Speed is crucial for distributed network success. At IoT Evolution Expo 2023 in Fort Lauderdale, FL, representatives from Internet of Things (IoT) solutions distributor GetWireless discussed ways transmission speeds in an IoT network can be enhanced.
Titled "The Importance of Speed: Locally and in the Cloud," the discussion began with the router topic.
"For me, it's all about the router," said Mike Anderson, Senior Director of Product Management at GetWireless. Processing data in the router (before it gets to the cloud) is the best way to alleviate data strain on your network. The goal is to process the data closest to where it been collected.
Anderson referenced a use case involving meter reading services, where human errors were leading to a lot of mistakes. By moving some of the human duties to the processing center in the router, the company was able to increase reliability. "The extra processing in the router helped mitigate the repeated human errors," he said.
"My edge is at the sensor," said speaker David Smith, Vice President of IoT solutions at GetWireless. The largest drain in device batteries comes from transmitting data; so, save transmission power and enhance the life of your batteries by processing data at the sensor level, Smith explained.
One way to control costs and improve performance is by processing at the sensor. "If you have a million sensors, edge processing can make a big difference," Smith said. Local processing also helps maintain privacy by allowing you to transmit to the cloud only the data you need.
“Intelligent endpoints can do more than just process data,” Smith continued. Moreover, IoT devices can assist with data filtering, unit changes, and decoding signals, too.
"All of this can be done locally," Smith said. "You can start using your sensors to predict changes. That's a huge benefit of doing processing at the edge."
Sensors can be programmed to recognize time-change models, which allows them to predict when breakdowns might occur based on previous experience. “Implementing proactive changes using software processing at the edge completely turns the way we do business on its ear.”
How do IoT sensors perform all these tasks? By using tiny machine learning, or Tiny ML, to process data in the sensor, Smith detailed. It's a version of AI that lets you perform on-device sensor analytics at extremely low power. The product typically carries 20 KB worth of memory running on an 8-bit processor.
"Its ML deployed in a highly constrained environment," Smith said. "The amazing things you can do with it makes it incredibly valuable."
As with all distributed systems, network security is another big issue. “An edge router allows you to run data through security protocols before the data reaches the cloud,” Anderson added. He also noted that the more sensors you have, the more attacks you're going to experience.
Transmitting wireless communications from devices to the network comes with its own set of security issues. The idea of employing encryption at the edge is there, but it's difficult to deploy computing power to the task. "That's an issue that still needs to be addressed," Smith concluded.
Edited by Alex Passett