Why the intelligent edge? Paul Miller, CTO of Wind River, will explain why AI on the edge is key to the machine led economy.
Paul Miller:
Hi everyone. My name is Paul Miller. I’m the Chief Technology Officer for Wind River, and I’m really excited to join you at this summit today and share some of the work that Chooch and Wind River have been doing together, and some of our vision for what we call the intelligent systems and machine-lead economy.
Paul Miller:
Without further ado, I’m going to use some slides to talk through this. You can see here that one of the things that our company really believes in, excuse me, is what we call the intelligent edge. And we’re seeing the emergence of a huge amount of data processing and compute moving to the edge of networks. And this is affecting both AIML, as well as data processing and the way that these things are realized intelligent compute systems. This is really being driven by a set of use cases that we see, and whether it be autonomous vehicles with vehicle-to-vehicle accident avoidance algorithms, or telemedicine, or even the things like commercial drone delivery, or military drone control, you’re seeing an incredible amount of compute, video, and machine learning processing happening throughout the network, and particularly, emerging at the edge.
Paul Miller:
Wind River as a company, we’ve been in business for 40 years and we’ve been involved in a variety of different market verticals. You can see here from personal cobots and energy, as well as telecommunications, infrastructure and aerospace and defense. We provide real time systems that a lot of our customer was used to build intelligent systems and intelligent edge products. As a result, we’re seeing some interesting things in the market as things evolve. We recently did a survey with Forbes, looking at the 13 characteristics of intelligent systems companies. And we found that a variety of technology attributes were present in what our customers are trying to build for products. And you can see here, some things that I think this audience would believe are common place, right? The ability to have cloud connectivity, the emergence of its strength and fire edge compute, the machine learning and AI based actions of total automation, event detection and resolution predictive outage avoidance, acting based on sensory data and even the learning and machine learning applied to these problems.
Paul Miller:
And so, as we look at our customers, we’re seeing them try to solve these problems. As a result, several years ago, we made some investments to shift from being purely an operating system company, one that provides embedded operating systems to one that provides an entire end to end solution that really embraces what our customers need for these types of systems. And this spans the gamut from development of these systems to actual deployment and live operation of these systems, as well as servicing them. And this comprehensive offering we call studio. And the work we’ve done with Chooch in the industry here is really tied of this offering that Wind River has created. We also see that one of the enablers for machine learning and AI and video processing at the edge of the network is really driven by this connectivity change that’s happening at the last mile. Wind River for decades has been in this device edge, in embedded systems, but now also with infrastructure edge with 5G, which we’ll talk about today, and with applications that span the entire gamut here.
Paul Miller:
And this is really the transition of the intelligent edge systems becoming hyper-connected, right, as devices and being able to use cloud infrastructure and backend processing, for example, for heuristic training, as well as the execution of the actual algorithms at the realtime edge, I’ll encourage each of you in the audience to take a look at our website and take a tour through what we do here for a development tool set, as well as operational environments. It’s a pretty interesting stuff and really the foundation of the work that we’re doing with Chooch.
Paul Miller:
In particular, the work we did recently was really focused on 5G. And we’re seeing a shift in this technology space from monolithic bespoke appliance based products to one that is now a disaggregated virtualized infrastructure structure. And this is kind of a revolutionary thing for the telecommunications environment. And this is really the layer cake of a hardware platform, of virtualization infrastructure, and then applications that run on top of that infrastructure, i.e. the cloudification of the edge of the carrier network. And this has really resulted in the emergence of a new of 5G deployments that are really based on what we call distributed cloud technology or edge based cloud. Wind River Studio includes this capability and as you’ll see shortly, this is the work we did together with Chooch and NVIDIA, at the recent NVIDIA event.
Paul Miller:
The foundation for this technology and Wind River Studio being used for 5G is driven by an open source community called StarlingX. And this was jointly founded several years by Wind River and other commercial entities to create and really embrace the new challenges of a distributed edge-based compute system. As you look at the challenges of doing things like 5G at the edge or edge-based virtualization, you have a significant operational challenge to embrace that distributed system and be able to operate it as a geo-separated function, right? These sites often number in the tens of thousands or devices in the tens of thousands that are separated across hundreds of thousands of miles of topography. And so the classical Kubernetes architectures and OpenStack architectures, which were really designed for a single monolithic data center environment are not really meeting the challenge of that type of architecture.
Paul Miller:
So StarlingX as an open source initiative was founded, and we do all the work in studio around our cloud technology in the open source community. Every feature we write and every bug we fix is contributed upstream into that initiative. And so as a result, the belief in open source and open source approaches is very much in the DNA at Wind River. The result of all this work, as some of you may have seen from press releases is that Wind River’s technology has been selected for the national deployment of 5G by Verizon. And we’re using this virtualized distributed Kubernetes technology for edge network applications that obviously clear the path for advanced edge applications, AI, and machine learning, and industrial environments, drone systems, automobiles, these sort of things that are really the new revenue enablers that are driving the build out of the 5G network.
Paul Miller:
Recently with the work we did with NVIDIA and Chooch, just a couple of slides in this here. So you can see how this all comes together. You can see here Chooch sitting in the application marketplace capable of being run on Wind River Studio as a virtualized environment. And if you see on the left hand side there, this EdgeX solution with NVIDIA DPU, GPU and Wind River Studio enables us to host applications anywhere from the core of the network, even in the public cloud, all the way to an incredibly small footprint single server model at far edge. And this is a applicable of course, for high volume 5G deployments, as well as on the left, you can see here a variety of enterprise applications like stadium, manufacturing, warehouse, et cetera, where AI based applications are of particular interest. We have one customer, for example, that’s T-Systems division of Deutsche Telekom in Germany that’s using these type of technologies in the manufacturing environment for augmented reality for the manufacturing environment.
Paul Miller:
Pretty exciting application there that is emblematic of exactly this type of architecture that you see here. And so obviously, running the Chooch application directly in Wind River Studio. We were incredibly pleased to see how smooth this went. The integration of the application and building the intelligent video analytics application to run was very, very seamless, integration being a… Wind River Studio being a Kubernetes-based environment is very straightforward to integrate that. And that enables you to have a highly scale edge-based artificial intelligent video analytics solution that can deploy at incredible scale in networks. We’re also seeing some interesting things happen over time in our other industries that I thought I’d mention today. And this is really the concept of convergence of edge with the cloud.
Paul Miller:
Where Wind River had been previously involved in embedded systems, you can see here in the aerospace and defense field, radar sensor systems, weapons control, navigation computers, these sort of things. But now these systems are becoming cloud-connected. And even the chief software officer of the US Air Force was quoted as saying, we’d like the airplane to land with different software than when it took off. And this is the modernization of these environments. And of course, these sensor systems that are on these airframes are making use of extensive video processing and machine learning algorithms. I’ll also mention in the industrial sector, we see the same transformation. As we mentioned, just a moment ago with what, with what T-Systems is doing in Germany, the emergence of industrial 4.0 and things that used to be for us, just simple embedded systems now becoming cloud-connected, AI-aware and processing machine learning at the edge, and especially tied to vision systems for the automation of the manufacturing environment.
Paul Miller:
Interestingly as well, in the energy sector, we’re seeing a true transition of the centralized power distribution technology that was coal fired power plants to now becoming distributed wind and solar based. And this changes the challenge of managing such diverse environments into a distributed architecture, right? We have control planes that are running across a variety of systems tied together with cloud virtualization, enabling the end to end automation of these entire environments. And of course, these hugely scaled systems that we’re talking about not really possible to operate these without the presence of artificial intelligence and machine learning, to provide a high level of software automation and control over these environments. We also see, as part of this evolution, the need to be able to dynamically manage workloads and operationalize this infrastructure. You can see here, the ability to deploy and operate these applications. Whether it be Chooch or other companion applications that run in these systems, we have the ability to now have a single pane of glass control to deploy these services in Kubernetes, private clouds, OpenStack as well as public cloud infrastructure, as you can see the demonstration running here.
Paul Miller:
But even running that in contemporary software methods, such as CI/CD and agile, where we can have Git and Jenkins and other automated systems automatically deploy and update the applications in the cloud infrastructure via infrastructure as code and CI/CD mechanisms. So a very modern approach to the way that you manage and deploy, this system is really important for the operators of these systems, our customers, because these diverse systems are incredibly difficult to manage and operate. And so just to wrap up a course to plug for Wind River here, I certainly really appreciate the invite here. We’re very excited about the work we’ve been doing and what we call Wind River Studio, and really looking forward to continuing to work with Chooch and our partnership there, doing great things with artificial intelligence, machine learning and image processing. Thanks for your time today and thanks for having me.
