[Published in Pipeline magazine https://pipelinepub.com/pervasive_mobility/improving-cx-with-5G/2]
Promoting customer experience (CX) is becoming a significant challenge for operators as 5G networks develop and broaden the number of services on offer. With such a wide span of services, operators are struggling to prioritise CX across all aspects as they are unable to sustain consistent customer maintenance. To manage this and diminish the risk of increased churn rate as a result, it is imperative that operators achieve a better understanding of their customers to deliver a high-quality service. This can be done by analysing customer behaviour so that their needs can be addressed and met with services in real-time. Neural Technologies have spent over 20 years developing solutions to such issues, generating positive results in the commercial application of their accurate Artificial Intelligence (AI) and Machine Learning (ML) technology.
The role of Machine Learning and the Internet of Things
Machine Learning ensures a quality 5G service by utilising unsupervised modelling in Internet of Things (IoT) technology. Currently, IoT devices are put under stress due to their inbuilt fault levels being exceeded. This causes disruption among interconnected devices as the entire system attempts to compensate and impacts performance levels. ML solves this by allowing the device to learn normal activity and recognise network elements moving outside of this for broader and faster adaptability to new environments. ML also allows for a more rapid response to work maintenance, being able to identify and solve issues within the system before subscribers can even be impacted. This is especially valuable in data rich environments where issues are difficult to manually define, and data is too vast to plausibly manually analyse.
The enhancement that ML brings to the fault tolerance of IoT is not only important around the home for the convenience of commercial users but also crucial in emergency scenarios for maintaining communication pathways. For example, in the development of the System for eDetecting and forecasting Natural Disasters based on IoT (SENDI). Here, ML enhanced and collaborated with IoT and wireless sensor networks to predict oncoming natural disasters, anticipate a failure in communication and provide a line for emergency broadcasting.
The development of these revolutionary systems in answer to such pressing real-world issues is valuable to our society on a global scale, and ML is proving highly effective in its role as a catalyst in the progression of IoT technology. The fact that such technology can have an impact far beyond the domestic scope is telling of its scale, which explains why operators are currently struggling to harness it for customer use. However, these scenarios showcase the importance of ML technologies in aiding this, demonstrating its ability to effectively improve IoT performance no matter the extent, whether it be improving CX or implementing safety measures.
Digital Twins: how does ML contribute?
Other answers to sub-optimal CX and rising churn rates include the analysis of subscribers using Digital Twin technology. Data is gathered from various sources, including subscriber devices, network logs, customer relationship management (CRM) systems and payment and billing records, to build a digital file on how they use the network. The data is then analysed to provide insight into the type of service they might prefer or be looking for. By monitoring customer usage, predictions can be made as to those likely to unsubscribe, allowing operators to put retention measures in place to deter them from doing so. As well as this, new subscribers can be scouted via cross-selling and uplifting; a company can profile their existing subscribers to identify who is benefitting the most from their services and hence identify potential subscribers who match the same behaviour to offer them the same services they would find beneficial.
Digital Twins technology uses ML and AI technology to collect data and learn independently. By doing this, a model of digital lifestyle can be created for each subscriber, showing areas of downtime and areas that need improving. This virtual representation can be accessed as a web service for operators to use to better their service performance which and improve their customer experience. Such implications to business mean that the company can take a more dedicated approach to solving customer issues and offering better 5G services, and this has impacted the progression of the Industrial Internet of Things (IIoT) greatly. The past industrial environment had significant issues with distributing widespread solutions, only having access to limited data at any one time. However, following the development of Digital Twins within the cloud and the introduction of ML, data can be automatically harvested and sifted through to provide rapid, targeted fixes.
Introductions of Digital Twins have also improved real-time customer care, with companies now having automated advice given to them on how to respond to an incoming service call. The system can also identify other matters of security, including predicting those who may move into bad debt, identifying subscription fraud and billing or payment issues, providing dynamic credit management and sending predictions of usage limit alerts. Such predictions are generated by the analytical modelling which, whilst more traditional, can provide proactive identification of potential issues when performed continuously. Once the problem has been recognised, digital integration ensures that technologies such as ML can take over and orchestrate the implementation of automatic or self-managed fix solutions.
Neural Technologies have been testing digital twin effectiveness using its Analytical Data Model in its latest development. Their findings proved that the technology is not limited to the IIoT, but also the telecommunications space. This led to their creation of the Customer Twin to address the challenges operators face in leveraging 5G services to address customer needs and improve CX. Such vast application of these technologies finds them undeniable useful in our modern, digitalised age and can also be applied to organisations, networks and processes in addition to customers. Their development of real-time communication between digital twins has also revolutionised the accuracy of the software by allowing individual twins to share recommended information and improve on one another. By making the already-tailored automated decisions even further well-informed and responsive, they have opened the potential for the future of the digital technology. Ultimately, they have succeeded in their mission to allow ‘what-if’ scenarios to be done in the virtual world without effecting the real world and, ultimately, eliminate uncertainty to offer the best possible, specified solution.
Generative Adversarial Networks: the evolution of ML
This has been advanced even further in some scenarios, with AI and ML technology being used to build generative adversarial networks (GANs). This offers the opportunity for virtual faces to replace bots in answering subscriber queries, transforming business approaches to customer service. The learning ability of GANs is incredible, being able to reconstruct human behaviours spanning a great range, from speech patterns to fine art. For example, a portrait recently sold for $432,000 was generated by GANs technology by compiling data on art history and reproducing its findings. This is possible due to ML discriminator networks continually predicting the outcomes that will most probably be received as authentic by users. Via an automatic loop of data connection from real and fake sources, the ML technology can decide on the most convincing replica of humanity. Such technology has been regarded by Yan LeCun, Facebook AI research director, as “the most interesting idea in the last 10 years in ML,” and is applicable to any range of data. Thus, by implementing it in a customer service context, subscribers will receive a personal feeling service without the need for large teams of customer service employees or more basic bots that have limited data reach and analytical ability. Being extremely advanced, such systems would excel the industry once put in place. CX would be valuably improved and companies would see better revenue as a result of the subsequent decrease in subscriber churn rate. By addressing every angle of the customer’s needs and predicting potential issues before they even require attention, the coverage of support would be deeply impactful, and customer satisfaction evident.
Summary
The integration of all such technologies has seen the development of Optimus, an Event Data Lake Analytics Platform that provides a Mediation Layer which uses ML and AI processes to direct IoT Application Program Interfaces (API) to relevant applications. Having launched Optimus, Neural Technologies can cater to a wide range of verticals, use cases and isolated IoT Ecosystems. The platform can also support several devices at once, initiating and integrating multiple different protocols. Such breadth and flexibility are crucial to ensuring open, global digital transformation via the effective implementation of IoT to advance 5G networks. Neural Technologies’ Optimus Ecosystem was recognised for its leading innovation as winners of the Outstanding Catalyst – Innovation award at TM Forum Live! 2017. The catalyst awarded was titled Open APIs for the Vibrant IoT Ecosystem (OAsIS).
Neural Technologies have worked on developing such AI and ML systems for decades. As is apparent, such technologies can cater to every issue and need that arises from the ever-transforming technological sphere. By supporting ML autonomous training and enhanced model analysis, Neural Technologies have allowed operators to use such technologies confidently and reliably. The personalised digital solutions that they provide have digitized even complex business processes, promoting Digital Integration and Advanced Automation to improve Analytics and Revenue Management. With a dedication to operators to provide the most accurate solutions to CX concerns and as one of the leading providers of Digital Transformation solutions in the industry, they continue to improve operators 5G network service offerings, improving customer experience and opening up new opportunities to generate revenue.
About the author
Dr George Bolt, Head of Analytics at Neural Technologies
Following an education covering general sciences, practical and theoretical computing, artificial intelligence and neural computing, George has developed a wide range of skills and received many awards in recognition of his abilities. During his first degree (First Class, Hons.) at Brunel University, George developed many software applications for various companies such as Esso, IBM and CERN, covering areas from hardware systems to high-level user interfaces. The University of York presented him with his doctorate following his extensive research into fault tolerance for neural computing. George joined Neural Technologies as a Neural Scientist in 1994. Since then, he has worked on a number of customised developments and has been one of the key drivers behind the development of Machine Learning within Fraud detection and Business Assurance functions. George is widely recognised as an expert in the field of fraud analysis with the application of machine learning techniques (such as neural computing).