Focusing on the notion of digital transformation through a technical lens, the realization of digital transformation depends on the realization of smart connections by establishing the synergy between 5G and artificial intelligence.
Mobile operators should consider AI as the key means to objectify the capacity of ultra-broadband 5G connections. However, telecom operators have often made limited progress in deploying AI. Among the obstacles having adverse effects in utilizing AI, the lack of strategic prioritization, the talent gap, limited investment, and regulatory restrictions can be mentioned. In this article, we would like to give some practical suggestions for AI for operators while categorizing the AI use cases by telecom operators.
Artificial intelligence will change the way of doing work and has a significant economic impact on different industries in the near future. For instance, Mckinsey& Co. expects that there will be about 70% of companies that use at least one AI use-case by 2030 in the world and the technology will increase global GDP by about 1.2 percent annually by that time. This is mainly due to labor automation and increased innovation in products and services. On the other hand, research conducted by Accenture on 12 developed economies indicates that AI will double the global annual economic growth rate by 2035 [1-2].
Due to the opportunities that lie in the utilization of AI in network operations, customer experience, and the provision of new services to enterprise customers, the acceleration of artificial intelligence in the mobile industry is increasing. Network automation management and digitalization of interactions with customers are dominant use-case areas, while more agile and innovative operators are also using AI to launch new services and platforms like chatbots and AI-as-a-Service (AIaaS) [2-3].
We discuss the capacities of AI in terms of easing the service provisions and optimizing the routine procedures and/or non-routine queries, thus playing the role of the main driver for 5G development.
Mobile Generations have evolved but, what about network operators?
Despite the evolution of mobile generations and significant progress in providing broadband connectivity services to end-users and enterprise customers, telecom operators often have implemented limited levels of automation. In other words, the mobile operators have not come through mature and integrated solutions that incorporate real-time data to simultaneously lower the operational efficiency, and improve both connectivity and their customers’ experience. For example, according to the research carried out by Analysis Mason Institute in 2018 on 76 operators, more than 50% of them did not use almost any type of automation.
Meanwhile, considering the potential of artificial intelligence in reducing costs and creating new revenue streams; and the increasing technical complexity of 5G (and maybe 5G-Advanced in the near future), for example, Massive MIMO and uplink and downlink decoupling, together with new service requirements, more agility in the field of operation and maintenance is needed, which makes the traditional operation methods less effective for the operators  and .
How MNOs benefit from artificial intelligence?
Automation of network processes is one of the accessible artificial intelligence use cases which quickly leads to the optimization of IT processes and OPEX reduction. In addition, with new types of industrial customers willing to build their blocks on 5G, generating specialized services for them and improving the quality of their quality of experience, is vital to survive in a new era. In general, artificial intelligence can be positioned in two areas of operations and services in the business strategies of telecommunications operators (Figure 1) .
AI focused on actions: where AI is used to design, optimize and operate the network, and helps operators and industry owners to analyze data from network infrastructure and perform predictive maintenance. This enables mobile operators to perform proactive maintenance, reducing downtime and improving network performance.
AI focused on services: where the ultimate goal is to improve the customer experience. Examples of these applications are smart advertising, dynamic pricing, data-driven marketing and sales, preventing churn and deploying virtual assistants (such as Tobi, Orange, Deutsch Telecom chatbot, and Telefónica Aura).
Figure 1- Applications of artificial intelligence in two groups of services and operations 
Some of the operators employ AI in more developed means. For example, Implementation AI as a platform is one of the ways for dealing with professional enterprise customers in a cloud-based fashion. Exploiting the big data, generated by the subscribers’ usage and performing behavior analysis - while respecting the principles of privacy protection - is another way of generating new revenue streams in dealing with third party verticals in energy, urban services and mobility sectors.
AI: the 4th dimension of 5G and beyond
5G is well known for its three fundamental dimensions of enhanced mobile broadband (eMBB), ultra-reliable and low-latency communications (URLLC) and massive machine-type communications (mMTC). These ‘three’, will play the orchestration role in technology, initiatives, and innovations in the mobile industry. However, the flourishing of 5G ecosystem depends on deploying AI as its fourth dimension which significantly improves the management efficiency of operators' operations (Figure 2) .
Figure 2- AI, the fourth dimension of the realization of the fifth generation 
In this regard, we explain some important applications of artificial intelligence in 5G networks:
The role of AI in 5G network planning
The entire process of network planning and configuration is widely recognized as the main technical complexity of operators which serves their core business, i.e., connectivity provision. A massive amount of engineering effort is required to ensure the proper site deployment, regardless of whether they are new sites or site upgrades. Especially with the rise of emerging use-case scenarios like the provision of 5G coverage in a sports stadium or ensuring low latency for a remote surgery application, taking advantage of automation to speed up network deployment and take leadership in the market becomes a key issue.
Leveraging AI to handle the site deployment automation is to provide an end-to-end process that includes the design of data exchange parameters on the radio platform, the determination of hardware in accordance with the technical specifications of the application, and speeding up the acceptance in test procedures.
However, this has not been realized yet, and there are gaps in the automated site deployment workflow that often require manual intervention. For example, coordinating between site design and site installation process workflows are among the operations that lead to the lengthening of the deployment process and sometimes cause unnecessary site visits, manpower costs, and falling behind schedule. Today, operators need to manage many parameters to manage the network planning process. Therefore, it is necessary for them to consider the implementation of simplified design and processing with the introduction of automation capabilities and the use of artificial intelligence technology in the industry .
Improving the back-end operations of the network
Monitoring and analysis of operational KPIs is a multi-dimensional, complex and critical issue for mobile operators. Almost all of the MNOs are handling the issue by the use of complicated organizational structures each of them is responsible for a particular type of issue within the network. By utilizing AI in different parts of this complicated and unstoppable process, MNOs can analyze network usage patterns, automate the root cause analysis of problems, decision-making, re-evaluation and confirmation processes, and optimize network resources to ensure optimal performance in terms of network speeds and reduce latency. In addition, other types of procedures like fraud detection, automated customer relationship management, offering personalized services or optimizing pricing strategies, can be performed by extensive use of AI-based data analysis of large volumes of customer data.
In the context of 5G, the use of AI in the back end becomes even more critical. This is because 5G networks are more complex than previous generations, with a higher number of connected devices and more advanced applications, which require the network to be flexible enough to adapt to users' expectations dynamics. The sheer volume of data generated by 5G networks will require advanced analytics tools to extract meaningful insights and identify patterns. AI-powered network optimization can help ensure that data is transmitted in the most efficient way possible, reducing latency and improving overall network performance. In addition, 5G will create new revenue opportunities for mobile operators, such as providing network slicing and offering new services to business-level customers, which will become overwhelming without using artificial intelligence.
Overall, the use of AI in the back end is essential for mobile operators to successfully provide 5G services. It can help them optimize their networks, reduce costs, enhance the customer experience, and identify new revenue opportunities.
Electricity consumption of the sites composes a high percentage of the network OPEX. Although network traffic drops significantly during idle hours, the equipment continues to operate in full power. In this regard, the ideal state of automation is to dynamically adjust power consumption based on data traffic flow to prevent the waste of energy. This requires predicting traffic usage patterns in real-time. For example, in the environment Multi-RAT, coordinated energy savings can be achieved through accurate traffic forecasting, without disrupting the user experience of its customers.
At present, most of the energy-saving solutions are based on non-dynamic shutdown systems. However, the energy-saving process should be dynamic by using other inputs like traffic forecasting and dynamic thresholding capabilities based on reinforcement learning .
The main goal of employing 5G technology is to grow the digital economy in all industries. In the meantime, by considering the pivotal role of mobile network operators, it should be accepted that the smart industries are not realized unless smart network operators arise. MNOs in order to respond to the new needs of users, should be more flexible and able to act as an enabling platform for industries and businesses in forming new partnerships. Achieving this maturity depends on the AI and 5G synergy level and the realization of smart connections.
Although this has been achieved at a limited and non-extensive level today, the realization of operational convergence requires the common understanding of all effective stakeholders and their convergent movements. The establishment and exploitation of AI is a multi-year path for operators, but the starting point of this path is ‘now’. It should be remembered that many factors slow down the deployment of AI and determine the success of operator strategies in using this technology. Some of these accelerating factors include the adoption and testing of AI in internal bureaucratic processes, presenting it to the market, adopting a customer-centric and data-centric approach with the aim of creating a competitive advantage, and adopting an approach based on open innovation.
The article is based on the authors' insights about telecom operators:
 Mckinsey & Company, The potential value of AI and how governments could look to capture it, 2022.
 European Parliamentary Research Service, Economic impacts of artificial intelligence (AI), 2019.
 PWC, What's the real value of AI for your business and how can you capitalize? , 2021.
 GSMA, AI & Automation , 2019.
 Deloitte, Fueling the AI transformation: Four key actions powering widespread value from AI, right now, 2022.
MSc in communications from Shahid Beheshti University, Tehran, Iran. She has done different projects related to the ICT era such as designing and implementing the LiFi system, implementing the Physical Layer authentication method using deep learning for IoT security matters, and investigating the necessity of using satellite technology and the potential of this technology in providing 5G telecommunication services and technical specifications of IoT services provided by LEO satellites. She started her professional career with MCI R&D Center as a technology strategist in 2021. In this institute, She has been performing different tasks related to technology management, including scouting new trends and technologies in ICT and Digital Transformation era and implementing technology roadmap.
Sayed Ali Khodam Hoseini
Ph.D. in communications from Shahed University, Tehran, Iran. He is a Smart City Project Manager at Mobile Communication Company of Iran (MCI). With a background of policymaking and regulation telecom ecosystem, he joined Smart Tehran Center in 2018 and provided consultancy services to the entity until 2022. Now, he is responsible for developing smart city projects and initiatives targeting the use of digital services and uptake of disruptive business models and technologies in MCI.