In this user-centric era, the update periods of software applications determine the pace at which user experience is improved. A/B testing that started in the Internet industry can deliver application updates to users in small, quick steps, allowing users to enjoy new features in no time.

In the last five to six years, the telecom industry has been transformed through network functions virtualization (NFV), which is a big leap to ICT convergence. To apply NFV technology, carriers spared no effort to decouple software from hardware and replaced proprietary infrastructure with cost-effective, general-purpose COTS products. They dismantled the silos in the traditional telecom networks and opened up the networks to future-oriented services. Now, carriers are shifting their focus towards service agility – the most important driver of their business success.

In the past two years, 5G has made great strides in a wide range of industries. More and more B2C services are emerging, like high-speed live streaming, ultra-HD video, cloud gaming, AR/VR applications, and IoT. Users are charged not only for the traffic they use, but also for the bitrate, latency, and connections. This model requires that 5G networks provide differentiated capabilities for various services. In addition to the B2C services, 5G networks are adapted to B2B applications, such as industrial vision, industrial control, and automatic navigation. The B2B adaptation is a tremendous change compared with traditional networks. There is an overwhelming number of industries which impose various requirements to 5G networks, and carriers need to leverage 5G technologies to enable these industries. To address such requirements, networks which previously concentrated on traffic distribution and scheduling now need to be enhanced to support thousands of industries.

To cope with the unprecedented diversity of services in these industries and the fast-paced service changes, reshaping networks is indispensable. In the 5G era, carriers aim to make the software more flexible and agile, which will require quick upgrades and trial-and-error. There is no doubt that A/B testing is the ideal choice.

Huawei innovatively introduces the A/B testing concept from the Internet industry to the Cloud Native-oriented 5G networks and tailors it based on CT service characteristics. Huawei’s 5G A/B testing solution has already been used commercially, making it a first in the telecom industry. This solution has the following advantages:

Microservice architecture as the basis, enabling quick and flexible releases and upgrades

The introduction of Service Based Architecture (SBA) lays the foundation for applying A/B testing.

SBA makes it possible to combine and roll out services as if they were pieces in a Lego set. In such a software architecture, applications or network functions are reconstructed as a suite of loosely coupled microservices. Each microservice is tied to a specific service goal. In a traditional, monolithic application, any slight change or update requires building and developing an entirely new application, which is time-consuming and labor-intensive. Microservices, on the other hand, are independently developed, tested, and released. Independent version numbers are assigned to different microservices. During an upgrade, the involved microservices are identified based on the source and target version numbers, avoiding the need to upgrade all microservices. In this way, software can be deployed incrementally without any wasted time.

Upgrades available any time, no impact on peripheral devices

On traditional networks, only one software version can run at a time. An upgrade requires heavy workloads and is error-prone. Before an upgrade, engineers need to evaluate hardware and software resources and device performance, check service models, and modify the configurations of peripheral wireless and datacom devices. They need to migrate users to other devices before the device is upgraded and migrate the users back after the upgrade. If users cannot be migrated, services will be interrupted for at least 30 minutes. If service interruption is inevitable, the upgrade must be performed during off-peak hours, for example, between midnight and dawn.

A/B testing breaks the rules that were in place with “black-or-white” software version releases and allows two software versions to run at the same time. Three-layer decoupling software architecture with stateless design makes session data compatible among different versions and allows modularized services for different versions to co-exist. The load balancer distributes service data intelligently, avoiding detours through external network devices. Software is upgraded to the target version through rolling updates, without any interaction with, or impact on, peripheral devices. Engineers are relieved from user migration and configuration on wireless and datacom devices. Upgrades can be performed at any time with zero service downtime.

Batch user migration, swift trial-and-error

If software is upgraded the traditional way, all users have to use the new version. Such a large one-off user migration is highly risky. Just a single problem may cause service interruption or inestimable losses. A/B testing allows dialing tests to be performed on the new version, decreasing or even avoiding tests on the test bed. The new features are available only to a small subgroup of users. If problems are found in the dialing tests, the feature update can be rolled back, and only the data of this subgroup of users needs to be deleted. In addition, users can be migrated in small batches. The first batch can be used to verify the feasibility of migration. Similar to the dial tests, any update can be rolled back. Old and new versions can co-exist for a long time while users are upgraded in batches, restricting the impact brought by service failures to the smallest scope while facilitating service monitoring. A batch will be migrated only after the previous batch has proven successful. These dialing test and migration principles improve upgrade reliability.

Huawei has been committed to ICT transformation of telecom services, and is continuously collaborating with carriers to develop future-oriented digital transformation solutions. Huawei aims to leverage agile service release paradigms to build flexible, robust telecom networks, promoting digitalization of the 5G industry. With Huawei’s A/B testing solution, services will be fast tracked to markets with small, quick steps.

The post A/B Testing: Fast Track Your 5G Services With Small, Quick Steps appeared first on Artificial Intelligence.

Over the past 12 months, the banking industry has become increasingly excited about AI. Virtually every leading consultancy has published research on the impact AI will have on the sector and investment continues to pour into developing innovative solutions. But, alongside all the buzz comes the inevitable concern that the implementation of this technology will reduce the need for actual human workers.

The notion here is simple – if a bank can automate a process then surely they don’t need a human to do it. The answer is not as simple, although these sorts of claims are not entirely unfounded. Over the past decade, the digitalization of customer services has led to a decline in the need for front-of-house staff in banks and the subsequent closure of many branches. Similarly, one of the primary areas where banks are implementing new AI solutions is customer services.

Several tier one institutions have developed AI-powered chatbots and virtual assistants. J.P. Morgan uses AI to answer customers’ questions and anticipate what their future needs are likely to be, while UBS’s virtual assistant is powered by Amazon Alexa. These products are the ones that are most likely to replace jobs. The more that these products are used, the more they learn, which means that they exponentially improve in their capacity to assist customers without requiring human involvement.

Aside from chatbots and Robotic Process Automation (RPA), which uses similar technology to automate simple administrative tasks such as inputting customer information, the way that banks are currently using AI is not a considerable threat to their employees’ jobs. A priority for several top-tier banks has been to use AI systems for detecting fraudulent activity or money laundering. This has been particularly successful, having dramatically reduced the time that investigators spent on false positive leads. In these instances, rather than reducing the need for human input, the AI-powered systems have alleviated time pressures on existing investigators and afforded them the time to investigate each case in more detail.

Otherwise, the areas of interest to banks in terms of AI varies considerably from one to the next. Some are focusing on using the technology within algorithmic trading, while others are developing solutions that can offer tailored products to each customer depending on their own circumstances. The crucial point here is that these projects are very much still in development and are yet to be deployed extensively.

The post Is Artificial Intelligence Replacing Jobs In Banking? appeared first on Artificial Intelligence.