The 5G network will support both To C and To B interconnection of all things, and the complexity and challenges are unprecedented. To meet these challenges, 5G needs to deeply integrate new technologies such as cloud and AI, and explore overall solutions such as cloud-network integration and network-industry collaboration.

At the fourth “i-RRM Wireless Intelligent Management and Control Technology Seminar” of the “i Wireless-Intelligent 5G Technology Series Seminars” held recently, Sun Qi, a senior researcher at the Institute of Wireless and Terminal Technology of China Mobile Research Institute, said that i- RRM is an important part of the “i Wireless-Intelligent 5G” concept. By introducing intelligent management and control algorithms, the innovation capability of wireless network technology is improved.

Sun Qi pointed out that 5G, as the first network oriented to both To C and To B, faces many challenges: First, it is “difficult” to guarantee deterministic wireless SLA. , jitter, and reliability requirements are challenging; second, the coordination requirements of the complex networking environment of multi-mode, multi-band, and multi-vendor on the existing network are “high”, and it is difficult to achieve optimization through simple algorithms; third, the flexibility of network configuration can be optimized “complex”, The 5G protocol and air interface features are more flexible, and the process is becoming more and more complex. It is necessary to expand new dimensions and adopt new algorithms to achieve active, refined and in-depth collaboration among networks, services, and users.

Sun Qi further pointed out that i-RRM (intelligent management and control) can effectively improve the efficiency of wireless network operations.

In terms of intelligent wireless SLA guarantee, industry scenarios have a high demand for deterministic networks, and wireless closed-loop SLA guarantee can be realized through “configuration”-“watch”-“control”. “Provisioning” is demand import/network configuration, industry business classification and classification, business requirements are automatically mapped to wireless network behavior logic; “watching” is online monitoring, which can be visualized and sensed to perceive network status, support anomaly detection and fault warning, based on Network data provides business KQI prediction; “control” refers to closed-loop adjustment/capability opening, which supports real-time network adjustment and service assurance based on closed-loop SLA monitoring, wireless capability opening, supports industry self-operation and maintenance, and realizes network industry collaboration.

In terms of multi-network collaboration, 4G/5G will coexist for a long time to jointly provide customers with high-quality wireless network services. For 1.8/1.9/2.6/4.9G and other multi-frequency point collaboration, it is necessary to introduce intelligent algorithms to achieve multi-frequency points and multi-standards Efficient load balancing, carrier selection and handover optimization under

In terms of intelligent scheduling, channel information optimization can take advantage of the nonlinear restoration of neural networks to restore the actual channel characteristics of the UE on the base station side and improve system throughput; modulation and coding method optimization can use supervised learning methods to establish a variety of wireless information and MCS mapping relationships , simplify the existing measurement-feedback process, and optimize the transmission efficiency of small packets; the scheduler optimization can realize a simple and flexible scheduling mechanism through artificial intelligence algorithms, improve user experience, and improve QoS indicators.

In terms of QoE service assurance, based on second-level and UE-level data, including wireless characteristics, transmission characteristics, service characteristics, and service quality label data, AI algorithms are used to analyze the impact of network and service history indicators on service quality, and predict service experience at the next moment . Based on the KQI prediction results, RRM parameter optimization suggestions are provided to the base station in real time to optimize resource scheduling and ensure service experience.

In terms of modulation and coding method optimization, the supervised learning algorithm is used to establish the connection between wireless indicators and MCS, simplify the wireless system’s decision-making chain for MCS, save CQI feedback overhead, provide fast and accurate modulation and coding methods, and improve the service experience of small packet users. According to the verification results, the gain is about 10% compared with the traditional method.

In addition, Sun Qi introduced that in terms of standardization, 3GPP carried out wireless + AI data collection enhancement and AI/ML functional framework and process research in Rel-17, and CCSA also launched 5G wireless network intelligence research related projects. In terms of test progress, in September 2020, China Mobile and its partners completed the field test of intelligent load balancing in Fujian, using wireless fingerprints to reduce the cost of inter-frequency measurement, improving load balancing efficiency and user experience speed; in December 2019, in Shanghai 5G The cloud VR QoE optimization test was completed in the field, effectively reducing service freezes and black borders; in 2018, the intelligent MCS optimization test was completed based on the data of the Hangzhou existing network, and the spectrum efficiency of users in medium and high-load cells increased by 10%.

In the future, China Mobile will continue to promote wireless intelligent testing and verification based on the needs of wireless network pain points, promote the maturity of technologies and standards, and promote widespread application. In 2019-2020, preliminary tests and verifications of intelligent load balancing, intelligent QoE optimization and intelligent MCS will be carried out. In 2021-2022, the test and verification of new intelligent features in vertical industries such as SLA guarantee will be further promoted, and the characteristics of wireless intelligent enterprise standards will be planned.

Responsible editor: gt

Leave a Reply

Your email address will not be published.