Fog computing provides robust networking framework for the IoT
The Internet of Things introduces new challenges regarding how and where data processing takes place. IoT devices, sensors, and systems tend to be so widely distributed that a central computing model doesn’t work. Likewise, processing data on the edge may produce mixed results. In a worst-case scenario, systems may not perform up to speed and latency requirements, producing subpar or even dangerous results, particularly in areas such as medicine and public transportation.
“Both the number of devices and the way they are used can impact IoT system performance,” observes Sam Bhattarai, director of technology and engagement at Toshiba America Research Institute in Irvine, CA. “Legacy IT platforms do not offer a viable or sustainable approach for IoT connectivity.”
As a result, a framework called fog computing is beginning to take hold among IoT solution providers. Indeed, a report from 451 Research predicts that the fog computing market will exceed $18 billion by 2022, with the energy, transportation, healthcare, and industrial markets leading in adoption.
Beyond the Cloud
Unlike traditional IT hardware, which tends to be clustered in offices and data centers, IoT devices may be distributed over a farm field or an entire city. “Transmitting and sharing data across devices and into the cloud becomes a problem using a conventional approach,” Bhattarai says. “It’s not practical, cost-effective, or even physically possible to deliver data where it needs to be at any given moment.” In addition, if a node or system in a traditional architecture goes down, the network can fail to operate.
Enter fog computing. By routing data through a “fog node” that connects multiple IoT devices and data points, it introduces a more flexible and efficient way to handle computing tasks. Essentially, processing can take place anywhere on the network when and where it’s needed at any given moment.
“Legacy IT platforms do not offer a viable or sustainable approach for IoT connectivity.” Sam Bhattarai
Fog platforms include three primary attributes: a horizontal architecture that supports multiple domains; a “cloud-to-things continuum” that distributes processing anywhere between clouds and devices; and a system-level framework that extends computing over network edges, through the cloud, and across multiple protocol layers.
They’re designed to deliver ultra-low latency at reduced costs and with less bandwidth, and typically encompass cellular technology—including emerging 5G networks—plus cloud technology, radio systems, protocol layers, and embedded artificial intelligence. Fog systems can incorporate storage devices, system control mechanisms, and more as well.
A Compelling Model
Experts like Bhattarai advise integrators entering the IoT space to familiarize themselves with fog computing and the requirements associated with building out robust IoT network infrastructures. Information is available from industry organizations like the OpenFog Consortium, which includes Cisco, Dell, Intel, Microsoft, Hitachi, Foxconn, and other major industry players, as well as thought leadership powerhouses like Princeton University and Shanghai Tech University.
“Fog computing offers a compelling model for addressing the challenges of highly distributed devices,” Bhattarai says. “It builds a level of intelligence into the network that is critical for the IoT. It has the potential to deliver maximum performance and efficiency.”
Samuel Greengard is a business and technology writer based in West Linn, Ore. He is author of the Internet of Things (MIT Press, 2015).