Ericsson has unveiled a company-wide approach to centrally position machine intelligence across its entire organization and customer operations.

This includes the introduction of artificial intelligence (AI) technologies that will enable networks to self-optimize, improve efficiency, and deliver optimal user experiences.

The complex reality of today’s telecommunications systems will only accelerate further with the introduction of 5G, IoT and ongoing industry digitalization. Machine intelligence, using machine learning and other AI technologies, is vital to handling this complexity with more efficiency.

Machine intelligence benefit examples:

  • By implementing AI technologies, Ericsson halves the handover time between cells – resulting in fewer dropped calls and shorter configuration and maintenance time for service providers. Similarly, machine learning algorithms can predict traffic patterns and dynamically put cells into dormant mode without impacting user experience – achieving energy efficiency of up to 10 percent
  • By applying machine intelligence to prevent future malfunctions, Ericsson can provide actionable recommendations to the Network Operation Center personnel thus reducing dispatches of services technicians by up to 30 percent
  • Self-learning agents are applied in virtualized core (vMME) for world first VNF Autoscaling using deep reinforcement learning which is improving system performance with 25 percent compared to predefined thresholds
  • Expert Analytics is now equipped with machine intelligence pattern recognition thru deep-learning, enabling detection and optimization, making it possible to achieve up to 20 percent fewer network quality calls to customer care

Erik Ekudden, CTO, Ericsson, says: “With automation and domain specific AI, the intelligence built into the network platform provides superior performance while optimizing use of scarce radio network resources. We are developing Machine Intelligence solutions across our product portfolio and services to provide the highest performance and most intuitive and easy to use network operations for our customers.”

Ericsson Research has been strengthening competence in machine learning and AI since 2007 and the company now holds hundreds of patents in this area. These technologies include everything from basic machine learning to more advanced Deep Learning and Reinforcement Learning techniques, as well as chatbot-like functionality with NLP and image recognition capabilities.