Ad block

The advantages of edge computing

Power of edge computing

With microcontrollers and SoCs becoming more powerful and portable, the demands by customers from even the simplest IoT products is quickly increasing in scale. The burden of this increased analysis and data collection is largely handled by data centers. There is a new solution, referred to as edge computing, that might soon provide some relief.

It is estimated that there will be 21 billion IoT devices connected to the internet by 2020. While this has led to many technological improvements, it has also resulted in the need for large data centers and huge amounts of processing power. This increased data exchange has opened the industry to unexpected situations such as cyber-attacks. Traditional cloud computing relies on collecting, storing and managing large amounts of data on remote servers in data centers. While cloud computing may seem to have infinite expansion potential, the fact is that remote services create increased system latency. In other words, a device may take longer to execute a task because of all the information handshaking with the data center servers. This is a downside in a world where fractions of a second count.

One concern that is not often addressed is the internet bandwidth needed for billions of devices, not to mention the potential strain that the IoT industry is expected to put on network providers. If IoT devices continue to utilize traditional cloud computing, network providers will have to improve communication lines to account for the additional data usage. This, in turn, will lead to increased internet costs and potentially bandwidth throttling.

To mitigate this unfavorable internet dilemma, what data does an IoT device really need to send to the cloud? If we look at an IoT sensor, such as a temperature sensor or door sensor, the device sends and receives information about the status of the current temperature or if the door is locked. This requires little processing or internet effort. What if the device was more intricate and had advanced features such as voice and speech recognition? This introduces a new level of complexity when it comes to information processing across the internet.

If we take a closer look at the voice and speech recognition example, this is like the devices many of us have in our homes. The traditional cloud approach requires a complex speech-recognition program and often requires the use of deep learning and neural networks living in a decentralized data center. The continuous stream of voice and speech sensor data is transferred via the internet and offloaded to a complex AI algorithm for processing. Now imagine this happening for billions of similar devices at the same time. With the microcontrollers becoming advanced and having increased processing speeds, what if we started building IoT devices that did some of their own processing?

Enter Edge computing

The key distinction between edge computing and cloud computing is where and how data is stored and processed. In cloud computing, IoT devices capture data and the entire data set is submitted for further processing to the cloud. In edge computing, information is gathered and partially, or entirely, processed locally on the IoT device. Then, only pertinent processed information is shared with the data center as necessary. Incorporating edge computing into your network can bring benefits to a cloud network by addressing the following:

  • Latency—Local data processing on the device can create a more responsive applications by not relying heavily on the cloud.
  • Security—Processing data locally makes it harder for attackers to access data by isolating data storage.
  • Connectivity Costs and Bandwidth—Reduced data stream loads saves on network and computing resources. Therefore, less congestion on the network and less energy demands for servers.
  • Business Uptime—When computations are performed at the edge, it creates a welcomed redundancy. Disruption can be isolated to single points on the network versus searching for the problem on the entire system.

The implementation of edge computing as an emerging solution could be favourable to many industries such as manufacturing, healthcare, smart buildings, and transportation. Here are a few technology trends that will impact the adoption of edge computing in these and other industries.

Trend 1: Localized computing power

Microcontrollers have seen incredible improvements in performance in the last decade. The rise of the ARM multiple core-based architectures has provided steep increases to processing power. Shrinking component sizes are reducing overall package dimensions. These two advancements make a perfect pairing for design-in opportunities for IoT devices. It is common for the simplest commercial microcontrollers to have clock speeds greater than 80 MHz and RAM amounts of 32 KB or more. This type of capability allows for fast and localized data edge computing.

Trend 2: Security concerns

IoT devices often transmit trivial information such as temperature of humidity. Designers have neglected implementing strong security measures into their devices. This has resulted in millions of IoT devices that are susceptible to malicious attacks. These attacks are largely associated with denial of service. Edge computing does not necessarily improve device security at the local level, but with the reduced online communication with data centers, it makes them difficult to identify. By not transmitting data, which could include audio information, attackers will not be able to perform man-in-the-middle attacks or even pretend to be a data center requesting information. Having much of this sensitive information kept local on the device, an attacker must either physically attack the device or attack the device remotely.

Trend 3: Configuration, implementation and support

The implementation of edge computing into a network is not a simple plug-and-play solution. Edge computing will force changes to current IT architectures and processes. Additionally, managed services will need to be developed to handle system expansion and evolution. This goes for both hardware and software aspects of the network. From gateways to servers to AI software and security, there will be a need for experienced resources to develop these changing systems. Engaging resources that have expertise in all these areas will be invaluable to an organization’s success.

Conclusion

Hardware, component, and software advancements are providing businesses a powerful opportunity to grow and improve their position. Moving towards edge computing is the future of IoT applications. Optimizing data handling and data processing at the local level is the advantage that edge computing offers. Thus, alleviating the drawbacks of the traditional decentralized data center approach. For many businesses the transition to the edge may appear insurmountable.

Interested in learning more?

Arrow Electronics understands this hesitation by businesses and sees the advantages of edge computing and has expert teams and products available to help you create your edge computing future.

For more information on this topic, or to get in touch with engineering specialist who can help answer any questions you might have, head to arrow.com

Ad block