By Michael Zurat, Senior Solutions Architect, GDIT
5G, Artificial Intelligence and Edge Computing have been areas of focus in the IT space for some time. What’s been given less attention, until recently, is how these three powerful technologies interact with one another and how, as one example, 5G is precisely what enables AI capabilities at the edge.
The proliferation of data-creating devices – whether that’s smart phones, Internet of Things (IoT) devices or drones – means that the sheer volume of data that exists is too large for humans to examine, and it has been for a while now. AI algorithms enable us to identify patterns and outliers and to either address a problem or pinpoint the things that require human intervention. 5G enables us to deploy those algorithms faster and more securely to devices at the edge, making them more powerful and capable than ever before.
Massive Machine-To-Machine Connectivity
Here’s why: 5G enables massive machine-to-machine connectivity. This allows for connected sensors and devices to “speak” to each other – sometimes without connecting to an enterprise or hub, such as in a mesh network. This allows for devices to manage themselves and react to data inputs from other devices, even without connecting to a cloud or centralized server. By leveraging embedded algorithms and even machine learning at the edge, we’re at the beginning of what will become a smart and independent mesh network of devices.
One example is how smart city traffic lights can automatically adjust to control or regulate the flow of traffic without communicating to a central server. As edge computing like this becomes ubiquitous, it will start to compete with the cloud as a compute and storage commodity. This may provide alternative physical locations for storage and compute for existing cloud vendors, and it could disrupt them altogether with potential new offerings from cellular network operators.
We’re already seeing the beginning of this from content distribution services. Netflix stores data in massive data centers in Virginia and California, but locally they’re using Akamai relays to cache data at the neighborhood level.
Blurred Lines Demand Advanced Cybersecurity
In this example and hundreds like it, as data moves from place to place with encryption at both ends, 5G and edge are, in effect, blurring the lines and creating a larger and more diverse attack surface in need of advanced cybersecurity. On top of that, more complex system architectures are built on more complex Infrastructure as a Service (IaaS), Software as a Service (SaaS) and Platform as a Service (PaaS) services that abstract the cloud and edge compute and storage locations. There are also more complex layers of containerization involved. No longer will cyber teams be able to think about security in local, cloud or on-prem contexts.
In environments where everything is virtualized and when it’s not exactly clear where the data is at any given moment, or when serverless systems are being leveraged, cybersecurity becomes more important than ever. So, too, does having a secure software supply chain. Look no further than the recent log4j breach.
As ever-more complex systems enable ever-more complex capabilities, it can be easy to focus on the delivery of functionality without looking back at how and what software is being used and how secure it is. The stakes are even higher for government and Department of Defense customers and their mission partners. As a systems integrator, GDIT has broad and deep capabilities that address these complexities and the stakes – both today and for the future. This includes our 5G Emerge Lab that proves out 5G solutions and demonstrates how we can securely connect and protect 5G-enabled devices, data, and applications across the broader enterprise.
Underpinning Technologies Paved the Way
Paving the way for what’s to come with regard to AI at the edge are a host of underpinning technologies. Containerization, as mentioned above, helps to quickly move applications into the cloud. Virtualization abstracts applications from physical hardware. Graphical Processing Units, or GPUs, allow us to accelerate workloads. Gaming engines gave us the blueprint for reusing software components or to more quickly port applications to other platforms.
Today, GDIT is leveraging these advances as well as the capabilities offered by 5G to enable AI at the edge on a diverse array of use cases – from training and simulations that leverage virtual reality on edge devices, to medical trainings with smart mannequins. We are also exploring how to use AI and guided assistance within Denied, Disrupted, Intermittent, and Limited bandwidth (DDIL) environments. Another use case involves deploying smart warehouse technology on edge devices for logistics and supply chain management purposes.
What is clearer than ever is that teams are just scratching the surface of the enormous potential of AI at the edge. 5G, and the continued bandwidth enhancements that will come after it, will expand that potential exponentially. And we stand ready to capitalize on it, exploring the art of the possible, for customers.