Now open for application: a PhD on mobile networks and 5G at Ericsson Hungary

Towards a comprehensive Quality of Service framework for next generation networks

Are you fascinated by mobile networks and all that potential 5G has to offer? Then read on. The EIT Digital Doctoral school is offering you an industrial doctoral programme run by Ericsson Hungary in association with ELTE University. If interested, this will open the window for you to be immersed in developing solutions to personalise network use to a large extent.

Future mobile networks will consist of many small cells that can provide large bandwidth connections to end users. This makes appropriate overprovisioning of the backhaul networks difficult and costly. Similar problems arise in multi-tenant data centre networks and Internet service providers' access networks where a large number of applications generate unknown traffic patterns. Modern applications also pose a number of new challenges against the communication network since different requirements (e.g. high bandwidth, low latency, low loss or the combination of these) should be satisfied for achieving high quality of user experience. In such scenarios, scalable and quality of service (QoS) aware bandwidth sharing and delay management are needed. Accordingly, network nodes should remain simple, not require per-user contexts and provide means for service differentiation.

In this PhD thesis we aim at developing a comprehensive QoS framework enabling controlled resource sharing in a core-stateless manner and thus resulting in higher flexibility for defining operator policies and higher scalability enabling per user policies than existing approaches. This will eventually lead to more personalized Internet services provided for all of us.

Challenge

QoS is one of the most studied area in networking with a vast array of valuable and practical knowledge. However, the issues of network QoS and resource management are still not fully solved problems. For example, QoS framework is considered a key issue in 5G standardization and one of the recent IRTF reports also lists a number of open challenges in the area of flexible resource sharing, QoS and congestion control.

An ideal QoS solution would:

  1. Allow a wide set of possible resource sharing policies. It should even enable different resource sharing policies for each user. This requires scheduling algorithms with higher scalability than the one of existing solutions.
  2. Be able to fulfil queueing delay requirements. The delay requirements of flows should be de-coupled from the throughput.
    requirement or the importance of traffic. In today systems, traffic with lower delay requirements are prioritized, providing those flows better access to shared resources.
  3. Be lightweight to implement. Low computational and memory complexities and stateless solutions are required for practical usability.
  4. Support network virtualization. The solution shall support flexible resource sharing policies across multiple layers of virtual networks.

There is no solution that can address all the above requirements.

Approach

The research topic is based on the Per Packet Value (PPV) concept that assigns values to packets according to a throughput value function (TVF). The operators encode their policies into TVFs that are then applied by the marker nodes. The design of operator policies though TVFs will require analytical skills. The new protocols, techniques and extensions researched and developed for a PPV-based comprehensive QoS framework will first be validated through packet-level simulations. Then a prototype implementation using kernel bypassing solutions like DPDK and data plane programming languages like P4 will be created to show the real-world applicability of the solution. We also plan to implement the features of the proposed framework in a cloud-native virtual router or other production-grade VNFs. A PPV-based Broadband Network Gateway has recently been demonstrated (industrial demo is submitted to SIGCOMM 2018) that can significantly improve the performance of the cloud-based BNG VNF, enabling to handle per user policies even for tens of thousands of subscribers where existing systems face with technical limitations. Many potential areas related to 5G QoS, slicing and service chaining have been identified where core-stateless solutions can be exploited for reducing OPEX and CAPEX. The deployability analysis of these directions are part of research.

Expected outcome

  • Contribution to keeping the leading role of Europe and Ericsson in 5G and NGN developments through new patents and standardization work (MEF, IETF, IRTF).
  • Novel solutions to leverage flexible and comprehensive QoS guarantees in NGNs.
  • Reducing OPEX (and CAPEX) of operators by the simple design and scalable approach.
  • Per user policies in a much more flexible way than nowadays.
  • More personalized services for end users: QoS-level customization and paying only for the service level that the user really requires.
  • Implementation integrated into Ericsson's cloud-native NF products.
  • Roadmap for real-world deployment.
  • High-quality publications.

Location

As an industrial doctoral student, you will reside in the EIT Digital Doctoral Training Centre in Budapest and share your time with the premises of Ericsson Hungary and the ELTE Eötvös Loránd University. A six-month mobility to another European university or research institution will be also part of the programme.

Facts

  • Industrial partner: Ericsson Hungary Ltd.
  • Academic/research partner: ELTE Eötvös Loránd University, Budapest, Hungary
  • Number of available PhD positions: 1
  • Duration: 4 years
  • This PhD will be funded by EIT Digital, ELTE Eötvös Loránd University, and Ericsson Hungary Ltd.

Apply

If you are interested in applying for this position, please send an e-mail to the EIT Digital Budapest DTC lead zoltan.istenes@eitdigital.eu, including a CV, a motivation letter, and documents showing your academic track records.

Please apply before 24 August 2018, 12.00 CET.

© 2010-2018 EIT Digital IVZW. All rights reserved. Legal notice