to be unnoticeable in city environments. One example would be city lighting (so-called smart lamps). The more people (and devices) there are in an area, the more base stations have to be placed there. One of the critical infrastructure solutions that enable 5G NR networks to provide high-quality user experiences in densely populated areas is massive MIMO (multiple input, multiple output). This technology allows more people to connect to a given network simultaneously while sustaining satisfactory network parameters.
5G implementations can also depend initially on 4G/LTE infrastructure (non-standalone models), which can cover larger areas thanks to their range. However, because they operate in lower band frequencies like sub-6 GHz, they have limited capabilities in terms of bandwidth and latency. Their role in 5G infrastructure is most significant in less densely populated areas.
Besides the radio hardware elements of 5G infrastructure, there’s also the inherent element of power supply, which is especially important in terms of site candidates for base stations.
Additional aspects of infrastructure are determined by the type of 5G deployment, i.e. whether it’s fiber supported or full radio. In the first case, the infrastructure includes a fiber network and distributed servers.
Besides all of the elements mentioned above, there's another essential infrastructure aspect in 5G network planning: the use of existing city infrastructure as potential site candidates.
Site candidates are all potential placements for base stations. As we said, they could be city lamps, facades of publicly-owned buildings, bus stops, or traffic lights, for example.
Finding a sufficient number of site candidates is one of the key challenges for governments and enterprises that wish to implement 5G NR. In order to achieve the required set of parameters, there has to be a satisfactory number of site candidates in order to cover an area densely enough. This task gets more complicated in highly populated areas with many buildings and other structures that may block signals.
Every site candidate, especially on private property, must be formally accepted by the owner, and there has to be a power supply available. The process of finding enough site candidates to meet latency and bandwidth requirements is undoubtedly difficult, time-consuming, and full of obstacles. For example, the logical strategy to use city lamps as site candidates might not be so straightforward. City lamps generally have a centralized power supply that only works during evenings and nights, so they would have to be modernized to be able to support base stations 24/7.
Besides choosing site candidates, it's necessary to gather different sets of spatial data such as precise terrain shapes and BIM data (Building Information Modeling), which includes information about the exact materials used in structures in addition to diameters. The fragile nature of millimeter waves requires a very particular approach to those pieces of information. Some such data is publicly available, but there are some types of data sets that usually need to be provided by the client. Using this data, we create a live-update virtual model of a given area (Digital Twin) so that we can test different planning configurations.
The conservative approach that was entirely sufficient for 4G/LTE networks turns out to be almost useless for 5G NR networks, especially in major cities. The amount of data required for cost-effective 5G network planning is virtually impossible to analyze without far-reaching automation. In any environment, there are countless possible base station placement scenarios. Our planning tool utilizes solutions based on artificial intelligence and machine learning in order to find the best possible configurations that meet the required standards. Read more about how we're creating reliable 5G networks in a cost-effective way here.