We’re in an unprecedented time with the accessibility and volume of data we can retrieve, especially with the prevalence of IoT devices. But data is only as good as what can be processed and the speed of computing devices.
All computer devices technologies have limitations related to the speed of electron motion, which is impacted by the speed of light. Ultimately, there will be a finite limit to how much computing power any device has, and with the ongoing data processing needs and capabilities, businesses need computing power to manage, route, and process data effectively.
Limitations to Computing Devices
Computing devices are limited by processing power and memory density – it’s about how fast they can move information from one place to another, and how quickly that information can be processed once it arrives.
As the name suggests, “electronics” compute by moving electrons around, so they’re limited by the physical bare board restrictions of an electron moving through matter. The information can move faster than the electrons themselves, however.
The wiring in electronic computers is already filled with electrons. The signals travel through the wires at the speed of light in metal. When the switches for information processioning are turned on, they regulate the current and act as a gate for electronic signals. So, electrons have to transfer from one side of the transistor to the other.
Then, there’s clock speed, which is usually measured in megahertz (MHz), one million ticks per second, or gigahertz (GHz), one billion ticks per second. These units estimate how fast a computer’s processing is, but it’s limited by the overall length the ticks have to travel and the speed of light and the transistors.
So, a computer’s computing speed can be increased by decreasing its size. The limits on computing speed are restricted by the principles of physics – signals can’t move any faster than the speed of light. Since we can’t make signals faster than the speed of light, we need to make the components smaller.
There is a limit to how small traditional computing parts, such as the NXP i.MX 8 single board computer, can become. Designers have to adjust for quantum tunneling of electrons, but a wire can only be so small. This leads to connected computers to handle the workload, or supercomputers. But in the absence of that, a computer’s power limit depends on how small the components can be.
Aside from the computing power itself, businesses struggle with data processing and analytics. Data is necessary for critical business processes and operations, and with IoT devices, the volumes of data businesses can access today are unprecedented.
This data is changing the way businesses address computing, however. The traditional foundation of a centralized data center and standard internet doesn’t perform at the level necessary to handle the onslaught of data.
As data is transferred from the devices to these storage centers, the network suffers from bandwidth limitations, latency issues, and unpredictable network disruptions that can impact the processing of high volumes of real-world data.
Benefits of Edge Computing
Edge computing addresses the challenges of data collection and processing by handling data on the periphery of the network, close to the source and the end user. Rather than transmitting high volumes of raw data to the centralized core, the processing and analytics take place on the edge. Then, only the insights are sent back to the storage center, giving business the opportunity to gain real-time business insights.
In many industrial IoT applications, IoT devices are in operation in remote or inhospitable environments that aren’t ideal for humans. Edge computing solves data processing problems with IoT devices that operate in real-time from remote locations and inhospitable operating environments.
Benefits and Limitations of Edge Computing vs. Cloud Computing
Both edge computing and cloud computing rely on distributed computing power and data, but they differ in where they’re located.
Edge computing stays close to the data source to process, filter, and analyze data before sending it to the cloud-based network core. The insights gained can be sent to the core for human review and a wider understanding of the whole picture the data illustrates.
Cloud computing is huge, highly scalable, and preferred for IoT deployments. For industrial applications, cloud storage resources are often located across distributed regions. Providers typically package the cloud services for businesses, making it easier to implement for IoT deployments.
These distributed cloud cores can be hundreds of miles from the source of the data, which can cause delays in the transfer and processing of data. Edge computing is decentralized and located near the data source, so it doesn’t suffer from the same delays. That comes with a trade off of higher maintenance, monitoring, and control needs, however.
Computing requires suitable architecture, and not all architectures are ideal for all computing tasks. Edge computing is a complement to centralized storage resources, taking some of the pressure off the network and the core to filter only the most valuable insights.
Edge Computing Use Cases
Retail: Retail businesses gain data from sales, surveillance, and other real-time sources. Edge computing, like the NXP i.MX 6, helps businesses analyze the data that comes from multiple sources and find opportunities to improve operations, such as starting a promotion or optimizing vendor ordering for inventory. Retail environments can change dramatically, so real-time data analytics is necessary to act on data insights.
Transportation: Autonomous vehicles are a source of huge amounts of data, and produce several terabytes per day on the vehicle, the location, the road and traffic conditions, and more. For autonomous vehicles to work well, this data must be analyzed in real time, which requires significant computing power without network delays.
Healthcare: Healthcare is one of the most dramatically changed industries because of data. Medical equipment, devices, and sensors can collect data about patients and allow immediate action. This volume of data requires edge computing, however, especially for automation or AI applications. The computing must ignore the standard data and identify abnormalities to facilitate prompt action.
Network optimization: Edge computing helps optimize network performance by monitoring the conditions for users and relying on analytics to discover the best network path for traffic. Like real-world traffic systems, edge computing can direct and redirect traffic to prioritize time-sensitive information.
Edge Computing for IoT
The rise of IoT brought unprecedented access to high volumes of data, bringing attention to the capabilities of edge computing. This data is useless without analytics, and cloud computing isn’t optimal for processing and analyzing data at the speed modern businesses require.
Jason is the Head of SEM at SolidRun which is a global leading developer of embedded systems and network solutions, focused on a wide range of energy-efficient, powerful and flexible products which help OEMs around the world simplify application development while overcoming deployment challenges.