Towards AI-driven, Cognitive Network Management:

Predictive Anomaly Pattern Recognition for 5G

Nokia, a main player in the telecommunication industry, wants to revolutionise network management with Artificial Intelligence (AI) technology. You can help making this ambition real as an EIT Digital Industrial doctorate student. You will develop an effective solution for predictive anomaly pattern recognition to apply in telecommunication networks.

Today’s communication networks are extremely complex systems consisting of countless network elements organised in cooperating, coexisting and overlapping technology layers. The network elements generate huge amounts of versatile data for performance monitoring, optimisation and troubleshooting purposes. It is already quite cumbersome to tackle these tasks with traditional approaches and the effects will be even more emphasised in case of 5G.  Therefore, Nokia is committed to deliver automated solutions that are capable to analyse the raw data and draw conclusions, generate actionable insights using AI. One important area in this field is the predictive detection of anomaly patterns that appear in the data. The main theme of this thesis is to develop an effective solution for this.


There are three big challenges to face. The performance data produced by live networks reflects the basic nature of the system that produced the data: it is complex, versatile, often unstructured and has huge volumes. To develop a successful anomaly pattern recognition solution, it is inevitable to deeply understand the source of the data. Moreover, there is no expert knowledge available to serve as a reference point of how the observed systems should behave. In most cases there is no ground truth that would set a reference point except statistical probability. Thus, the main reason that anomaly pattern recognition is a challenging task is that  it must be solved in an unsupervised way. Another big challenge is the strong pressure from Nokia’s customers like the telecom operators for predictive solutions that are capable to reliable detect early symptoms of failures and send notification before serious degradations occur. The third big challenge is that the analysed systems are non-stationary. Solutions need to be capable to continuously learn new information contained in the data. The results need to be easily interpretable by domain experts without data science knowledge. The system needs to be scalable and easily deployable. Textbook solutions do not work.


First the state of the art should be reviewed along with Nokia’s current internal understanding and status of the problem. Then, the new concepts should be formulated, implemented and verified with real network data from several real networks and domains. Initially, the AI,machine learning and concept development part should be in focus, before implementing the ideas as a scalable solution on decent hardware and with best-in-class big data technology. Cooperation with Nokia engineers, researchers and engineers of Nokia’s customers is expected. The doctoral student will have a chance to get insight of unparalleled depth about live telecommunication network.

Expected outcome

The expected results of the industrial doctorate are new solutions to predictive anomaly pattern recognition. The results shall include:

  • Working, verified prototype of the new concepts
  • Detailed analysis of the impact on network performance and deployment considerations
  • Published papers describing the findings in high-quality academic journals
  • Potential patents, working in close cooperation with Nokia Solution Networks Kft.


The doctoral student involved in this industrial doctorate programme, will share its time between the Co-Location Center of the EIT Digital Budapest Node, the premises of Nokia Solutions and Networks Kft. in Hungary, and the Budapest university of Technology and Economics.


  • Industrial partner: Nokia Solutions and Networks Kft.
  • Academic/research partner: Budapest University of Technology and Economics
  • Number of available PhD positions: 1
  • Duration: 4 years 
  • This PhD will be funded by EIT Digital, Budapest University of Technology and Economics, and Nokia Solutions and Networks Kft.


Those interested in applying should send an e-mail to, including a CV, a motivation letter, and documents showing their academic track records. Please apply before April 20, 2018.