Autonomous Intelligent Networks (AIS)

Main article: Artificial intelligence (AI, Artificial intelligence, AI)

2021: Artificial intelligence will develop networks and learn how to earn

On November 17, 2021, Rostelecom, together with Huawei, presented the study "Autonomous Networks: Analysis of Technologies, Efficiency and Applications," prepared jointly with the analytical company J'son & Partners Consulting. The study confirms that in the long term there is no alternative to fully automating the development and management of networks at all levels, which involves the total replacement of key processes with autonomous and intelligent ones. The consumer will play a decisive role in the production and business processes of operators, who will himself manage the parameters and price of the service. And artificial intelligence (AI) will begin to offer personalized services and tariffs for the needs of the client, and at a specific time and in a specific place.

According to the company, autonomous networks are controlled using software applications using artificial intelligence algorithms. While networks are configured by engineers using basic automation tools, such as tips and recommendations for changing power parameters, network coverage, or data rates. But soon they will be almost completely controlled by AI and human intervention will be required only in exceptional cases. The need to increase autonomy is associated with the rapidly growing complexity of organizing and managing communication networks. If in the fourth generation (LTE) networks the number of settings was measured in dozens, then in the fifth generation (5G) networks these parameters became many times more, so that they cannot be controlled manually.

According to J'son & Partners Consulting, the economic effect of implementing the concept of autonomous intelligent networks (AIS) on a global scale could exceed $800 billion per year if operators go to the L4/L5 level (highest in the classification), which implies complete autonomy (L0 level - all processes are only under human control). The amount includes both savings on operating costs (for example, intelligent energy saving systems allow you to optimize your energy consumption) and the additional revenue that comes from optimizing resource flexibility and value-added services.

The higher the degree of automation, the greater the economic effect. When switching to partial automation (L1/L2), the effect is estimated at "only" $6 billion, and at the level of L3/L4 - already at $149 billion. In conditions where the revenue from basic telecom services does not increase, and the opportunities for reducing operating costs are exhausted, the transition to AIS is actually the only option for business growth and profitability of operators.

In the study, the future functioning of telecom networks is compared with how cloud services are arranged. The market for cloud computing services (services that monetize computing, not communication infrastructure) is growing. Gartner Research Company estimates the average annual growth (CAGR) of the global IaaS and PaaS market at 33% during the 2019-2022 years. But the cloud model requires a high level of automation of all production and business processes. According to analysts, if operators learn to provide network infrastructure according to the model of cloud service (the so-called Network as a Service, NaaS), then they are likely to be able to overcome stagnation and begin to increase revenue in this segment at a double-digit pace over the years.

According to J'son & Partners Consulting, if the concept of distributed cloud computing is commercially successful, the global NaaS market could reach $76 billion in 2022, and grow to $600 billion in 2025. And with the successful digital transformation of the sectors of the real sector of the world economy, the volume of consumption of NaaS services may exceed $900 billion in 2030.

The transition to autonomy and the widespread use of artificial intelligence technologies dictates a change in service requirements and an increase in the complexity of networks. Managing infrastructure and processes becomes more difficult - and the further, the faster. If in LTE the number of parameters for network configuration was measured in dozens, then in 5G there are already hundreds of them and each of them can be critical for a particular user at a given place and at a given time. Making these settings manually is not just difficult - it is impossible. Our study shows that as of November 2021, it is understood by the concept of AIS and at what stages of its implementation are world operators. Some traditional operators have already begun the transition to network autonomy, but most are not yet ready for these changes, which means that they risk ceding the market to more flexible and ambitious newcomers. noted Pavel Sirotinkin, senior marketing manager for Huawei in the Eurasia region

Increasing the use of digital technologies and increasing user requirements are changing approaches to the development of services and tariffs, which are largely becoming individual. In turn, due to the growing number and characteristics of services, management becomes more difficult. Traditional services no longer provide operators with tangible revenue growth, so you need the services that customers expect. And they require a different quality of process control at all levels - the transition from manual control to artificial intelligence. A person will determine goals, and "machines" will have to find ways to achieve them. said Alexey Mitkin, Director of Strategic Programs and Innovative Development of Rostelecom