Next Generation Converged Digital infrastructure (NG-CDI)

Lead Research Organisation: Lancaster University
Department Name: Computing & Communications

Abstract

This programme will forge the research required to underpin the next generation converged digital infrastructure for BT, creating a radically new technology architecture for autonomous operation of future networks and services.

Digital infrastructure networks of the future will be highly reliable and resilient to disruptions through autonomous operations and will be able to cope with increasing demands on its capacity and types of services. These networks will be equipped with programmable and virtualised network functions that can be flexibly placed at specific network locations. New and unpredicted services will be capable of being supported without the need to make costly changes to the infrastructure at the physical level. The physical nodes of the infrastructure network will be represented digitally by a 'digital twin' or 'software agent' which will make them "autonomic" - i.e., the capability to perceive its state and environment, understand and predict its behaviour, and react to disruptions and opportunities autonomously with an aim to enhance customer experience. The network will be able to detect and predict possible disruptions, analyse the risk to service provision, make autonomous decisions regarding the (re)deployment of functions to least risky network locations and arrange remedial actions such as repair or replacement of risk-prone nodes. This will lead to new services, improved resilience of the network, better customer experiences and greater operational efficiency ensuring that the UK remains a leading digital economy.

In order to realise this vision, the research carried out in this programme is structured around 5 challenges:
Research Challenge 1: Agile Converged Infrastructure Systems Architecture. The fundamental question addressed here is "How to build an agile digital infrastructure that is amenable to autonomous operations?" This will be achieved through developments in new technologies such as MicroNFV and SDN.

Research Challenge 2: Future Networks Operations and Services. The fundamental question addressed here is "How to ensure service reliability of the agile autonomic digital infrastructure?" This will be achieved by developing an automated service ecosystem capable of placing virtualised network functions at specific network locations.

Research Challenge 3: Autonomic Knowledge Framework. This challenge addresses the question "How to enable autonomous operational ability for the digital infrastructure?" This is tackled by a multi-agent system architecture and through creating data sources that are intelligent.

Research Challenge 4: Autonomous Diagnostics and Response. This challenge addresses the question "How can the digital infrastructure respond to disruptions autonomously?" This will be answered by developing novel automated change detection and statistical learning techniques.

Research Challenge 5: Future Organisational Dynamics. The question addressed here is "How can the organisation exploit the autonomic agile capabilities of the digital infrastructure?" This is addressed by developing decision-support algorithms for risk-based function redeployment and predictive asset management.

Planned Impact

Society: Today's society is entirely dependent on technology and the availability of "connected" services. The UK currently has one of the most advanced digital infrastructures in the world with over 95% of its citizens and businesses having access to super-fast broadband and digital infrastructure overall was used by over 41million adults daily in 2016. The ongoing economic and societal importance that the UK Government places on the nation's digital infrastructure was highlighted in the 2016 Autumn Statement as well as the National Infrastructure Commission consultation. The technologies and solutions delivered by this project will bring direct benefits to society by improving the resilience of the digital telecommunications network, enhancing service assurance and thereby customer experience, and producing the next generation converged digital infrastructure that is sufficiently agile to respond to unforeseen new ICT services.

Business Impact: Future ICT services will change ever more rapidly - and unpredictably - and a fundamental change to the economic model for infrastructure development is required. The project will produce an infrastructure which drastically reduces the time taken for new services to be launched. By enabling this agility, our project will reduce BT's operational and development costs, producing huge capital and operational cost savings, and growing value significantly.

Economic Impact: By bringing the world of cyber-physical systems to the digital infrastructure domain, the project delivers significant scientific advances in an area that is critical to the success of the UK economy. The output from this programme will have a direct impact on the rate of adoption of technologies such as NFV and 5G by telecoms operators. By providing a resilient infrastructure with the flexibility for operators to configure and offer new services, and greater agility for businesses who need digital services that adapt as they grow, these outputs will directly address one of the new government industrial challenges - Transformative Digital Technologies.

Publications

10 25 50

publication icon
Fawcett L (2018) Tennison: A Distributed SDN Framework for Scalable Network Security in IEEE Journal on Selected Areas in Communications

publication icon
Li H (2018) A social network of collaborating industrial assets in Proceedings of the Institution of Mechanical Engineers, Part O: Journal of Risk and Reliability

publication icon
Petchrompo S (2019) A review of asset management literature on multi-asset systems in Reliability Engineering & System Safety

publication icon
Salvador Palau A (2019) Collaborative prognostics in Social Asset Networks in Future Generation Computer Systems

 
Description A summary of the key findings obtained thus far within the project:

- Intent-based networking has emerged as a key concept to support the architectural concepts within the project. This has led to the development of a top-level framework, use cases and basic testbed facilities.

- A technology demonstrator showcasing the emerging concepts within the project has been developed, combining expertise across Universities and involving the research disciplines of networking, statistics and systems engineering.

- The demonstrator has been used to explore intent-based networking, utilising both network and environment data to make intelligent decisions as to how traffic is forwarded within a network.

- This demonstrator will provide a platform to investigate new approaches to developing and deploying code in increasingly softwarised networks. This is driven by closer collaboration between development and operational processes and enabled through the emerging technologies.

- Our research into intent-based networking has been disseminated to the ITU-T Network-2030 Focus Group as a key enabling technology for future network management.

- The project is pioneering strategies based on Reinforcement Learning (RL), a sub-domain of Machine Learning (ML), to orchestrate network operations. Reinforcement Learning strives to produce general purpose algorithms that can learn how to behave directly from interaction, culminating in more intelligent decision making.

- The project is designing infrastructures capable of supporting Connected and Automated Mobility (CAM). The safe and secure operation of such networks is contingent on the early detection of anomalies. The project has developed a range of novel unsupervised learning models to detect anomalies in CAM systems.

- BT have carried out successful trials of online anomaly detection methods developed within the project to help manage network performance. Given the scale of data complexity within the network, in terms of equipment and the numbers of metrics, it is impossible for these to be monitored by human eye. Hence methods such as these are crucial.
Exploitation Route The programme's primary impact pathway will be via the integral partnership between BT and the consortium universities. From co-creation of the proposal to shared oversight of the research, the programme has been designed to facilitate bidirectional transfer of understanding and skills, enabling efficient and trusted pathways to impact. Impact will be achieved via the direct engagement with BT's researchers, engagement with BT's network & system architects and operations centres, BT's strategic partners (including Huawei, Intel, Cisco) and the ecosystem of associated small businesses at Adastral Park.
Sectors Digital/Communication/Information Technologies (including Software)

URL http://www.ng-cdi.org
 
Description This is an important strategy-setting project for the industry, and is there is a range of timescales of benefit. Each has a different pathway to downstreaming: On the shorter timescale to direct impact, we have a number of detailed collaborations between the academic and BT researchers. These have resulted for example in anomaly detection algorithms generated by the project being used in the business already to help diagnose and thus resolve customer and network issues. Network faults are statistically rare, which would limit the efficacy of machine learning, so the project has generated sophisticated techniques to augment apparent fault data to speed up AI learning speeds. This solves a fundamental problem with many machine-learning autonomic applications within and beyond telecoms. These techniques scale up autonomic performance to industrial scale and near real-time operation. The network modelling capability from the project has been used to assess the benefits of different content caching schemes for content delivery on the network. On the medium timescale, we are working with a number of strategic programme Directors in BT to focus benefits in those areas. These include the assessment of risks to the network, eg from flood vulnerability, or from connectivity to other networks, such as the interdependence with power networks; optimising the load on mobile networks of users' session tracking; and building network capacity models based on real time-dependent user demand. We are in discussion with the Chief Data Architect and Network Automation Programme. BT Directors have attended, presented and interacted at NG-CDI Plenary meetings. Valuable insights in a variety of other technical areas have benefitted BT; such as deep reinforcement learning and unikernels. A proof-of-concept prototype of agent-based network control has been used to illustrate the potential for managing the balance between different policies in the network, for example between the operator's and cutomers' requirements. Key results are planned also to be disseminated more broadly through our annual Data Science Week: a practical training session run by BT Research, each year involving 100 participants from across the company. In the longer term it important that we influence the direction of the industry. This is not limited to BT, but will include the wider research community, standards bodies, vendors and other operators. Telecoms is a very interconnected business of its nature, and so commonality of approach is a make or break issue. At this stage the project has concentrated on reaching out to the research community, and to standards bodies. These include exchanging insights with the NG-CDI and the ITU Network 2030 Focus Group. We have initiated discussions and provided thought leadership with senior BT IT Architects. The Industry Impact Board held in February BT included the BT Dynamic Infrastructure Director and a member of the Tele Management Forum. This proved to be a valuable session in providing further alignment between the views of the different groups.
First Year Of Impact 2019
Sector Digital/Communication/Information Technologies (including Software)
Impact Types Economic

 
Description 5G Testbeds and Trials Programme
Amount £2,100,000 (GBP)
Organisation Department for Digital, Culture, Media & Sport 
Sector Public
Country United Kingdom
Start 04/2018 
End 03/2019
 
Description NEWTRIP: New Transport-layer Intelligence and Protocols
Amount £12,000 (GBP)
Funding ID IEC\NSFC\170090 
Organisation The Royal Society 
Sector Charity/Non Profit
Country United Kingdom
Start 04/2018 
End 03/2020
 
Description Network 2030
Amount £1,050,000 (GBP)
Organisation Huawei Technologies 
Sector Private
Country China
Start 01/2019 
End 12/2021
 
Description • Digital Twin Demonstrator
Amount £250,000 (GBP)
Organisation Digital Built Britain 
Sector Private
Country United Kingdom
Start  
 
Description • Digitally optimised through-life engineering services
Amount £630,000 (GBP)
Organisation Aerospace Technology Institute 
Sector Charity/Non Profit
Country United Kingdom
Start  
 
Description • GIS-Based Infrastructure Management System for Optimised Response to Extreme Events on Terrestrial Transport Networks (2018-22
Amount € 295,800 (EUR)
Funding ID 769255 
Organisation European Commission 
Department Horizon 2020
Sector Public
Country European Union (EU)
Start  
 
Description • Social Networks of Gas Turbines for Collaborative Fault Prognosis
Amount £130,500 (GBP)
Organisation Siemens AG 
Department Siemens Power and Gas
Sector Private
Country Global
Start  
 
Description • Socially aware infrastructure assets
Amount £170,000 (GBP)
Funding ID 104257 
Organisation Innovate UK 
Sector Public
Country United Kingdom
Start  
 
Description BT 
Organisation BT Group
Country United Kingdom 
Sector Private 
PI Contribution Shared research
Collaborator Contribution Funding and joint research
Impact TBC
Start Year 2017
 
Description Collaboration with Scania 
Organisation Scania
Country Sweden 
Sector Private 
PI Contribution This research showed the potential for improving equipment failure prognosis, and was demonstrated using a public dataset published by Scania. This has led to a collaboration agreement with Scania to test the methodology on more truck components.
Collaborator Contribution Scania has shared datasets and access to engineering personnel.
Impact Ongoing partnership. Further results will be published later this year.
Start Year 2019
 
Description University of Bristol 
Organisation University of Bristol
Country United Kingdom 
Sector Academic/University 
PI Contribution TBC
Collaborator Contribution TBC
Impact TBC
Start Year 2017
 
Description University of Cambridge 
Organisation University of Cambridge
Country United Kingdom 
Sector Academic/University 
PI Contribution TBC
Collaborator Contribution TBC
Impact TBC
Start Year 2017
 
Description University of Lancaster 
Organisation Lancaster University
Department School of Computing and Communications
Country United Kingdom 
Sector Academic/University 
PI Contribution TBC
Collaborator Contribution TBC
Impact TBC
Start Year 2017
 
Description University of Surrey 
Organisation University of Surrey
Country United Kingdom 
Sector Academic/University 
PI Contribution TBC
Collaborator Contribution TBC
Impact tbc
Start Year 2017
 
Title anomaly R package 
Description Provides a number of functions to enable the ease of use of a variety of different time series anomaly detection methods for a wide range of users. 
Type Of Technology Software 
Year Produced 2018 
Open Source License? Yes  
Impact Researchers from other institutions have used the software, including at Nokia Bell Labs. They published some research with DOI:10.1109/ISSREW.2019.00041 
URL https://cran.r-project.org/web/packages/anomaly/index.html
 
Company Name MAVERICK ML LTD 
Description The company is a spin off from the Asset Management Group, and the focus of the company is to provide machine-learning based solutions for detection and prediction of industrial equipment failures. The unique value proposition for the company is based on the research carried out by Gishan Don Ranasinghe (CEO of the company), a PhD student under the supervision of Dr Ajith Parlikad. 
Year Established 2020 
Impact The company was established in January 2020. It is too early to describe impacts from the company.
 
Description Dissemination through International Telecommunication Standardization Focus Group of Network 2030 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Industry/Business
Results and Impact On October 2nd 2018, Prof. Ning Wang was invited to give a keynote talk at the first technical workshop of ITU-T Focus Group on Network 2030. The presentation focus is on network intelligence in future networks and the technical work carried out in NG-CDI was highlighted in the talk. The workshop was held at University of New York and there were around 70 people from both ICT industry and academia.
Year(s) Of Engagement Activity 2018
URL https://www.itu.int/en/ITU-T/Workshops-and-Seminars/201810/Pages/default.aspx
 
Description Fourth Workshop on Network 2030 and the fourth meeting of force group Network 2030 
Form Of Engagement Activity A formal working group, expert panel or dialogue
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Industry/Business
Results and Impact Dr. Mehdi Bezahaf has attended both the Fourth Workshop on Network 2030 and the fourth meeting of force group Network 2030 organized by the International Telecommunication Union (ITU). He has met some interesting people, such as Prof. Zhani who invited him to visit ETS Montreal or Dr. Shen Yan from Huawei who invited him to participate in a workshop on the future of the Internet. These invitations allow Dr. Mehdi Bezahaf to get involve with standardization bodies such as IETF and have a better understanding of intent-based networking that becomes one of an important aspect of the project.
Year(s) Of Engagement Activity 2019
URL https://www.itu.int/en/ITU-T/Workshops-and-Seminars/201905/Pages/default.aspx
 
Description Network Management (NMRG) research group of IETF 
Form Of Engagement Activity A formal working group, expert panel or dialogue
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Industry/Business
Results and Impact Through different discussions during his different meetings and workshops, Dr. Mehdi Bezahaf has started picking interest in the Network Management (NMRG) research group of IETF. He has taken advantage of being in Montreal to attend the NMRG sessions at the IETF 105. After exchanging with the chair of the research group (Laurent Ciavaglia from Nokia), He gets invited to the NMRG workshop. Network automation is one of the most important aspects of our project, and intent-based networking is one way to achieve it.
Year(s) Of Engagement Activity 2019
 
Description Network Management (NMRG) research group IBN workshop 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Industry/Business
Results and Impact Dr. Mehdi Bezahaf has participated in the Network Management (NMRG) research group IBN workshop. After this meeting, Dr. Mehdi Bezahaf had a better view of intent and how it can impact our project.
Year(s) Of Engagement Activity 2019
 
Description The 10th International Conference on Network of the Future (NoF 2019) 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Schools
Results and Impact Dr. Mehdi Bezahaf has presented his paper on Self-Generated Intent-Based Systems at the 10th International Conference on Network of the Future (NoF 2019). During the conference, He has caught up with the chair of NMRG. Dr. Mehdi Bezahaf has proposed to work on a survey on the intents. As a follow up to this discussion, they have gathered people from UCL London, Nokia, Orange, Telefonica, University of Bologna, and UFRGS Brasil to work on collaborative papers on Intent. Dr. Mehdi Bezahaf is currently leading the team working on common definitions of intents.
Year(s) Of Engagement Activity 2019
URL https://web.uniroma1.it/nof2019/home
 
Description The fifth Workshop on Network 2030 and the fifth meeting of force group Network 2030 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Industry/Business
Results and Impact Dr. Mehdi Bezahaf has attended both the fifth Workshop on Network 2030 and the fifth meeting of force group Network 2030 organized by the International Telecommunication Union (ITU). He has been elected to be a champion to harmonize different contributions and check if all the used terms are consistent. He has also presented his views on intents. He has been responsible for contributing to the Net2030 architecture (intent and automation section). This contribution gets used for the project deliverables.
Year(s) Of Engagement Activity 2019
URL https://www.itu.int/en/ITU-T/Workshops-and-Seminars/2019101416/Pages/default.aspx
 
Description The sixth Workshop on Network 2030 and the sixth meeting of force group Network 2030 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Industry/Business
Results and Impact Dr. Mehdi Bezahaf has attended both the sixth Workshop on Network 2030 and the sixth meeting of force group Network 2030 organized by the International Telecommunication Union (ITU). Ms. Eleanor Davies and Dr. Mehdi Bezahaf presented our project's demo at the event, which gives more impact on our project.
Year(s) Of Engagement Activity 2020
URL https://www.itu.int/en/ITU-T/Workshops-and-Seminars/2019101416/Pages/default.aspx
 
Description Third Workshop on Network 2030 and the third meeting of force group Network 2030 ITU 
Form Of Engagement Activity A formal working group, expert panel or dialogue
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Industry/Business
Results and Impact Dr. Mehdi Bezahaf has attended both the Third Workshop on Network 2030 and the third meeting of force group Network 2030 organized by the International Telecommunication Union (ITU). Both events were really interesting and it was the first contact for him with standardization body people. He gets involved with sub-group 3 on the topic of network architecture and infrastructure. Dr. Mehdi Bezahaf has also the chance to discuss with Prof. Mostafa Ammar (Georgia Tech - US) about network architectures and Internet evolution. This meeting and his engagement with sub-group 3 helped him, later on, to write a paper on Internet evolution and submit it to the IEEE Internet Computing magazine.
Year(s) Of Engagement Activity 2019
URL https://www.itu.int/en/ITU-T/Workshops-and-Seminars/20190218/Pages/default.aspx
 
Description Visit of ETS Montreal 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Industry/Business
Results and Impact During his last Network 2030 workshop at Saint Petersburg, Dr. Mehdi Bezahaf gets invited by Prof. Zhani to visit ETS Montreal. He has spent almost a week visiting and discussing network management and intent topics with Prof. Zhani. He has introduced the NG-CDI project and did a presentation on the evolution of the Internet.
Year(s) Of Engagement Activity 2019
 
Description Visit to BT by Lancaster colleagues to meet researchers at BT in anomaly detection. 
Form Of Engagement Activity Participation in an open day or visit at my research institution
Part Of Official Scheme? No
Geographic Reach Local
Primary Audience Industry/Business
Results and Impact The purpose of the visit was to try and embed the anomaly detection methods for real time data developed, into BT's systems.

BT researchers had developed a system to obtain a feed to get real time data on measurements from a variety of instruments, such as edge routers. These measurements were typically quite high frequency, i.e. every 30 seconds. The challenge they encountered was in identifying anomalous behavior and sounding an alarm when enough evidence had been observed but quickly enough to still be useful in practice.

The first challenge was to model the behavior of the data in its "normal" state. A group of researchers from Lancaster visited BT for a week to discuss these problems. We were given a sample of training data, reminiscent of this "normal" state and estimated a model, based on which we could define a measure of how anomalous a datum was.

The method we developed has been used by BT researches and has been tested in internal systems to more quickly identify potential faults in real time.

After the event BT researchers expressed interest in other areas where anomaly detection.
Year(s) Of Engagement Activity 2019
 
Description Workshop on the future of the Internet 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Industry/Business
Results and Impact During the fourth workshop on Network 2030, Dr. Mehdi Bezahaf has met some people from Huawei that invited him to a workshop on the future of the Internet organized by Huawei. He has participated and debated on self-driving networks and future network architecture. During this event, Dr. Mehdi Bezahaf met with Cheng Zhou (from China Mobile), with whom he writes a section on intent-base networking as a contribution for ITU Network 2030.
Year(s) Of Engagement Activity 2019