In today's data-driven world, the evolution of communication systems is closely tied to data traffic growth. This evolution not only influences innovations but also unlocks new markets and services. As the deployment of 5G networks continues worldwide, data-driven insights derived from traffic demand and analytics play a crucial role in shaping the future of 6G systems. An article published in the journal Sensors and Actuator Networks explored the significance of data-driven insights and discussed the potential evolution towards 6G by analyzing network evolution and data forecasts. Additionally, it highlighted the importance of technology trends and standards in driving this evolution and presented a case study on leveraging data analytics to enhance network operations and quality of service predictions.
Data traffic and network impact
The rise of smartphones has led to exponential growth in global data traffic demand. Forecasts indicate that the majority of mobile traffic will be supported by 4G networks in the next two years, with 5G networks gradually taking over from the following year. By 2027, 5G is projected to serve the majority of mobile subscriptions. These forecasts also reveal regional variations, with the Asian market, particularly China, playing a vital role in the global landscape. As data consumption continues to increase, the convergence of 5G with Wi-Fi and other technologies is becoming crucial to meet users' needs, especially in indoor environments.
Study: Unleashing the Potential of Data: Paving the Way for 6G Systems. Image credit: Photon photo / Shutterstock
Mobile data usage
Data consumption per smartphone is expected to surpass 15 GB globally by 2022, reaching 40 GB per month by 2027. Video usage remains a significant driver of data traffic, and it is projected to contribute to 79% of global data volume by 2027. In addition to cellular networks, Wi-Fi connections are gaining importance, particularly in indoor scenarios. The evolution towards 6G systems must consider multi-access convergence, combining cellular, Wi-Fi, and potentially satellite links to ensure seamless connectivity and meet the growing demand for data.
Technology trends and standards
Various technology trends and standards are emerging to support the evolution toward 6G. Convergence between 5G and Wi-Fi is being addressed at different levels, although widespread adoption is still underway. Additionally, the exploration of satellite links and their integration into communication networks is gaining interest, especially in industries like automotive. These trends reflect the continuous growth of data traffic and the need for innovative solutions to provide efficient and reliable connectivity.
Case study: Leveraging data for AI operations and QoS predictions
This paper presents a case study demonstrating the practical application of data collected from live networks. By leveraging data analytics, network operators can enhance AI operations and improve the design of services. The Hexa-X activity exemplifies how data-driven insights can inform decision-making processes, optimize network performance, and predict the quality of service.
Hexa-X is a collaborative initiative involving leading companies and institutions in the communication industry. It aims to drive the research and development of 6G systems by addressing key challenges and opportunities. One of its activities, Fed-XAI, focuses on integrating Federated Learning (FL) with explainable artificial intelligence (XAI) models.
Fed-XAI combines FL and XAI to enable collaborative learning of transparent AI models. FL allows multiple parties to train AI models locally using their private data and share the obtained models instead of raw data, preserving privacy while benefiting from collective experiences. XAI focuses on developing explainable AI models, providing insights into their decision-making processes. By integrating FL and XAI, Fed-XAI ensures privacy preservation and enhances trust in AI results.
Data plays a crucial role in driving the Fed-XAI innovation. Real-time, emulated network testbeds were created using data collected from live mobile networks. By incorporating realistic sources of live data, such as base station positions and user data volume, the prototype implementation of Fed-XAI demonstrated the reliability of its output. Using real data and live measurements enhances privacy preservation through FL and improves the trustworthiness of the models.
Combining live network data with simulations further strengthens the capabilities of Fed-XAI. The Simu5G simulator generated a dataset including quality of service (QoS) metrics from the mobile network. The scenarios learned by the training algorithms became more meaningful by configuring the simulated network topology based on actual network conditions and using data from live networks. This approach allowed the prediction of video quality during teleoperated driving scenarios, showcasing the practical applications of data-driven insights.
The Fed-XAI prototype involved an offline training phase and an online inference phase. For training, an XAI model based on the Takagi-Sugeno-Kang Fuzzy Rule-based system was trained using FL. The implementation employed the Intel Open FL library, extended to support FL of interpretable models. The real-time testbed, which included a video server and receiver, utilized Simu5G's network emulation capabilities to emulate the mobile network in real-time and predict video quality based on network conditions.
Conclusions
The power of data and analytics in driving network evolution towards 6G systems cannot be overstated. As data traffic continues to grow, insights derived from data analysis and forecasts play a pivotal role in shaping the future of communication networks. The evolution to 6G will require multi-access convergence, combining cellular, Wi-Fi, and potentially satellite technologies to meet the increasing demand for data connectivity. By harnessing the potential of data, network operators can improve network operations, enhance user experiences, and pave the way for innovative services in the era of 6G systems.
The case study on Fed-XAI exemplifies the potential of data-driven innovation in developing 6G networks. By integrating FL and XAI, Fed-XAI enhances AI models' privacy, trust, and explainability. The utilization of real data and live network measurements in the prototype implementation demonstrates the practicality and reliability of the approach. As the world progresses towards 6G networks, leveraging data innovatively, like in the Fed-XAI project, will play a crucial role in shaping the future of communication technology. With the power of data and analytics, network operators can optimize network performance, gain valuable insights into user behavior, and deliver exceptional services to meet the evolving needs of users in the era of 6G systems.