The digital communications infrastructure of the UK forms a significant component of the Critical National Infrastructure, upon which the commercial and social wellbeing of the country depends. BT, as the major telecoms provider, recognises the importance not only of running a highly resilient network, but also one which keeps pace with the increasingly dynamic demands of customers and the wider digital ecosystem. This requires a fundamental rethink of the supporting infrastructure and operations. The scope extends beyond the UK to the provision of service across the globe through BT’s international business, and also in the sense that industry-wide agreements are needed to achieve the global scale required for successful economics and service interoperability.
The NG-CDI project is at the forefront of new research areas which are creating transformational solutions which address this future, and is engaged with the industry as a whole through standards bodies and thought-leading publication in the field.
The provision of new network service capabilities has historically depended upon the rollout of new equipment. To change the pace of service innovation radically requires an infrastructure able to respond to new demands as they emerge, rather than having to rely on uncertain forecasts tied into longer-term investment decisions and rollout planning.
NG-CDI has addressed these limitations by developing capabilities and processes to deliver software-enabled service capability into the network. Intents Based Networking capabilities have been demonstrated which allow services to be defined in terms of business requirements rather than the requirements of the technology. The customer relationship becomes more open and creative in the development of new services. Network resources in both the fixed and mobile (5G) networks can automatically align to optimise the desired service. This can be done iteratively with customers (SPs and large corporates) as art of the service setup, or, in principle, even during an individual service session.
An ambitious programme geared to create a radically new architecture for the UK’s internet and telecommunications infrastructure
Telecoms infrastructures comprise many layers of technology and capability which build together to provide the end service. Service delivery is therefore complex, involving many separate technical and business processes across this range of domains. Legacy technologies have necessitated a whole range of separated processes, which increases organisational complexity and limits efficiency. Different processes are focused on the various architectural layers (such as equipment, network, service), and on different timescales of operation (such as design, strategic planning, maintenance scheduling and on-the-day-operations). Decisions taken in these separate domains requires considerable co-ordination activity to ensure optimal outcomes.
NG-CDI is demonstrating that the complexity of the interrelated layers of capability and timescale can be managed more easily as a whole, optimising operations from a balanced set of perspectives. New network modelling techniques based on live network data enable different perspectives to be taken, enabling fundamental improvements in end to end decision making. Different possible service/cost impacts of different interventions and investment options can be explored. The resilience of different design options can be modelled.
The balance between proactive and reactive maintenance can be assessed. Sophisticated statistical methods have been demonstrated in live operations which can identify changes in patterns of massive live data streams to alert operations for any necessary response.
Transformational change involves not only the technology but equally the human and organisational components. Taking full advantage of the technical capabilities offered by the project requires changes to organisation and culture. Advancing the technical capability without wider change merely freezes in any historical limitations. The project has therefore embraced these organisational and cultural aspects to understand and prepare for the future.
The technical/modelling capabilities described increase the power of the technology, enabling human control and reducing the effort devoted to low level repetitive processes. This releases creativity to improve and develop new services. Commercial risk is reduced compared with historic fixed automation, which locks in investments and creates barriers to system change. However, more powerful technology can also carry risks. A range of protection mechanisms has been identified, embracing technical, organisational and cultural elements. Integrating these into culture and tools promises to reduce risk compared with historical norms based as they are on fixed automated processes and risk management processes which are disconnected from daily decision making.
The output of the project falls into four key business/technical headings which reflect these different areas of benefit. Each is an active collaboration between the university academics and BT’s researchers, developers and business areas. These vary in nature from the trialling of real algorithms in live business operations, through to thought leadership and strategy, working with and influencing decision makers in the business and the wider industry though standards organisations.
This work is transforming the customer relationship and growing the ability to experiment in the marketplace. Instead of the long linear development process from requirements capture to mass rollout, the relationship becomes continuous and iterative. Expressing requirements in terms closer to business language, and receiving feedback from the infrastructure on the nearest available matches to requirements, helps iterate to more successful services.
Software-based deployments of the service enable scaling up and down to match demand, making more efficient use of resources. The customer relationship extends over longer timescales through DevOps tool chains, which enable continuous adjustments to existing services. Service-specific performance monitoring can be rolled out in parallel with the service, giving customers assurance on service levels.
This area has collaborated with the TMF Autonomous Networks Group to test and extend their existing and proposed Intents API. This links the work firmly with existing developments and deployments as well as contributing to the direction of the technology across multiple vendors and operators.
Intents & DevOps...
The work in 5G mobile networks has extended the research into Intents Based Networking as well as other areas such as AI-optimised design and “digital twin” modelling. Intents Based Network slicing has been demonstrated, optimised against a set of desired characteristics.
Network service resources can be tuned to the requirements of a particular service, such as low delay or high throughput. This improves the user experience for a wider range of services, including, in principle, ones which have yet to be invented. Responding to a user or application request during an individual call session has also been demonstrated, such as a request to add a video stream or change resolution.
AI techniques have been developed which optimise the distribution of functionality between components in the design of a 5G network. What-if models have been used to explore the response of mobile networks to different possible traffic scenarios (such as the mix of people and future densities of self-drive cars).
Optimising the digital infrastructure over a range of business and service drivers across a wide range of timescales requires powerful new techniques. The ultimate aim is the best customer service at minimum cost, but this can be achieved in different ways. Greater spare capacity can allow the headroom to reduce operating costs for example. When and and how service is impacted are therefore dependent on these decisions. Significant advances in network modelling have demonstrated the ability to visualise and investigate the digital infrastructure from these different perspectives. For example, the priorities on maintenance and scheduling, informed by data from the network, can be determined by the end service quality required and overall risk effect on the network.
This is a more global optimum, rather than resulting from a set of local technical requirements. The models enable the investigation of different design scenarios which can increase overall resilience, reducing the likelihood of cascade effects which could result from particular events.
Service Assurance and Infrastructure Management...
Trends in traffic patterns can inform the business regarding service trends and possible hotspots. Sophisticated techniques have been applied to the detection of anomalies in very large real-time data flows. These have been deployed in BT’s operational systems to inform operations of possible problems or trends. Determining the difference between a normal – but still stochastic – flow of data and one which represents a change that operational attention is a non-trivial problem. The robust algorithms deployed incorporate the ability to set different sensitivities in order to optimise operational resource.
Risk & Governance...
A well thought out Governance picture is essential not only to control the risks associated with these more powerful technologies, but also to reduce the risk from the levels already inherent in a heavily automated business. The approaches taken include three levels: Organisational, Cultural and Behavioural.
To address the Organisational level, a number of case studies was undertaken with businesses that have already gone down the path of AI enablement. These include companies from the Finance, Construction, Logistics and Power sectors. This gave insights into options for organisational structures and processes for risk control. The studies also inform transitional concerns such as the communication of change and the building of employee trust in the new capabilities. At the Cultural level, management approaches to enabling and encouraging creativity in the workforce have been workshopped with the business. The language used around “risk” has been found to be freighted with cultural history, and so communication and empowerment is key.
Working with BT’s Group Risk area, and BT’s new Risk Framework has encouraged project attention on Behavioural Risk. The aim here is to change risk management away from a separate offline “tick list” process, to one in which risk is simply part of the everyday professional judgement and quality decision making across all roles. These judgements and decisions are greatly aided by the new modelling techniques developed by the project and so there is a very close relationship between the models in Service Assurance and Infrastructure Management and the work in Risk and Governance.
To this end, a number of interviews with a range of senior roles across the different timescales of design and operations has been held to gain deep insight into the richness of decision making in those roles. This helps shape the models to best effect to those roles, whilst maintaining the ability to look across these multiple perspectives so as to optimise the infrastructure holistically.