What is edge computing in IoT?

The growing number of “connected” devices is generating an excessive amount of data, and this will continue as Internet of Things (IoT) technologies and use cases grow in the coming years. According to research firm Gartner, by 2020, there will be as many as 20 billion connected devices generating billions of bytes of data per user. These devices are not just smartphones or laptops, but also connected cars, vending machines, smart wearables, surgical medical robots, and more.

[[242312]]

The large amount of data generated by countless types of such devices needs to be pushed to a centralized cloud for retention (data management), analysis, and decision-making. Then, the analyzed data results are transmitted back to the device. This round trip of data consumes a lot of network infrastructure and cloud infrastructure resources, further increasing latency and bandwidth consumption issues, thus affecting mission-critical IoT use. For example, in self-driving connected cars, a large amount of data is generated every hour; the data must be uploaded to the cloud, analyzed, and instructions sent back to the car. Low latency or resource congestion may delay the response to the car, which may cause traffic accidents in serious cases.

IoT Edge Computing

This is where edge computing comes in. Edge computing architecture can be used to optimize cloud computing systems so that data processing and analysis are performed at the edge of the network, closer to the data source. With this approach, data can be collected and processed near the device itself, rather than sending it to the cloud or data center.

Benefits of edge computing:

• Edge computing can reduce the required network bandwidth between sensors and the central cloud (i.e. lower latency) and reduce the burden on the entire IT infrastructure.
• Data is stored and processed at the edge device without the need for a network connection for cloud computing. This eliminates the need for a high-bandwidth, constant network connection.
• With edge computing, endpoint devices only send the information required by the cloud computing instead of raw data. It helps reduce the cost of connections and redundant resources for cloud infrastructure. This is beneficial when large amounts of data generated by industrial machinery are analyzed at the edge and only filtered data is pushed to the cloud, resulting in significant savings on IT infrastructure.
• Leveraging computing power enables edge devices to behave similarly to cloud-like operations. Applications can execute quickly and establish reliable and highly responsive communications with endpoints.
• Data security and privacy through edge computing: Sensitive data is generated, processed, and saved on edge devices instead of being transmitted over insecure networks and potentially breaching centralized data centers. The edge computing ecosystem can provide common policies (which can be implemented in an automated manner) for each edge to achieve data integrity and privacy.

The advent of edge computing does not replace the need for traditional data centers or cloud computing infrastructure. Instead, it coexists with the cloud as the computing power of the cloud is distributed to endpoints.

Machine Learning at the Network Edge

Machine learning (ML) is a complementary technology to edge computing. In machine learning, the generated data is fed to the ML system to produce an analytical decision model. In IoT and edge computing scenarios, machine learning can be implemented in two ways.

• First approach: ML algorithms require huge computing power to produce decisions in the cloud. Data collected from the edge will be sent to the ML system, where a learning analysis decision model will be generated, and then this model will be pushed to the edge of the network. In this way, analytical decisions can be made on all edge devices. In this model, edge devices will be used to collect, analyze, and take action in the cloud, thereby enhancing intelligence.
• Second approach: If the endpoint device sends sensor-generated data to the ML system in the cloud, the ML system will spend a lot of time transmitting and processing the data to generate analytical decisions. For this purpose, smart machine learning or artificial intelligence (AI) chips can be introduced, and the endpoint device sends the data to the cloud only for storage purposes. Enabling machine learning capabilities at the edge of the network requires less computing power.

Edge computing and the Internet of Things

Edge computing, together with machine learning technology, lays the foundation for the agility of future communications for IoT. The upcoming 5G telecommunication network will provide a more advanced network for IoT use cases. In addition to high-speed and low-latency data transmission, 5G will also provide a telecommunication network based on mobile edge computing (MEC), enabling automatic implementation and deployment of edge services and resources. In this revolution, IoT device manufacturers and software application developers will be more eager to take advantage of edge computing and analytics. We will see more intelligent IoT use cases and an increase in intelligent edge devices.

Original link:

http://www.futuriom.com/articles/news/what-is-edge-computing-for-iot/2018/08