Solution-ready packages overcome IoT challenges

It’s relatively easy to imagine an IoT solution that can improve business performance by applying powerful analytics to data captured from numerous diverse channels. Bringing it to life is rather more challenging.

Any IoT solution has multiple aspects that call for extensive OT and IT expertise, not to mention the large numbers of sensors and suitable edge computing, communication, and cloud services needed. Key challenges include integrating the collection of industrial equipment data based on a variety of standards with edge device management, wireless communication, and data analysis.

Moreover, creation of visually intuitive reports needs to be integrated on cloud platforms. AI models may also need to be created and combined with expert knowledge. It is also vital to ensure compliance with data storage regulations prevailing in each territory. And, of course, nothing can be connected to the IoT until trust in hardware platforms and data-security mechanisms is established.

Co-creation strategy

These barriers could slow the pace of IoT adoption, and thus business improvement, despite the technology’s enormous potential to deliver improvements far more quickly than traditional human effort alone could possibly contemplate. Solving the diverse challenges to create a cohesive solution is no small feat, but specialised industry-sector knowledge is also needed to create a turnkey IoT solution.

A new way to break the apparent impasse is emerging, leveraging a comprehensive IoT Platform as a Service (PaaS), and off-the-shelf Solution Ready Packages (SRP) developed by co-creation partners from a variety of industries. System integrators can install these SRPs at a customer’s site and so shorten the time to deliver new IoT solutions.

WISE-PaaS

Advantech’s WISE-PaaS industrial cloud platform provides a suitable basis for this model. Indeed, Advantech has begun working with co-creation partners, Domain-Focused Solution Integrators (DFSI) to create SRPs that are ready to run on WISE-PaaS.

There are four key elements of WISE-PaaS that provide the basic services and frameworks on which the SRPs can run. The first of these is WISE-PaaS/SaaS Composer, a cloud configuration tool with visible workflow that supports customised component plotting for simple and intuitive 3D modelling application and interaction. Together with WISE-PaaS/Dashboard, Composer presents critical management data in an intuitive display to help extract valuable data and improve efficiency.

In addition, WISE-PaaS/AFS is an AI framework service that supports neural network training and deployment. A simple drag and drop interface allows developers to quickly input industrial data. When combined with AI algorithms, the service builds an inference engine that can be automatically deployed to edge computing platforms. Model accuracy management, model retraining, and automated redeployment are all built in, and WISE-PaaS/AFS simultaneously controls multiple AI models in the application field - offering automated model accuracy improvements and lifecycle management.

The equipment network connection remote maintenance service framework WISE-PaaS/APM (Asset Performance Management) connects to a wide array of on-site industrial equipment controls and communication protocols. It supports the latest edge computing open standard, EdgeX Foundry, and includes built-in equipment management and workflow integration templates. Jointly with the AFS, APM accelerates Machine to Intelligence (M2I) application development.

There is also a micro-service development framework that simplifies and accelerates program development. Micro-service functions such as service finding, load balancing, service administration, and configuration centre, all offer built-in flexible support mechanisms.

WISE-PaaS can move across multiple clouds and offers flexible expansion, multi-tenancy, superior reliability, multiple database services, AI model training and deployment framework services, visualisation services for dashboards, and multi-level data security/management services.

So far, 32 SRPs have been announced within categories such as Industry 4.0 and smart factory, energy and environment, edge intelligence, smart city, retail, logistics, and healthcare. Individual SRPs are now available for factory energy management, operating and managing CNC machines remotely, smart parking, electric vehicle charging management, solar power management, IoT security, intelligent water treatment, equipment vibration monitoring, streetlight control, license plate recognition, retail AI for sales prediction, hospital management and security, and many others.

Edge compute for IoT

To further strengthen the IoT value chain, leveraging WISE-PaaS and co-created SRPs, Advantech recently announced an edge-computing initiative with partners AMD, and Mentor, a Siemens company. Aiming to make AI technology more accessible and easier to implement, the three partners have created a powerful edge computing platform that leverages open standards and allows customers to concentrate on application development.

Integrating AI technology into embedded systems remains a major challenge, despite the many open-source AI technologies now available in the market. Because so many different elements must be brought together, including data collection, model training, and deployment of the inference system, nobody can do Advantech IoT (AIoT) alone. To address this, Advantech and its co-creation partners have created an edge computing platform to accelerate AI deployment.

This is based on Advantech’s embedded platform, containing the latest AMD Ryzen Embedded V1000 processors, and running the Mentor Embedded Linux operating system. The AMD Ryzen Embedded V1000 processor supports frameworks, libraries, tools, and compilers for machine vison applications that leverage the powerful Ryzen processor with Radeon Vega GPU technology. Open standards such as OpenVX and OpenCL API supported on the Linux kernel, this platform allows users to migrate machine learning across diverse hardware architectures for a variety of AI applications.