From an idea to a simulator – How we develop simulators at Magister Solutions
February 19, 2025
At Magister Solutions, simulator development has been at the core of our work since our company was first established in 2005. Initially focused on mobile network simulators, we have since expanded our expertise to simulate a wide range of use cases across the SatCom and space industries, as well. Over the years, we have created simulators for use cases such as European satellite constellation research, developing UAV satellite terminals, as well as comparing the performance of 5G-NR NTN and DVB technologies.
But what actually goes on in the daily work of Magisterians? How exactly do we design and develop these simulators?
Today, we’re going to take a peek “behind the curtain” of simulator development at Magister Solutions. We asked experts here at Magister to shed some light on the simulator development process. What tasks are involved, what’s the most challenging part, and how long does a simulator development project typically take? By reading this article, you will find answers to these questions, and more.
In this blog article:
1. Customer requests driving simulator development
2. Inside Magister’s simulator development process
3. Balancing accuracy and performance in simulator development
4. Assessing simulator functionality, performance and quality
5. Evolving simulators: Supporting network planning, 5G TN/NTN standardization, and industry evolution
1. Customer requests driving simulator development
First step – what’s the reason or need for a new simulator?
According to Timo Nihtilä, Principal Scientist at Magister Solutions, customer requests are usually the main driver for developing a new simulator.
For example: our customers might be involved in the standardization group of a certain technology. These groups often involve parties such as hardware vendors and operators. The group members are responsible for deciding how the technology in question should work. For example, how technology devices should communicate with each other.
Utilizing simulators is the cost-effective solution for these situations.
“It’s much cheaper to write software that simulates the technology, in comparison to building actual devices and forming a network with those to find out how it operates”, explains Nihtilä.
Magister’s customers also include large organizations in sectors such as satellite, space, and device manufacturing. For example, over the years, we have completed various simulator development projects for the European Space Agency. We’ve both collaborated with other organizations and delivered projects as the sole contractor.
2. Inside Magister’s simulator development process
The team involved in the development of a simulator varies depending on the customer. In the history of Magister Solutions, there have been teams of even ten or more people. On the other hand, some “teams” have only consisted of a single person.
There is usually no hierarchy in the roles of the team members. However, senior researchers are the ones with more responsibility, and they also guide and mentor the juniors.
“Even in bigger projects, Magister is usually fully responsible for our simulator development”, describes Henrik Martikainen, Principal Scientist at Magister Solutions. “Other companies offer input data, research, or might implement a different part of the system that our simulator is part of.”
“However, there have also been some projects in which another company has been responsible for one section of the simulator. In these cases, it’s crucial to have clear responsibilities and interfaces between the sections.”
Simulator development requires both independent and team work
Various kinds of work tasks are involved in the simulator development process, and it requires both independent and team work. These include, for example, the following:
- Planning and design
- Learning how to build the simulator
- Meetings
- Testing, debugging and finding errors
- Running simulations
The length of a simulator development project can range from a few months to a few years. The shorter projects only last for a month or two. In that case, the needed features are smaller, and the focus is usually on providing results.
New simulator development projects can take from one to two years. Projects that also involve research goals and reporting can take longer, even if the simulator already exists.
3. Balancing accuracy and performance in simulator development
According to Magisterians, the most challenging part of simulator development is related to the balance of implementation accuracy and performance of the simulation.
“When starting a new simulation project, it’s always important to keep the focus on the desired outcome of the project”, Nihtilä emphasizes. “What are the questions we want to find answers for in this project? What does the customer expect to get out of it?”
If the customer is interested in the bigger picture – for example, the overall network performance – we may want to implement the technology more broadly than accurately.
This means that we may choose not to implement a detailed physical layer of a technology with all the details, such as coding gains, antenna patterns and other detailed techniques. Doing so would require more calculations, leading to slower simulations. Additionally, a precise implementation may be slower to carry out, the simulation scenarios more complex to configure, and results more difficult to understand.
We can choose to model only the expected effect of these techniques on the data transmission rate. This saves precious computing power and eventually time.
On the other hand, if the customer is interested in the performance of an implementation detail, e.g. a patented physical layer modulation scheme, we have to model the scheme accurately to the physical layer to provide an answer. However, this also means that it will be more difficult, or at least slower, to evaluate the model on a large scale network when there are thousands of devices. In many cases, such a detailed level of modeling is not necessary.
Finding the right balance with experience
Thankfully, these challenges can be navigated with experience.
“With experience, our researchers are able to estimate which techniques are essential to model in detail to achieve adequately accurate results. We can evaluate which techniques can be modeled more broadly without compromising the accuracy of the results”, Nihtilä points out.
“When it comes to implementation balance, software development is often faster when shortcuts are taken. For example, by using models that capture the essentials rather than implementing every little detail”, Martikainen adds.
“However, this can raise the question of whether the model is appropriate and correct. It must be developed and validated to ensure reliability. In some cases, it’s easier to stick with established models and more detailed implementations, as they have already been researched and documented in specifications.”
4. Assessing simulator functionality, performance and quality
Magisterians use various technologies when developing simulators, including programming languages like C++ and Python, the network simulator ns-3, as well as telecommunications technologies such as 5G NR, NTN, DVB-RCS2, and DVB-S2X.
But how do we test the functionality and performance of the simulator we’re developing?
“For assessing simulator functionality, we utilize 3GPP and ITU cross company calibration”, illustrates Verneri Rönty, Researcher at Magister Solutions.
However, before comparing our results with those of other companies, we must examine the implemented features individually, and finally in a larger context. Here, the importance of expertise is highlighted again. One must understand how changing a single parameter affects the results, and confirm that it does. If not, the reason must be investigated. Comprehensive statistics are essential, and for this purpose, we have a versatile statistical framework that is used in different simulators.
These methods allow us to check that the output of the simulator – results, statistics, and visualization – is as correct as possible.
“Simulator performance as a software is also under continuous monitoring, starting from efficient design and using efficient programming methodologies”, Martikainen continues.
“We record simulation runtime and memory consumption as statistics when running functional or regression tests, or when running simulation campaigns. Finally, when we see performance issues with some feature, we use toolsets like Valgrind to see detailed memory and process CPU usage logs. They can be used to identify problematic implementations.”
High quality is one of the most important company values here at Magister. How do we ensure that it’s at the core of our operations and daily work?
“We never compromise the integrity and correctness of results, even when a deadline is quickly approaching”, Nihtilä affirms.
“We try to be experts in the field, not just a software house”, Martikainen states.
5. Evolving simulators: Supporting network planning, 5G TN/NTN standardization, and industry evolution
Over the span of our operations, we have provided simulators for various use cases and purposes.
Since the beginning, one of our main focuses has been to develop simulators for network deployment and configuration planning and verification. At first, this mostly centered around mobile network simulations. However, throughout the years, we’ve increasingly expanded our services to include the SatCom industry, as well.
With our simulators, our customers can design satellite communication networks, such as low earth orbit (LEO) constellations consisting of thousands of small satellites. Our simulators can be used for optimizing LEO satellite constellation protocols and performance. We’ve also developed a resource management algorithm that optimizes resource allocation in non-geostationary (NGSO) satellite constellations. The constellation can provide more capacity to peak times and areas, improving service quality.
Standardization has become one of the most important use cases for our simulators in recent years. Since joining the 3GPP, we have been part of the global 5G TN & NTN standardization process, which promotes higher quality systems and products in the telecom and SatCom fields. We have developed simulators for technology protocol evaluation (NR/NTN, DECT, etc.), radio frequency planning and coexistence.
For example, the EAGER project researched 5G-Advanced and 6G NTN technologies in the context of 3GPP standardization. We led the simulative analysis of the project, further enhancing our own simulators for the research of these future technologies. In the 5G-SPECTRA project, on the other hand, we developed simulation tools for evaluating TN/NTN coexistence, with the goal of examining the extent of adjacent TN/NTN channel interference in various scenarios.
Continuous development, research and marketing
One last question – what happens once Magister has completed a simulator development project?
“We use the simulator in future development projects, enhancing it further. We also utilize it for our own research purposes. Other purposes include selling and marketing the service. And repeat”, describes Martikainen.
That’s right! Most of our simulators have been developed in projects and further enhanced towards new use cases throughout the years. This reflects our commitment to constantly improving our solutions and services to keep up with the changing demands of the SatCom, telecommunications and space industries.
Interested in simulator development services?
Discover how our simulator development services can benefit you, and reach out today!
