September 9, 2021
When it comes to network planning, 5G has brought a set of entirely different challenges compared to the previous standard (4G/LTE). The task used to be simple: build a tower or find an existing one (like a church tower or roof) that a base station antenna can be installed on. 5G network architecture, on the other hand, requires much more preparation, data, and analysis. In this article, we'll answer:
A difference in data speed and latency between 5G and 4G is enormous, but what makes 5G NR fast also makes it much more troublesome to implement. And the reason for that is frequency. New generation networks are based on millimeter waves (mmWaves) up to even 52.6 GHz, while LTE is, in most cases, capped at around 2,600 MHz. The problem is that 5G NR high-frequency bands have a significantly shorter range and are much more greatly influenced by environmental aspects. While LTE antenna towers are able to cover kilometers of range, 5G mmWave antennas have to be installed within about 100-150 meters from one another. Therefore, depending on the required network parameters and the type of environment, it may take even a hundred separate devices to cover the same area with a sufficient 5G network as 4G/LTE one. Additionally, the placement of those devices is incomparably more challenging due to the characteristics of 5G mmWaves. The planning process is therefore also much more complicated compared to previous network standards. It requires an automated analysis of an extensive number of data sets.
In order to create an efficient 5G planning tool, we had to understand how exactly the propagation of mmWave behaves and what types of data we need to analyze in order to plan networks efficiently. We elaborate a bit more on that in our article "Understanding mmWaves as a gateway to seamless 5G network planning."
In short, previous cellular network generations didn't have to account much for factors like weather, road traffic, or even obstacles like trees or traffic signs. The millimeter wavelength range brings all of those challenges and many more. This makes 5G planning a complex process that requires massive amounts of data and testing, which can't practically be done without an automated, software-defined 5G network planning tool. Or at least it can't be done cost-effectively, which brings us to the next point.
The growing need for IoT-based solutions, more reliable connectivity, better user experiences, and autonomous vehicles, for example, dramatically increases the demand for 5G networks. Among other things, they are potentially much cheaper to implement for industrial or enterprise uses compared to optical fiber networks. One of the most important reasons why 5G networks are not widely implemented yet though, is the lack of highly effective network planning tools to address the challenges arising from mmWave characteristics. Solutions that were perfectly viable for 4G/LTE deployments can’t be used effectively to plan 5G NR networks due to their much higher frequency band and all of the phenomena and challenges that come with it.
For instance, 4G/LTE solutions rely only on static data that may include limited information about city architecture, but not much more than that. Without a comprehensive set of data about weather, vegetation, and other elements and parameters, planners would have to simply add more antennae to account for unexpected circumstances such as aggregation of traffic or extreme weather conditions. This would only increase deployment costs and further energy consumption, which stands in opposition to the modern idea of sustainable Smart Cities. But, of course, there is still a smart way to provide complete coverage for any area without wasting resources. Our tool brings 5G network planning to a whole new level.
A groundbreaking advancement for 5G network planning comes with the use of cloud computing and Digital Twins, which are real-time updated, virtual copies of objects or, as in this case, entire cities or regions. For example, in the test project LuxTurrim5G, we collected multiple types of spatial data to recreate a whole area with all of the elements and parameters in a 3D real-time environment. With the live-updated model, we were able to test multiple planning scenarios to find the cheapest and most efficient solution.
Once a Digital Twin model is created, the user has to mark all potential site candidates, which is one of the biggest challenges of building 5G architecture. Site candidates are all of the possible placements for 5G antennae in the area. They're not usually built from scratch, but are based on existing infrastructure. Some of the most common types are street lamp posts or the facades of municipal- and state-owned buildings, or other municipal infrastructure. Private-owned buildings and facilities on the other hand must be approved by the owners in order for them to be considered as site candidates. After establishing a sufficient number of site candidates, we’re able to let the tool do the work. In most cases, there are hundreds of thousands of possible various configurations.
For example, new generation network planning for a highway should seem like a pretty straightforward task, even using static data. And, of course, a significant part of a highway won’t be particularly complicated, but the task gets much more complex with multi-level highway junctions that are usually very different from one another. Each one requires a tailormade configuration, which can’t be made without the analysis of spatial data and traffic density. However, thanks to Digital Twins, we're able to thoroughly analyze every junction and identify potential congestion points.
Using cloud computing, a voxel propagation engine, and real-time data, we're able to simulate and test all of those setups and analyze network performance in search of the optimal configuration. What allows the tool to be accurate, fast, and efficient, is its utilization of machine learning and AI. Compared to other tools that are available on the market, the 5G radio network planning process can be up to 90% faster and 10 times cheaper with our next-gen tool.
Of course, taking into account terrain shape, buildings, vegetation, and other possible obstacles is critical, but that's not the only challenge. Perhaps the most crucial change of modern network planning is how we approach the entire premise of providing an assured level of service and the best experience for users and enterprises. It may sound surprising, but previous generation networks were designed in the opposite way. The exact network performance parameters and possible level of service were only known exactly after the deployment. Based on those results, cellular providers were able to establish what kind of service they were able to provide with the given network.
In case of 5G NR, the starting point of the network design process is deciding what will be its purpose in terms of the services that are to be provided through it and what parameters are required for the network to fulfill that purpose. If, for example, one is aiming to seamlessly deliver 8K ULTRA HD streaming for users, enable 5G-based traffic, and provide reliable mission-critical use, then they have to make sure to have proper network resources in the area. Each of those applications/services requires a different set of parameters and, therefore, different 5G network architecture. It is that which determines how densely the antennae need to be placed, what hardware is needed, and what radio standard will be used.
There are three defined “clean” deployment models for 5G NR depending on the planned usage.
eMBB (Enhanced Mobile Broadband) is the most widespread model at this stage, partly because the infrastructural requirements for its deployment are lower than with other models. eMBB networks must be sufficient for the widespread usage of AR/VR live media, Ultra HD, and 360-degree streaming. Those implementations focus on providing much higher data throughput (up to 10 Gbit/s) but don’t require a significant latency improvement compared to 4G/LTE.
URLLC (Ultra Reliable Low Latency Communications) is a model that requires a network to be sufficient for applications such as autonomous driving, industrial automation, or remote surgeries. URLLC networks must be able to reliably work with under 1 ms latency. At the same time, no more than 1 packet loss in 10⁵ packets is acceptable.
Massive Machine Type Communications (mMTC) is a case in which networks must be suitable to provide connectivity to a huge number of IoT devices. Those implementations prioritize low-energy consumption: mMTC networks must be able to support even 1 mln devices per 1 km², which is 10 times more than current 4G/LTE networks.
Depending on the purpose of a network, planners have to determine what set of network parameters is necessary to provide a given service or application to the users. Of course, those models describe the ideal parameters required for a given type of network usage. Real-life deployments will probably only come close to the one that planners aim toward. And, in some cases, there might be hybrid implementations as well, such as eMBB+URLLC.
The process of 5G radio architecture planning is, in our case, 100% cloud native in the SaaS model, which means we're responsible for all computing power. We start by collecting all necessary spatial data and feeding it into our Digital Twin Smart Map. Next, we identify areas that require additional bandwidth and choose site candidates. Most of the work afterward is automated thanks to our planning tool, which is based upon the mmWave propagation model.
What's crucial for the 5G network revolution is that it shouldn't be oriented only around mobile phones. It's also a necessary prerequisite for a number of Smart City applications based on thousands of IoT connected devices.
If you're interested in our Digital Twin-based 5G planning tool, make sure to read this article. You can also contact us using the form below.