Electronic Products & Technology

Balancing scalability, complexity in embedded systems

By Emily Newton, Editor-in-Chief, Revolutionized   

Electronics Embedded Systems Engineering architectures complexity digital-twin embedded IoT scalability systems

Digital twins aren’t an option in every case. However, successes show people should strongly consider exploring them.

Designers are frequently called upon for associated input as embedded systems become more commonplace. That may mean making tradeoffs in a product’s complexity if customers want architectural scalability in embedded systems. Conversely, a highly complex system is harder to mass-produce, which could limit its commercial appeal by making it too costly. Fortunately, electronics engineers and other design team members can create embedded system architectures that check both boxes. Here are some practical ways to do that.

Create a digital twin for easier designing

One possibility is to deploy a digital twin when working with potential designs. It helps people determine which options will most likely get the desired results. They can run various scenarios via the digital twin, then get valuable takeaways about how embedded systems will perform in real life.

Engineers could see specifics, such as when the system becomes too complex to remain easily scalable. Alternatively, they could tweak various aspects to see which are most beneficial for retaining complexity while aiding scalability.

The convenient thing about a digital twin is that people can run several scenarios in relatively short time frames, thereby accelerating the entire design process. Digital twins can also show the likely impacts of adding specific features or changing design parts. This leads to better visibility and fewer surprises.


An example of what’s possible came from a University of York team. They built digital twins for embedded systems to answer key questions. Some related to identifying the quality of digital twins and understanding when they needed improvements to achieve maximum effectiveness.

Source: Adobe

In this case, the designers focused on developing digital twins for complex on real-time embedded systems. They recognized that a best practice is to determine the worst-case execution time for a planned device. However, that approach lacks adaptability and fault-mitigation potential, leading the team to look for alternative methods.

This research centered on the hope that digital twins could allow people to reduce inaccuracies in their embedded systems while making valuable observations that lead to improvements. They also explored how to continually improve the model with a feedback loop. The results showed the digital twin could enhance embedded system architectures by giving users better planning potential. The researchers also hoped to further develop it by adding new features.

Digital twins aren’t an option in every case. However, successes show people should strongly consider exploring them.

Plan around AI in embedded system architectures

Achieving architectural scalability in embedded systems means accounting for all relevant aspects early on and continually until completion. Sometimes, that means choosing appropriate IoT sensors for planned use cases. It’s also increasingly common for an embedded system to include an artificial intelligence algorithm. Additionally, some design teams use artificial intelligence as a focal point when designing for scalability and complexity.

Researchers from the Wuchang Institute of Technology took that approach with an embedded system. They made hardware and software-related choices while considering AI requirements. They believed it would allow them to create a highly complex but scalable system well-suited to demanding needs. The team included a multitasking, real-time processing platform for their embedded system.

The outcome of the AI-centric design showed highly optimized results associated with the project. The algorithm steered many associated decisions, enabling designers to focus on the possibilities that were most likely to work. Doing that saved them time while boosting overall performance.

As the team analyzed how their AI-driven system performed, they found that the algorithm could simulate the real-world environment, similar to how the digital twin did in the previous example. The group also found that AI made it easier and more effective for them to test the embedded system.

Stringent testing is a vital part of scrutinizing all embedded system architectures. That means examining the three main components of electromagnetic compatibility, testing the system in various environments and more. However, this example highlights how AI could be a foundational part of designing for architectural scalability in embedded systems and ensuring the finished system has the desired capabilities.

Explore different design strategies

A McKinsey report about embedded systems complexity showed it can vary tremendously based on the industry needing such technology. For example, the typical car contains 100 control units and thousands of software components. The associated subsystems exchange tens of thousands of signals as the automobile operates.

Thus, people who build embedded systems must always envision how future customers will need to use them. The requirements for an embedded system in an automobile are much more intensive than for an application that does not require such frequent data transmission and vast amounts of information processed.

In other cases, engineers and designers must select less high-tech options than those described above. One possibility is to overdesign the system so it’s more likely to handle future needs that designers cannot necessarily anticipate.

However, that costly approach doesn’t always achieve its intended purpose. That’s because it’s often not specific enough, and it may mean designers build embedded system architectures with unnecessary features and capabilities.

An alternative is to list the embedded system’s ultimate requirements and split them into groups associated with low, middle and high-end users. That will help designers prioritize what they need to work on immediately while catering to the largest number of potential customers.

However, this may mean designers must create multiple systems. Budgets and timelines don’t necessarily allow for that option. Additionally, this strategy could make getting products on the market take longer.

People assessing scalability must ask how quickly they intend to scale and to what extent. Will the products only be available in one country or region or several? Achieving better architectural scalability in embedded systems goes more smoothly when individuals consider the right questions.

Proceeding thoughtfully for architectural scalability in embedded systems

The main thing to remember is that there’s no single best way to accomplish scalability and complexity. People facing the issue must carefully consider all factors before proceeding. That entails assessing budgets, must-have features and how the embedded system will function.

Embedded system architectures are evolving, and as they become more complex, people must not overlook that the associated products must be scalable, too. The architectural scalability in embedded systems is essential for marketplace feasibility and wider adoption among target customers.


Emily Newton is an industrial journalist with over seven years of experience writing technical articles for the manufacturing, engineering and electronics sectors.

Emily Newton, Editor-in-Chief, Revolutionized


Stories continue below