Fredrik Nilsson is Vice President of the Americas for Axis Communications, overseeing the company’s operations in North and South America.
This year marks a major milestone for the security industry: 25 years ago, my company, Axis Communications, introduced the first internet protocol (IP) camera. Might not sound like a big deal, but this marked the beginning of the shift from analog surveillance to today’s network solutions. Without the IP camera, modern video/audio solutions and analytics would not be possible. And even as cloud computing has experienced a major rise over the past decade, today’s organizations are re-embracing the network edge.
Twenty years ago, another first occurred: the first IP camera with built-in edge analytics (video motion detection) was released. Today, edge devices use analytics for a broad range of purposes, ranging from security to business intelligence, but in the early days of analytics, limiting factors like bandwidth, processing capacity and storage issues hampered the technology’s ability to find mainstream success. As these elements have improved, so has the power — and usefulness — of modern analytics.
The prevalence of hybrid systems incorporating both cloud and edge solutions has helped analytics live up to its early promise. Edge devices are reemerging as an essential tool for today’s organizations, and the broad range of available analytics tools have helped those organizations make improvements that go far beyond security.
The Rise Of Artificial Intelligence
Artificial intelligence (AI) has become an overused term — today, it is often applied to anything that has to do with any form of digitalization, even if it doesn’t technically qualify. It is the relatively recent rise of more complex forms of artificial intelligence, including machine learning and deep learning, that has boosted camera/sensor capabilities. Today, high-powered Deep Learning Processing Units (DLPUs) are enhancing and opening opportunities for new analytics applications. Without this technology, modern analytics would not be possible.
MORE FOR YOU
The technology has enabled cameras to do things like object identification and tracking, adding the ability to differentiate between a human, or a car or a truck quite accurately. And thanks to modern cameras and chipsets, these analytics can now run at the network edge, rather than the cloud. This helps address issues like bandwidth and storage while also increasing responsiveness, since analysis and the resultant action can all occur at the edge, reducing latency.
Of course, AI has powered the development of edge analytics that go beyond security — and beyond cameras. Anytime a user unlocks their phone with facial recognition, that is edge analytics in action. Retailers use analytics to track inventory through both video surveillance and point-of-sale (POS) systems. Self-driving cars use AI-powered analytics to improve their object recognition and reaction capabilities. It is almost hard to believe, but today a car or a bus can be an edge device, no different from a phone or a camera.
Going Beyond Security
What is it that makes the edge unique — and more advantageous than the cloud in many instances?
For starters, data and storage are a major concern. A lot of data is needed to train and refine analytics. The more data a system is fed, the greater its accuracy and the greater degree of classification is possible. Unfortunately, devices that generate a lot of data but need to send it to the cloud for processing consume a lot of bandwidth and power.
Today, that data can often be processed at the network edge, and only metadata needs to be sent to the cloud for classification. The case of recorded video, that data might be “a blue car is coming left to right at 35 MPH at 2:51 p.m.,” making it possible to pinpoint what is happening in the video and understand it without needing to send the entire video to the cloud. Similarly, a retailer tracking inventory levels might upload metadata from self-checkout systems, but not things like security footage or payment information. And users who opt for facial recognition to unlock their phones need that to be done at the edge — or they would risk being locked out of their own devices in areas with poor coverage.
Edge computing has also enabled real-time data gathering and situation monitoring that would have been possible otherwise. A chemical plant might train sensors to gauge heat levels or even track leaks and spillages. Retailers can analyze POS data to track consumer spending habits and optimize staffing and inventory needs. Healthcare centers can use audio analytics to detect coughing, cries for help and other signs of patient distress. Modern analytics has allowed these devices to become more than just security tools.
Getting Started With Modern Analytics
Before getting started with edge technologies, organizations should ask themselves a few questions to ensure it’s a good fit for them:
• What is the nature of the problem you are trying to solve? Is security the primary need, or is there a safety or business intelligence component?
• What analytics are available, and what are their capabilities? Manufacturers and integrators can work with customers to identify the right tools and explain their capabilities and limitations.
• Can your business infrastructure support the analytics you need? System operations and training, budget, maintenance and support are all important considerations.
• How scalable is the solution? Organizations grow, and they need solutions that grow with them. Open-architecture solutions and platforms with strong third-party support can help.
Shaping the Future with Analytics
Edge computing has revolutionized analytics, enabling new use cases across a wide range of devices. Analytics algorithms that were once too inaccurate (or required too much centralized processing power) to be functional can today run efficiently at the network edge. Cloud computing remains important, but the ability to process vast amounts of information on the edge has brought technology like self-driving cars out of science fiction and into reality.
The dramatic improvement in analytics capabilities has a growing number of organizations reassessing their information processing priorities and taking their operations back to the edge. With so many employees continuing to work remotely in the aftermath of the Covid-19 pandemic, bandwidth is scarcer than ever — making this the perfect time for today’s businesses to look to the edge and the vast new possibilities that come with it.