Projects

Ongoing funded projects

ORLANDO

O-RAN intelligent adaptive Load balancing and efficiency in highly Dense deployments

The ORLANDO project aims to develop intelligent, spectrum-efficient load balancing solutions for dense Open Radio Access Network (O-RAN) deployments. Benefiting from University of York’s partnerships made within previous successful DSIT-funded projects (YO-RAN and REACH), ORLANDO has access to a large data set of real users’ movement and traffic. The data set will be used to train a traffic prediction Machine Learning (ML) model and also fine tune a generative AI to generate synthetic data for new environments and unforeseen events. Two load balancing ML-based approaches will be investigated.

A centralised deep reinforcement learning-based xApp for small area load-balancing, and a scalable federated reinforcement learning-based architecture for larger RAN environments. The FRL architecture is expected to reduce the training energy consumption by 35%, in addition to its privacy preserving benefits. ORLANDO’s solutions will be tested on an Digital Twin like emulation environment (VIAVI’s AI-RSG), and trialed on York testbed and Blackpool winter gardens increasing its TRL from 2 to 5.

YO-RAN

YO-RAN (Yorkshire Open RAN) brings together two leading Yorkshire universities with a number of suppliers and network operators, several based in and around Yorkshire, to develop Open RAN components and a RIC for Neutral Host Networks – another development which allows the same infrastructure to be used by multiple operators, and also by private or enterprise-based networks. This requires in particular low-cost generic Radio Units with configurable multiband capabilities which are not currently available, and efficient fronthaul interfaces to connect them with suitable Distributed Units – all configurable via a RAN Intelligent Controller which can respond to the dynamic needs of all the network users. Open-RAN is a new approach for Radio Access Networks (RANs) – which consist of all the infrastructure including base stations and associated connections to allow a mobile phone to connect to the telecommunications network. It creates an open network allowing different suppliers to provide different components, including the Radio Units (RUs) at the base stations and the Distributed Units (DUs) which process radio signals, creating a more open and competitive supplier “ecosystem”. The RAN network is also virtualised so that different functions can be performed in different locations under the control of software called the RAN Intelligent Controller (RIC).

ORAN-TWIN

ORAN-TWIN is funded by EPSRC CHEDDAR Hub TMF uplift, Federated Telecoms Hub 6G Research Partnership Funds (THRPF). The task for ORAN-TWIN project is to integrate advanced machine learning techniques into ORAN xApps for dynamic resource allocation and test time data augmentation.

We plan to simulate diverse network conditions to generate synthetic data, optimize resource allocation and ensure real-time performance adaptation. Leveraging digital twin, the xApps can be interacting with virtual environments to avoid hardware cost and search overhead in the physical systems. Digital twin helps to detect and diagnose potential issues, thereby improving physical systems reliability and efficiency.

Send air

ScalablE aNd trustful crowDsourcing of AIR pollution sensings (SEND-AIR)

Air pollution is a challenge that each metropolitan is facing in a way. With diverse, frequent and high resolution air quality/pollution sensing data, municipality policy makers, health authorities and businesses can plan their policies and activities more efficiently. While installing/building a large number of air quality sensing stations is costly and high maintenance, crowdsourcing is a low cost approach to gather massive amounts of data. Crowdsourcing requires incentive mechanisms to be widely adopted by users. The incentives can be in the form of small payments and/or a game which needs to be protected with fraud/cheat detection and prevention mechanisms.

This is a Royal Society international exchange project with ITU.

Selected completed projects

REACH

Tourist hotspots drive significant volumes of mobile traffic, presenting a challenge to existing mobile technologies. The REACH project brings together an exciting group of partners to find new solutions to address this challenge by improving the delivery of data in High Density Demand (HDD) areas.

By developing new tools for an intelligent controller, the partners will enhance different aspects of data delivery. The project will deliver 3 distinct O-RAN (Open Radio Access Networks) HDD trials that will demonstrate the technology using realistic data.

MANY

Mobile Access North Yorkshire

DCMS funded Interdisciplinary 5G Testbed and Trials project providing 5G connectivity to uncovered areas of North Yorkshire.

We designed and developed rapidly deployable terrestrial and aerial 5G networks for providing coverage for temporary events and disaster relief tasks.
Supported two main use cases mountain rescue and outdoor large social events.
SDR and open-source SRS software allow the mobile network to adapt to future 3GPP releases

Macbook

MAChine learning-Based Online sOcial media marKeting (MACBOOK)

Everyday on our social media we see advertisements of the local or global business. Some of them are relevant to us and many are not in our interests. While these advertisements are a main revenue source of social media companies like facebook, they are costly to local businesses and marketing companies. The social media companies can display the ads on pages of any requested groups of their users, but the local businesses often do not know which groups to target and where to find them. They rely on the experience of marketing agents who also struggle to perfectly reach the target audience. This wastes their advertising budget. In this project partnering with Pick & Mix marketing we design, train and develop a Machine learning algorithm that learns from social media data and identifies target audience for their advertisement and marketing campaigns. Our ML only uses public data and is fully compliant with UK and university of York ethical code.

KTP with AEMT

Virtual twins for hands on education

This KTP aims to convert a hands-on practical course into an online virtual learning system-based course which uses virtual twins of real world equipment for training and assessment. Assessments will be undertaken through the development of dynamic grading, based on various different types of virtually twinned machinery.

Unleashing the power of influencer marketing to stimulate prococial causes

White Rose Consortium led by University of Sheffield.

White Rose funds a project entitled ‘Unleashing The Power of Influencer Marketing to Stimulate Prosocial Causes’ to fill above research gaps by extending the current knowledge on conditions where influencer marketing can be most effective in promoting prosocial behaviour to address the climate change emergency. Using an interdisciplinary approach, this project, which is funded by White Rose, brings together a team of Social Science, Engineering, and Environmental Management researchers from leading universities of Sheffield, Leeds, and York to decode complexity of influencer.

https://sheffield.ac.uk/bristt/research/research-projects/influencer-marketing-and-climate-change

aware

AI-based Wireless coverAge pRediction and Enhancement (AWARE)

Wireless coverage analysis is an important task for network planning and deployment. Having an accurate coverage map significantly reduces unnecessary field trips of telecommunications engineers, and operators’ capital and operational expenditure. Existing coverage analysis software, like WISDM by Wireless Coverage Ltd, work on a point-to-point Line-of-sight basis which degrade their efficiency. In this project we will use machine learning algorithms to efficiently find the point(s) in 3D domain with the best link conditions. To further enhance the accuracy, we aim to integrate real sensing data with map-based information. We will demonstrate our developed software patch on WISDM.

Open access paper: https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9523543

Page updated

Google Sites

Report abuse