How AI-on-5G Drives the Industrial Transformation - Mavenir

Blog

How AI-on-5G Drives the Industrial Transformation

15 April 2021

AUTHORED BY: Kuntal Chowdhury

How 5G, AI, and Edge Come Together

5G is the latest generation of wireless technology providing faster speed, lower latency, and the ability to connect a very high density of sensors.

5G enables a new connectivity fabric designed to handle the explosion of “things” and new disruptive applications.

Artificial Intelligence (AI) brings a new computing paradigm in which algorithms learn from the data to efficiently handle the ever-growing amount of information sensors generate, including being able to infer patterns and trends.

This leads to a new architecture in which compute is extended to the Edge, where high volume of raw data from sensors connected over 5G is processed with AI/Deep Learning at the Edge.

This is considered the Edge-AI Transformation driven by 5G, in which new applications require graphics processing units (GPUs) and data processing units (DPUs) as units of accelerated compute.

AI-on-5G for Industry 4.0

Digital transformation has been an elusive goal for businesses looking to leverage data, automation, and digitalization as tools for differentiation and competitive advantage.

Governments around the globe realized 5G is the foundation upon which digital transformation can happen and have been allocating spectrum for Enterprises willing to leverage 5G to drive AI and Automation. 5G enables the private wireless infrastructure for IoT and digital transformation to happen, powered by AI and Edge compute.

5G provides the high-performance wireless infrastructure that eliminates wired networks, making it easier to reconfigure factory layouts and connect all sensors and machinery. Edge Computing enables new applications such as Intelligent Video Analytics (IVA), machine learning, and extended reality (XR) that further increase efficiency and automation.

Manufacturing Inspection

Converting complex, manual, repetitive and stressful inspection tasks create a better work environment while delivering quick, accurate, and reliable results. Optical inspection in 20% of the manufacturing workflow can reduce 30% of the manufacturing cost.

Predictive Maintenance

Reliably predicting the remaining life of mechanical equipment avoids unscheduled maintenance and extends equipment utility. State-of-the-art deep learning models are proving capable of predicting the remaining useful life with up to 49% better accuracy than traditional methods.

Extended Reality

Augmented reality (AR), virtual reality (VR), mixed reality (MR) and XR have several use cases in a manufacturing environment, including visualization of factory re-layout and re-configuration, assist maintenance of sophisticated machinery, and assist product reviews. Photorealistic XR is now streamed from the Edge over 5G to connected Head Mount Displays, which enables untethered experiences with real scale and context.

Introducing Mavenir’s Hyperconverged 5G Edge AI Solution

Mavenir is collaborating with NVIDIA to build an AI-on-5G solution that brings together a simplified infrastructure stackable to run 5G RAN, 5G Core, AI, and Graphical Edge computing applications. The solution leverages NVIDIA EGX Platform that provides systems, software, and application frameworks to enable the different workloads.

Enterprise customers can implement a single IT infrastructure that enables the full stack including 5G RAN, 5G Core and Multi-access Edge Computing applications.

Mavenir brings a new paradigm to 5G RAN virtualization leveraging NVIDIA’s highly flexible, general-purpose GPUs and CUDA software, to efficiently offload all compute-intensive applications from the host CPU, while augmenting the system flexibility to be simpler to re-program and easier to upgrade and reuse, while also minimizing maintenance costs. The 5G Core is another key component to ensure the handling of high data throughput. NVIDIA DPUs provide hardware acceleration capabilities so 5G Core user plane function (UPF) software handles all required tasks such as the decapsulation and encapsulation of packets, packet inspection, routing, and forwarding much more efficiently handling more throughput with fewer CPU cores.

AI applications benefiting from 5G such as Internet of Things, (IoT), Computer Vision, Robotics, AR, and VR can also be deployed at the same GPU-Accelerated Edge infrastructure offering innovative new services.

Early Indicators of Future Success

Mavenir has a few ongoing Hyperconverged 5G Edge AI trials including Center Connected Industry (CCI) in Aachen Germany, as well as a set of trials in APAC.

CCI brings research and industries into the community and helps them to build new applications on CCI’s unique testing environment. Mavenir brings a Hyperconverged 5G Edge AI solution to CCI’s logistic Campus Eschweiler and builds applications together with industrial members from the community.

Learn more

For more information, see Mavenir Launches 5G Hyperconverged Edge AI Solution for Enterprise & Telco Market Based on NVIDIA Platforms.
Read about Mavenir AI and Analytics solutions.

Kuntal Chowdhury
Kuntal ChowdhuryClick for Bio
Kuntal Chowdhury is responsible for building products and solutions at Mavenir utilizing cutting-edge Big Data Analytics, AI/ML based applications for network automation and Edge AI applications for 5G Enterprise & IoT segments. He is an industry veteran with decades of experience in various roles such as technology development, product management, systems engineering, marketing, business development, standards & strategy.

Prior to joining Mavenir, Mr. Chowdhury held strategic role at Cisco Systems where defined and drove Cisco’s mobile internet strategy. In his illustrious career path, he contributed to leading standards organizations such as 3GPP, WiMAX and IETF, authoring numerous technical standards. He also held engineering positions at Nortel and Motorola where he led systems engineering and RF optimization of large-scale deployments of CDMA, GSM and UMTS networks.

Mr. Chowdhury has over 50 granted patents and many filed patent applications.

Need More Information?

Building the future of networks with cloud-native software that runs on any cloud and transforms the way the world connects.