Understanding the intelligent edge
Connecting state and local government leaders
By bringing artificial intelligence and compute power directly to edge devices, agencies can turn data from their expanding number of internet-of-things sensors into rapid, actionable insights.
Imagine a federal agency wants to improve physical security at its offices around the country. In the past, that might have meant hiring security guards to work around the clock at each location. But now, advanced technology can fill those shoes. Edge devices, which are essentially sensors with additional computing power, can evaluate security footage, detect unauthorized entry and alert on-site personnel to respond to the threat in question -- all in near-real time.
Edge computing sits at the intersection of artificial intelligence, the internet of things and big data and provides the flexibility of a hybrid model that can take advantage of both data center infrastructure and the cloud. By 2025, experts expect there will be 41.6 billion IoT devices, each equipped with sensors gathering information each second. It’s simply not reasonable for all that data to be streamed back to a central data center or traditional on-premises computing environment. By bringing compute power to the edge and integrating increasingly mature AI -- a combination referred to as the “intelligent edge” -- agencies can actually use this rapid expansion of data to help their missions.
Applications for the intelligent edge are wide ranging; it can be used for everything from unmanned systems to predictive maintenance. In this article, we will dive a bit more into the technology before focusing on two government use cases in particular: fleet management and computer vision.
The intelligent edge model
Normally, IoT devices rely on a hub-and-spoke model and a centralized hosting infrastructure. The intelligent edge model, on the other hand, relies on distributed computing. It uses small form-factor servers to move compute and decision making to AI-enhanced devices located close to the actual events as they occur. This decreases latency and conserves bandwidth; depending on the hardware, edge computing can achieve AI inference speeds of 10 to 15 milliseconds -- the blink of an eye.
When it comes to government missions, every second counts. Regarding physical security, for instance, getting alerts about intruders hours or days after their unauthorized entry is useless. By bypassing a long chain of networks and computations, the intelligent edge eliminates delays between sensors and the ability to act on what they detect. Moreover, by reducing the need to transmit or store the full data stream, the transmission is more secure because it is less subject to interception or decoding.
The edge in action
To collect and analyze security footage at scale requires computer vision, which uses advanced machine learning to train algorithms to automatically process and analyze digital images or video files from edge devices. It can enable everything from object detection for employee safety to foot traffic and hotspot tracking in smart buildings. Computer vision at the edge can count people and track the amount of time they spend in a particular spot, detect objects of interest in augmented reality, verify workers are following safety measures and detect intrusions.
Sometimes, image recognition at the intelligent edge doesn’t require optical cameras and sensors but can use LiDAR or depth-sensing technologies. Such newer sensors have the benefit of offering greater privacy, as the images are abstracted into non-identifiable formats.
Another promising use case of the intelligent edge is fleet maintenance. Accelerometers, which sense motion and velocity, and proximity sensors both can be deployed in fleet management and are critical for agencies like the U.S. Postal Service. By equipping mail delivery vehicles with these and other sensors, USPS can understand both operational data about the vehicles themselves, including fuel efficiency and run time and the behavior and safety of drivers.
The bottom line
Sensors at the edge can collect a wide swath of data -- machine vision, position, motion, temperature, humidity, sound, force, leaks and more. However, that data is only useful when it’s analyzed, and that analysis loses value each second it’s delayed. By bringing AI and compute power directly to edge devices, agencies can turn data from the growing number of IoT sensors into rapid, actionable insights.
NEXT STORY: What’s the best cloud for OLTP?