As athletes pushed the boundaries of human potential during the Summer Olympic Games, technology—particularly advanced 5G networks—played a pivotal role in capturing, analyzing, and sharing those incredible moments. In an exclusive interview with Telecom Review, SmartCIC CEO, Toby Forman, explained how the company stayed at the forefront of 5G technology by testing and predicting network performance to ensure seamless connectivity during the event.
By collecting extensive data and applying advanced AI and machine learning (ML) capabilities, SmartCIC set a new standard in 5G performance, supporting the high demands of the Olympics and enhancing the overall experience.
Could you elucidate SmartCIC’s role, and describe how the company became intricately involved in the Olympic Games?
SmartCIC was founded about 15 years ago by my Co-Founder, Catherine Hemingray, and myself. Our core business focuses on providing professional services as a connectivity aggregator and field services. We cater primarily to large global telecommunications companies and systems integrators, offering connectivity and field operations solutions.
Over the years, we have developed a strong expertise in fixed wireless access by leveraging publicly available cellular networks. Our work has included deploying these technologies in challenging environments such as airports and railway stations, where traditional 4G networks often need to catch up.
As 5G technology evolved, particularly in Europe, our customers sought our help to effectively understand and utilize these next-generation networks. They needed detailed insights to share with their enterprise clients about how to maximize the potential of 5G. We initially explored creating an app to gather performance data from crowdsourced sources to address this. However, we soon realized this approach needed more control and granularity for precise analysis.
This led us to develop our specialized hardware platform. This platform enables us to run detailed tests and collect extensive data, including radio frequency information and performance metrics such as download and upload speeds, latency, and jitter. We have gathered 65 million geo-positioned data points for accurate, comparative analysis. This robust data collection helps us understand how different networks perform in real time and supports our AI-driven methodologies for predicting network behavior.
Our involvement in the Olympic Games was a natural extension of our efforts to refine and validate our methodologies. We saw the Olympics as a unique opportunity to test how networks handle the high demand from international visitors. By driving around Paris and collecting data during the event, we could assess network performance under the pressure of large-scale usage, further enhancing our understanding and the value we provide to our customers. It was an excellent opportunity to find out what the networks were doing to cope with the influx of visitors from all nations supporting the Olympics.
Our work is not confined to a single location, but spans across various cities, including Washington, D.C.; Paris, France; Barcelona, and Madrid, Spain. This international testing is a testament to our commitment to continuously improving our AI models and methodologies. It ensures that we deliver accurate and actionable insights for enterprises looking to optimize their next-generation networks, regardless of their location or scale.
Did the Olympic Games boost your visibility as a company?
Significant events like the Olympics are pivotal, especially given that SmartCIC is still in a relatively early growth stage. These high-profile events help raise our company’s profile and provide us with invaluable opportunities to gather and analyze data.
The real value lies in understanding how networks perform under the intense demands of such large-scale events. We’re particularly focused on examining how the network capacity provided by operators supports the various applications used by attendees and other stakeholders. This insight not only helps us fine-tune our models and technologies but also showcases our adaptability, ensuring they are robust and effective in real-world scenarios.
In essence, while the increased visibility is beneficial, the key advantage we gain from participating in these events is the chance to delve deeply into network performance and application interaction. This data-driven approach allows us to enhance our offerings and better serve our clients in future projects.
As stated above, data is crucial for testing. SmartCIC collected data from over 10 million data points throughout a week of drive-testing Bouygues Telecom, Free and Orange 5G networks ahead of the Olympics. What challenges did you face during the data collection and testing phases?
Collecting data from over 10 million data points presented several challenges, particularly given Paris’s high density and unique layout. While our systems were well-equipped to handle large volumes of data, the real challenge lay in the practical aspects of data collection and the city’s distinctive characteristics.
One significant difficulty was adapting our models and methodologies to fit Paris’s radial street layout. Unlike cities with a grid system, such as those in the U.S. or Barcelona, Paris’s streets have evolved over centuries in a more organic pattern. This radial development made it challenging to design effective driving routes and ensure comprehensive coverage for data collection.
Our traditional approach, often based on grid systems, had to be rethought to account for Paris’s historical and irregular layout. This necessitated the development of new strategies for driving routes and data collection plans that could accurately reflect the city’s unique geography.
In addition to the logistical complexities, integrating and analyzing the massive amounts of collected data was another significant task. We gathered over 10 million radio frequencies and three million performance data points. Combining these datasets effectively and creating a coherent story for specific geographical locations required advanced clustering technologies and AI models. We employed machine learning (ML) and neural networks to process and interpret this data, which allowed us to gain valuable insights into network performance.
Adapting our technology and methodologies to Paris’s distinctive urban landscape was not just a challenge but a crucial learning experience. It underscored the need for flexible and sophisticated approaches when testing networks in diverse cities, and it has prepared us for future projects in other European cities with their unique layouts.
Can you elaborate on how SmartCIC predictive analytics achieved a 96% accuracy in forecasting 5G network performance for the 2024 Olympics in Paris?
We achieved 96% accuracy from a download perspective, 94% accuracy and 96% accuracy for uploads, 90 calls for downloads, and over 90% latency—these results reflect the sophisticated algorithms and models employed by the company.
To achieve this level of precision, SmartCIC utilized a combination of real-world data and predictive models. First, we collected physical data from testing equipment, which provided actual values for downloads, uploads, and latency. This empirical data was then compared with the predictions generated by the algorithms. The process involved meticulously aligning recorded values with forecasted outcomes, ensuring minimal deviations.
The innovation hub of SmartCIC, which is located in Barcelona, played a crucial role in refining these predictions.
Taking your Spanish presence into account, how did the hubs in Barcelona help to achieve such high predictive accuracy?
SmartCIC Technologies has made a significant investment in Barcelona, a city with a rich pool of technical talent and strong educational institutions. While our headquarters is in Cannes, in the South of France, we’ve recognized the strategic importance of Barcelona for our global operations. It was here that analysts worked with detailed spreadsheets, where one column showed the real-world data, and another displayed the projections. They fine-tuned their models by analyzing the differences between these columns to enhance accuracy. The city’s technical talent and educational institutions have played a crucial role in our growth, and this rigorous comparison and adjustment process was vital to achieving the high predictive accuracy reported.
Our Global Services division has established a base in Barcelona, which includes a warehouse, support staff, and technical resources. The decision to invest in this city was driven by the exceptional talent available and the strong partnerships we’ve formed with local technical universities, such as the Universitat Autònoma de Barcelona (UAB). These institutions have been instrumental in helping us recruit both interns and full-time employees who bring valuable knowledge and enthusiasm to our projects.
Currently, our innovation hub in Barcelona focuses heavily on developing our predictive analytics capabilities, particularly for next-generation networks. This hub is crucial in advancing our technology and exploring new opportunities within the connectivity space.
Looking ahead, the Barcelona hub will have an even broader scope, contributing to various innovative projects and solutions. This investment enhances our ability to develop cutting-edge technologies and ensures that we stay at the forefront of advancements in connectivity.
However, Barcelona and our Global Services division operate in over 200 territories worldwide, reflecting our truly international reach.
We have a dedicated team of nearly 60 people spread across various countries. While our main offices are in France and Spain, we also have team members in Poland, Germany, the UK, the Philippines, India, West Africa, Latin America, and North America.
Our commitment to serving international markets is unwavering and is a cornerstone of our strategy. Both my co-founder and I have always been committed to leveraging our technology to meet the needs of a diverse customer base. It’s an exciting opportunity for us to understand and optimize networks on a global scale, ensuring we can provide top-notch solutions wherever our clients are located.
What were the key factors that led to Bouygues Telecom being identified as the top performer in central Paris, and how does this compare with the performance of Free and Orange across different areas of the city?
Bouygues Telecom stood out as the top performer in central Paris due to its exceptionally well-implemented 5G network in that area. The company’s performance was notably superior to Free and Orange’s competitors, particularly within the city’s heart. In our testing, Bouygues consistently ranked in the top 10% across various performance metrics, meaning that in every 10% segment of our sample, Bouygues was in the leading position.
This strong performance was expected because Bouygues had focused heavily on rolling out a high-quality, standalone 5G network in central Paris. This area was crucial for the Olympics, which likely influenced Bouygues’s decision to invest heavily there.
In contrast, when we moved out of central Paris into the suburbs, the performance of Bouygues and the other carriers, including Free and Orange, significantly decreased. This drop was anticipated because the networks in these suburban areas are less advanced, with more reliance on non-standard 5G or older 4G LTE technologies.
The key takeaway is that while Bouygues Telecom excelled in central Paris due to its advanced network setup and strategic investments, other city areas showed more performance variability. This pattern is not unique to Paris but is also seen in other cities like Barcelona and Madrid. Different carriers are at varying stages of network development and investment, often prioritizing central areas or locations with high returns on investment.
Bouygues Telecom’s focus on high-performance 5G infrastructure in central Paris gave them a clear edge in our assessments. This success underscores the importance of targeted investment and development in critical areas, a strategy that can significantly impact a company’s performance.
Based on real-world testing, enterprises and visitors to the Olympic Games saw Bouygues Telecom delivering the fastest maximum download (661.6 Mbps) and upload speeds (161.62 Mbps) in central Paris. Could you please elaborate on the implementation of these tests?
Real-world testing involves evaluating network performance in the actual environments where users will be rather than just examining theoretical or lab-based scenarios. For us, this means going beyond simple drive tests or focusing solely on the radio frequency (RF) environment. While these methods provide some insights, they must capture how users experience the network entirely.
In our testing, we simulate the behavior of real users using their devices to connect to services through the network. This approach helps us understand how the network performs under everyday conditions. We consider various factors, such as the built environment, reflective surfaces, and materials that can absorb signals. These elements all affect network performance, and by including them in our tests, we get a more accurate picture of how the network will behave in real-world scenarios.
The data we gather from these tests, including geolocation and other metadata, is not just for the present; it serves as a foundation for building detailed network performance models that reflect current performance and help us predict the future performance of various applications on the network. As we continue, we are committed to using this data to enhance our predictive analytics, enabling us to forecast the performance of various applications on the network.
When we observed Bouygues Telecom delivering maximum download speeds of 661.6 Mbps and upload speeds of 161.62 Mbps in central Paris, it was a direct result of this comprehensive real-world testing approach. This method ensures that the reported speeds accurately represent the user experience, factoring in all the complexities of the natural environment.
Given the insights gained from your testing, what role do you see artificial intelligence (AI) and machine learning (ML) playing in future network performance prediction and optimization?
Artificial intelligence and machine learning are not just a part, but rather the central pillars of the evolution of network performance prediction and optimization. Our technology business model is firmly built around leveraging these advanced tools to enhance how we understand and manage network performance, providing a solid foundation for our future endeavors.
AI plays a crucial role in creating detailed network behavior models. One of the critical approaches we’re developing involves using AI to build digital twins—virtual representations of real-world networks. These digital twins enable us to simulate and analyze network performance in various scenarios, helping us answer critical questions about how networks will behave in different environments.
From an enterprise perspective, we aim to provide actionable insights and information that enterprises can use to optimize their network usage. This involves creating and refining models that capture the complexities of network performance, both inside and outside of built environments. By building a comprehensive data lake and continuously applying and improving our analytical methodologies, we ensure that our models are robust and accurate.
Currently, we are in the phase of conducting a thorough analytical review of our models to prepare for the next stages of development. This ongoing refinement process, a testament to our commitment to excellence, is essential to ensure that the tools and insights we provide are not just reliable, but also valuable for carriers and their enterprise customers.
In summary, AI and ML are driving the next generation of network performance tools, enabling us to predict and optimize network behavior more effectively and provide deeper insights to help enterprises fully leverage their networks.
