Find out how AI Everywhere is key to the infinite edge
If you look at any of the main themes driving innovation in telecom and connectivity right now – 5G, edge computing and IoT to name the big ones – operators and inventors need an ecosystem and network that can deliver at scale the artificial intelligence (AI) experiences and services people will crave in an ever-more connected world. Networks need to evolve from connectivity to predictivity. And AI is a horizontally critical enabler for newly intelligent, reliable networks to autonomously identify and self-adapt to dynamic future demands.
AI for 5G, IoT and edge computing
5G is not only about unlimited bandwidth but also smart networks that can self-heal and self-learn, as well as make accurate predictions. AI embedded into the network can not only predict future demands and problems but also factor in external influences – like mobility and weather, for example – into consideration to make predictions more accurate and context oriented. To give another example, AI can predict the spike in load based on predicted increase in user mobility in a particular location because of special events like baseball games.
"There is a growing need for networks to adapt to dynamic application demands as well as address dynamically to special events, seasonality and so on," Diomedes Kastanis, head of technology and innovation for Ericsson, said recently to CIO.com. Although we have a lot of automation systems and rules to manage and operate networks, he said, it still is not enough to make networks intelligently adapt to the intense changing environment and proactively self-optimize to changing demands.
AI can not only bring in mobility as an external influence to make predictions for network performance, but also turn around networks to add a new value to mobility in general. For example, with self-driving cars next generation networks need to be more than just a fast connection between cars and the cloud. AI can make networks offer “predictivity as a service,” sending data to the car on which locations ahead in their path will there be less or no network connectivity and re-route the car accordingly to ensure that the car stays connected. AI can also notify the car about a truck that is coming down the road and allow the car to make adjustments to avoid an accident.
AI on the infinite edge
Edge computing is all about the shift from the cloud (in centralized datacenters) to the edge. What does the edge mean? It is defined by its context and can be found in sensors, devices or base stations. Let’s say that you are running AI on video feeds from an IP-based camera. There will always be latency if the AI is in a datacenter. What if we can run the AI on the camera itself, moving the compute to the edge?
As the demand for rich, robust experiences booms, the cloud can’t deliver what the network edge can. As the cloud players come to this realization, the landscape changes drastically.
The same goes for IoT deployments. If we are running an enterprise managed service for an operator, we can have applications running in our enterprise cloud. Or on prem in their labs. But this will still not be close enough to the IoT devices producing the data. The AI needs to be in the last mile to be truly real time and be able to predict actions, with an autonomous critical care unit in a hospital only one of many examples.
Machine intelligence combines AI and machine learning
Erik Ekudden, Ericsson CTO, has identified machine intelligence as one of five leading tech trends. Until earlier this year, Erik worked in Silicon Valley as Head of Technology Strategy for the Ericsson Group and CTO Americas, and he defines machine intelligence as a combination of machine learning and AI methods “to create data-driven intelligent, non-fragile systems for automation, augmentation and amplifications” and compares the situation between humans and machines to a teacher in a classroom both mentoring and learning from the students.
You can read more about Erik’s views on machine intelligence and how it will bring simplicity through automation, as well as the other four trends, at Tech Review.
Highlighting innovation at our 2nd Annual Startup Day
On Sunday and Monday, September 10-11, we are hosting our 2nd Annual Startup Day event at the Ericsson Experience Center in Silicon Valley. We will be featuring 14 startups at the daylong event.
Skymind is a great example. It is a deep learning company that supports the open source deep learning framework Deeplearning4j and the JVM-based scientific computing library ND4J. We have been working closely with them on machine learning engagements.
Skymind CEO Chris Nicholson talked about the benefits of predictive analytics in the same CIO.com article referenced above:
"When you can predict capacity problems accurately (for example), you can act pre-emptively to rebalance the load on your network and provision the network with more capacity," he said.
Learn about the Ericsson Garage Startup Challenge and Ericsson Ventures
Want to explore more of our work with startups and innovation? We just held the Ericsson Garage Startup Challenge in Stockholm this week - see who won in front of our Dragon's Den. Or you can read about Ericsson Ventures, through which we invest in leading companies to drive innovation in new areas, accelerate our core business, and generate strong returns.