The average inference speed for cloud-based AI hovers around 1.5 seconds. The intelligence edge cuts it down sharply to 10–15 milliseconds. This drastic reduction in latency alone makes a number of futuristic technologies — such as autonomous vehicles — possible.
The advent of cloud computing has set off a colossal centralization fever that has caught almost every business that understands the importance of a digital-first business strategy. Even the world’s governments and public sector organizations are leveraging the advantages offered by cloud computing. Easy access to data, powerful analytical tools, and improved business agility have enabled organizations to make more “intelligent” and informed decisions than ever before.
However, over the next few years, a rival computing architecture approach — decentralization — will witness a sharp uptick in popularity, fueled by edge computing. A combination of edge computing and artificial intelligence technologies will spawn a whole new generation of capabilities that were hitherto mere science fiction.
Before we explore the tremendous potential of these technologies, let’s take a minute to understand them.
Edge Computing and Intelligence Edge: An Introduction
In the cloud computing approach, data is collected from various endpoint devices and sent to centralized cloud servers, where the data is stored and analyzed. Insights or responses generated by cloud-based tools are then sent to the endpoint devices.
Edge Computing brings the computing resources to the endpoint devices, thereby making it possible to analyze incoming data at or near the point of data collection.
Intelligent Edge is applying artificial intelligence technologies to edge devices to collect, analyze, and respond to data in near-real-time. The primary objective of Intelligent Edge is to make split-second decisions a reality at the point of interaction for the endpoint or end-user.
Applications of Intelligence Edge in the Real World
International Data Corporation predicts that there will be a whopping 55.7 billion connected devices by 2025, which will collectively generate 73.1 ZB of data. As consumers and businesses alike will continue to rely on smart devices for everything from communication to physical security, they will need and demand faster computing. Intelligent Edge is uniquely equipped to fulfill this need.
Here are some use cases where intelligent Edge plays a pivotal role:
1. Autonomous Vehicles
The present generation of autonomous vehicles collects environmental data — road conditions, vehicle positions, road signs, and other data — in real-time and sends them wirelessly to cloud servers, from where they receive driving instructions. As we know, the AI inference speed stands at an average of 1.5 seconds for cloud computing.
For a vehicle traveling at 60mph, a 1.5 second lag in response translates to 132ft of distance on the road! That’s after ignoring the inherent latency of 4G technologies, as these challenges are being addressed with 5G.
Intelligent Edge reduces this decision-making process to a split-second activity, thereby allowing autonomous cars to make near-real-time decisions.
2. Patient Monitoring
Patients are routinely hooked up to an array of medical devices that do not share data with each other. Some hospitals collect all data and store them on third-party cloud servers, and analyze them there. This poses several privacy and security concerns. Not to mention the strain on the hospital’s resources employed in storing so much data.
Intelligent Edge analyzes patient data at the collection point — the monitoring devices — and provides the staff with real-time insights and alerts to act upon.
3. Smart Homes
From refrigerators to printers to security systems, almost every device we use at home connects to the internet. Intelligent Edge allows these IoT devices to be equipped with computing capabilities that can deliver us faster information. More importantly, sensitive data they collect must not be stored on remote servers that come with their own sets of privacy and security problems.
Voice-based assistants like Google Home and Amazon Alexa are excellent examples of intelligent Edge in action.
4. Predictive Maintenance & Improvements
Manufacturers have forever desired to detect potential failures and fix them before they disrupt the production lines. Edge computing makes this possible using an array of monitoring sensors that analyze machine health data in real-time and alert relevant staff on potential issues.
Intelligent Edge can even make real-time operational optimizations a reality. Currently, operational data is analyzed centrally, and production improvements are rolled out periodically. With intelligent Edge, manufacturers will be able to identify inefficiencies in real-time and make improvements swiftly across their production lines.
There are numerous other applications such as predictive maintenance, facial identification, remote surgery, and so on, where intelligent Edge’s low latency is proving to be a gamechanger.
Components of Intelligent Edge Ecosystem
Intelligent Edge’s real-world applications are varied, complex, and potentially pervasive. Here’s a quick overview of the various components that support and power the Intelligent Edge ecosystem:
1. Edge Computing
You are already familiar with edge computing. It decentralized computing resources and brought them closer to the endpoint devices. Some excellent examples of edge computing include P2P computing, blockchain, content delivery networks, grid computing, and so on. All of these examples have one thing in common — they pool computing resources from several different devices without the necessity of central coordination.
Edge computing offers several benefits like enhanced endpoint security, low latency, cost reduction on bandwidth for data transfer, and resilience of device functionality against network disruptions.
2. Edge AI
For years, data scientists have considered AI to be a software challenge. They have been traditionally hosted on cloud servers, so physical hardware never drew their attention. The emergence of Edge AI has turned AI development on its head. With AI being hosted on endpoint devices, the need for special-purpose chips, which are optimized for AI and their tasks, has risen.
iPhone’s A11 Bionic chips are a great example. The chip can perform 600 billion operations per second, and this is what made real-time facial recognition possible for the iPhone X series. So, edge AI is as much about software as it is about hardware.
Edge AI’s many benefits include real-time data processing, extreme scalability made possible by advancements in consumer-grade devices, and contextual analysis for real-time feedback.
3. Edge Devices
Smart devices make it possible for AIs to “understand” the world around them in a richer, deeper, and “real” manner. Instead of relying on humans to input structured and formatted data into computers, AI can use various sensors to see, smell, taste, feel, and hear the world around them.
Besides the standard sensors like accelerometer, monitor detectors, humidity sensors, light sensors, and so on, edge devices powered by AI use far-infrared cameras, ground-penetrating RADAR, and other sophisticated sensors to offer unprecedented insights in real-time.
An array of sensors, powered by AI, makes it possible to automate data collection, analysis, and even operations using that data.
4. Edge Data Management
The amount of data generated by edge devices is astronomical. By way of example, an autonomous car generates 19TB of data per hour. Now, multiply that by the number of autonomous cars that will ply the roads in the near future. Forget storing that data, our current infrastructure is not equipped to handle the data transfer from edge devices to the cloud, which is set to exceed the capacity of underground fiber infrastructure.
Naturally, a bulk of edge data will be analyzed and stored within edge devices. Presently, only 10% of data is processed in decentralized locations. According to Gartner, that number will hit 75%. Edge data management offers two significant benefits over centralized data management — decreased data management costs, and real-time feedback and response.
5. Edge Infrastructure
Edge infrastructure is more than just edge devices. It involves three critical elements — edge devices, connectivity, and centralized data center (cloud or in-house physical data center).
If the computing happens only at the edge device, then it’s local computing, not edge computing. Edge computing involves consistent, low-power connections to the core (cloud or centralized in-house servers). While most of the data is analyzed at the end, some data may be processed or stored centrally. Usually, this means that most of the data collected by edge devices are analyzed and discarded by the edge devices. However, some of the critical data is transferred and stored at a centralized location for the longer term.
Edge devices and the central core are not technological challenges anymore. However, consistent, low-power connectivity continues to be a challenge. The various currently available options are vastly incapable of supporting the edge infrastructure. Although a low-power option, Bluetooth is not consistent; Wi-Fi, although it offers consistency, has a very limited range; 4G LTE, despite its excellent consistency, is power-hungry and low on bandwidth.
5G connectivity is poised to address all of these issues and revolutionize Intelligent Edge.
Intelligent Edge: Rocket Fuel to Technological Innovation
A staggering number of use cases are waiting for intelligent Edge to mature and become widely available. Remote surgery, for instance, which involves a real surgeon operating on a patient through a robot, needs haptic feedback and real-time contextual insights with near-zero latency. Even millisecond lags can have disastrous consequences.
Likewise, many use cases demand low-latency infrastructure, such as smart traffic management, smart grid management, AI-driven building inspections, and more. Intelligent Edge not only fulfills this demand but also ushers in a new revolution of hardware and software development, which will be optimized to support and take advantage of Intelligent Edge.