When it comes to processing huge mounds of data, it's always been about efficiency for businesses as their databases continue to expand.
Latency and connectivity issues have been consistent roadblocks for businesses that process high volumes of data from computers, applications, and Industrial Internet of Things (IIoT) devices.
Edge computing has, well, taken the edge off these issues.
The edge computing model allows businesses to optimize their systems by managing real-time process control or detecting anomalies in their manufacturing processes closer to where data is generated rather than relying on data centers or third-party cloud services.
This also alleviates some security concerns. Edge computing provides better data security and privacy protection because data is processed at the edge rather than traveling to central servers. Privacy can be compromised when data hosted on centralized servers are hacked because they can contain more comprehensive information.
The growing presence of artificial intelligence (AI) and machine learning (ML) also reduces latency and connectivity issues because these applications work best by fetching and processing data in large batches.
Edge computing devices usually are lightweight and supported in various form factors. Sometimes, edge computing can also refer to a hardware-agnostic software component that can run on-premise hardware or in virtualized environments.
In this week's New Tech Tuesday, let's look at AI-powered edge computing devices and sensors from Kneron, Advantech, and Axiomtek.
Kneron’s MINI-AI-520 AI Edge Computing Module brings AI and ML inferencing to existing systems. The device is based on the KL520 Neural Processor Unit (NPU) and can connect to embedded systems that support PCI Express Mini Card (mPCIe) modules. The KL520 provides an AI computing performance of 0.35 TOPS per watt, making it ideal for remote, mobile, and unmanned applications. Kneron also has the M2AI-2242-520 AI Edge Computing Module, which is available individually (PER-T520-MIAI-A11-0001) or paired with an EverFocus EUA1200 Full HD Camera (PER-T520-MIAI-A11-K001).
Advantech’s VEGA-340 Edge AI Acceleration Module can boost video infrastructure performance from edge contribution to cloud distribution. The low-power device has plug-and-play AI inference capabilities for vision applications, including video surveillance, medical diagnostics, and machine vision. The modules also feature Intel® Movidius™ Myriad™ X Vision Processing Units (VPU) that are optimized for edge processing, computer vision, and deep learning. Developers also can leverage the Intel OpenVINO™ toolkit for inference performance optimization.
Axiomtek's AIE100-903-FL Edge AI Powered by NVIDIA® Jetson™ Nano is designed for AI and edge computing, smart retail, and smart city applications. The box-level system runs the Ubuntu 18.04 operating system on the NVIDIA® Jetson Nano module. The ultra-compact (147.4mm x 129.8mm x 34.6mm) system is equipped with a 15W GbE Power over Ethernet (PoE) port for power and video transmission. The 1kg system has a -30°C to 50°C operating range, 3Grms vibration resistance, and 10 percent to 95 percent non-condensing humidity tolerance. Combined with an optional IP42 cover kit that supports semi-outdoor applications, the system is suitable for wet and cold environments.
Edge computing allows businesses to control real-time processes, improve bandwidth utilization, and ensure smooth operations even without an internet connection. Edge computing does not replace but augments cloud computing, which means designers and developers can create innovations to improve the technology.
Tommy Cummings is a freelance writer/editor based in Texas. He's had a journalism career that has spanned more than 40 years. He contributes to Texas Monthly and Oklahoma Today magazines. He's also worked at The Dallas Morning News, Fort Worth Star-Telegram, San Francisco Chronicle, and others. Tommy covered the dot-com boom in Silicon Valley and has been a digital content and audience engagement editor at news outlets. Tommy worked at Mouser Electronics from 2018 to 2021 as a technical content and product content specialist.