AI Model + 5G: Empowering Industry Intelligence

Since the second half of 2022, the popularity of AI models, represented by ChatGPT, has swept the world, signaling the official entrance of AI technology into the era of AI models. However, in 2023, the model development and applications have followed different trajectories. While the models are rapidly evolving, their applications have not kept pace, lagging behind the advancement of AI models. As a new technology, AI models strongly promote the development of basic intelligent science. However, their wide applications need to be integrated with the industry and other technologies. This paper discusses the new paradigm of applying AI models combined with 5G in various industries and explores the prospects and evolution of the “AI model + 5G” to promote the development of industry intelligence.

In China, the government held six consecutive sessions of the “Blooming Cup” 5G application competitions to promote in-depth integration of the digital economy and the real economy. Throughout this process, various issues have been exposed, such as the high threshold of 5G network technologies, maintenance and operation difficulties for industry users, and the failure to realize the value of network data. Additionally, model applications in the industry encounter numerous challenges, such as acquiring high-quality datasets and achieving rapid industrial deployment, which remain significant concerns.

The development of industry intelligence requires that AI model and 5G complement each other, with both being jointly promoted: AI model enables 5G and drives the digital and intelligent transformation of the industry, while 5G empowers AI model and accelerates its applications, as illustrated in Fig. 1.

AI Model Enables 5G

The challenges encountered in the development of 5G within the industry include complicated network O&M, unguaranteed service level agreements (SLAs), and insufficiently differentiated processing methods for specific services. These issues can be addressed by introducing AI models into the 5G network.

• AI Model Facilitates Intelligent 5G Network Operation

After industry customers deploy 5G networks, professional knowledge and personnel are required for service provisioning and routine O&M, thereby increasing network costs and investment. To mitigate this, AI models are applied in 5G O&M to achieve intelligent network operation, significantly reducing industry investment. These AI models aid industry customers in converting service requirements into network planning and configuration, enabling intent-driven services and improving service provisioning efficiency. By analyzing a vast amount of network logs and alarm data, AI models can identify or predict faults and provide corresponding solutions. Through in-depth analysis of customers’ behavior data, AI models can predict network requirements and service preferences, generate related service bundles, and facilitate better network operation.

• AI Model Guarantees Network SLA  

Applications in various industries have unique demands concerning network bandwidth and latency. For example, video surveillance services in the intelligent manufacturing field necessitate substantial bandwidth, while production control services prioritize low latency. Therefore, 5G networks need to provide SLA guarantees tailored to different users or services. On the core network side, AI models can be used to evaluate and predict user service quality and generate service guarantee policies in time. On the wireless side, these AI models analyze wireless signals and sensor data to implement more accurate resource allocation and scheduling, thereby enhancing network efficiency and quality.

• AI Model Enables Intelligent Service Analysis

In industrial application scenarios, stringent security protocols often dictate that data must remain confined within specific areas to mitigate potential risks. However, the emergence of advanced AI models and algorithms has heralded a transformative solution to this challenge. With the introduction of AI models and algorithms, the 5G network can analyze and intelligently classify data traffic in real time without compromising security protocols. This innovative approach enables industries not only to uphold confidentiality protocols but also to extract valuable insights from the data generated within their confines. By deploying AI models at the edge of the network, where data is generated and consumed, the system can harness the potential of intelligent service analysis. The integration of AI models into the 5G network empowers industries to tailor their services according to specific requirements, ensuring optimal performance and responsiveness. Through continuous monitoring and analysis, the system can dynamically adapt to changing conditions, providing differentiated service guarantees tailored to the unique needs of each industrial sector.

5G Empowers AI Model

As an information infrastructure in the digital era, the 5G network boasts high bandwidth, low latency, and massive access, greatly enhancing the industry’s data collection capabilities and enriching AI model datasets. At the same time, as 5G networks are widely used in industries, AI models will play a pivotal role in the digital and intelligent transformation of these industries.

• 5G Network Enhances AI Data Sets

In the industry intelligence process, data stands as the core element shaping the competitiveness of AI models. The 5G network supports multiple access modes like IP/LAN, overcoming mobility restrictions and facilitating access to voice, image, and video data from anywhere in the industry, thereby greatly enriching the datasets of AI models. In addition, the 5G network provides user authentication and transmission encryption technologies to ensure the security and effectiveness of data access, thus greatly enhancing the quality of AI model datasets.

• 5G Network Expands AI Model Application Scenarios

5G, as a dynamic innovation catalyst, is rapidly gaining ground across industries, expanding the application scenarios for AI models. On the one hand, through the deployment of 5G private networks, heterogeneous and massive connections of monitoring devices, such as sensors and cameras at front-line production sites, can be supported. These new scenarios also pose additional requirements for AI models in the industry. Leveraging the multi-mode analysis capabilities of AI models, intelligent and accurate fault warning and risk management can be achieved, significantly improving production efficiency. On the other hand, 5G networks also promote the collaboration among AI models. Currently, AI models primarily collaborate between the cloud and edge. With the applications of 5G networks, AI agents can be extended to 5G intelligent terminals to establish a comprehensive cloud-edge-end collaboration system.

Conclusion

To foster the industrial applications of “AI model + 5G”, ZTE has launched the AiCube all-in-one machine. The AiCube consists of computing hardware, a cloud platform, and an AI platform. The computing hardware is compatible with CPUs/GPUs from various manufacturers and models. The cloud platform implements unified resource management, allocates computing resources on demand, and can deploy both 5G networks and AI models. The AI platform provides operators and enterprise customers with tools such as data management, model development, model training, model inference, and application statistics to enhance usability and efficiency.

As a full-stack intelligent computing solution provider, ZTE will work closely with operators and industry customers to continually advance “AI model + 5G” applications, co-create a new intelligent computing ecosystem, embrace a future of intelligent computing, and inject new impetus into the development of the digital economy.