Master Thesis | Fall 2019 | Predicting Synchronization Reference Behavior Using AI Methods

Job Description

Date: May 8, 2019

PDU Base Band and Interconnect


We at Business Area Networks are responsible for the company’s industry leading radio products and solutions and is the foundation for Ericsson’s technology leadership. Our Product Development Unit (PDU) Base Band and Interconnect consist of approximately 2000 R&D professionals spanning R&D sites in Europe and Asia. We are responsible for all aspects of Base Band and Interconnect HW and SW module development and their integration. The products we develop are part of Ericsson Radio Systems and our PDU is heavily involved in product development for 5G.


We are now looking for a student to conduct master thesis work on Modeling Synchronization references, in collaboration with our Synchronization teams. The period will start from Fall 2019 (August/September).

We welcome the opportunity to meet you!


Thesis Subject


Mobile broadband for cellular networks are continuously being evolved to meet the future demands for higher data rates, improved coverage and capacity. The enormous success of Smartphones boosts mobile broadband data requirements. 4G or Long Term Evolution (LTE) is commercial since several year back, and is continuous being evolved. Mobile broadband applications require sub micros second time accuracy.


To fulfill necessary timing requirements in the Radio Access Network, the Base Station implements a synchronization solution where measurements provided by an external synchronization reference are used to control the radio equipment clock to maintain stable frequency and phase.

Available synchronization references are validated according to specific criteria such as quality level and priority, bad references are deselected. However, it is very difficult to detect slow reference degradation. Also, it is rather common that network operators are not able to provide synchronization references of sufficient quality.


This thesis will focus on building a learning algorithm for different types of synchronization references that will predict the external reference behavior by analyzing measurements provided by several sources, taking into account their own nature and consequent possible inaccuracies.

A possible solution could include an Artificial Neural Network (ANN), by exploiting their powerful processing capabilities and therefore developing a model performing computational tasks faster than the
traditional systems.



You will drive this master thesis work with support of an experiences mentor and collaboration with experienced developers in Synchronization area.

We are looking for people with genuine passion for software development who really wants to make a difference and won´t hesitate to strive towards quality results.




  • Develop a method or algorithm as described in the master thesis subject
  • Simulate, Prototype and Verify the proposed method\algorithm with baseband SW and radio HW.


Key Qualifications:


  • Student of M.Sc. major in Applied Physics, Mathematics , Machine Learning or Artificial intelligence.
  • Student of M.Sc in Computer science, Electrical or Telecommunication Engineering with a passion for the above mentioned.
  • Knowledge in control theory and analytical modeling is required.
  • Programming skills are essential, Matlab and/or C and/or C++ are our primary tool for modeling and simulation work but the applicant is free to choose suitable tools.
  • Good communication skills and fluent in English is required
    This position will be based in Ericsson HQ in Stockholm, Kista for the duration of the thesis period.