Using machine learning to optimise critical cloud native application performance for railway operation

Location

Online seminar

Date & Time

Wednesday 25 Nov 2020 11:00 - Wednesday 25 Nov 2020 12:00

Availability

Open to all. Please contact us to be sent a link to join the seminar.

Join us for an online seminar by Jude Fletcher, DPhil Candidate at Oxford e-Research Centre. 

Due to the proliferation of Internet of Things (IoTs) and the stringent performance requirements for low latency and high bandwidth in future railway 5G applications, there is a need for a facility that allows railway native cloud applications to flexibly run on ‘unlimited’ and reliable compute, network and storage, to meet these performance requirements.

Unfortunately, the traditional cloud computing and radio access network setups do not sufficiently meet this need, compromising on the quality of service (QoS) of the railway applications. This talk proposes a machine learning-based performance-enabling facility called MECRAN, which addresses this problem by optimally scheduling applications to run on base stations at the railway edge (of a moving train) instead of in the remote cloud, to achieve the desired latency and bandwidth required.

This is an implementation of Jude's DPhil work (in collaboration with Network Rail) to demonstrate the value of his research in industry

If you would like to attend please write to events@oerc.ox.ac.uk and we will send you a link to join the seminar.