In edge computing, computing capabilities are decentralized and performed closer to the “edge” of the network, where data from sensors, devices and equipment originate

The Internet of Things or IoT has been called the next technology revolution after the Internet – the dawn of a new era promising exciting possibilities with wearables, smart homes, autonomous cars, smart factories, smart cities and literally anything with “smart” on it.

The term the “Internet of Things” implies the interconnection of devices — “things” — to the internet and to each other. The adoption in IoT adoption is opportune due to a combination of underlying trends: cheap and widely available bandwidth; developments in semiconductors that has resulted cheaper, less energy consuming microprocessors; developments in big data analytics tools. IoT adoption will be enabled by two developing technological trends – edge computing and machine learning.

The Rise of Edge Computing

The proliferation of connected devices, sensors, gateways – IoT endpoints – is under way. The IT industry analyst, Gartner, estimates a rapid proliferation of IoT endpoints at a rate of 30%, potentially reaching 20 billion by 2020. The implosion of connected devices brings about new challenges. Massive amounts of data will be generated from all these endpoints. Transporting all these data directly to the cloud for data management, analysis and decision-making would not only be costly and inefficient but could also choke the network infrastructure, potentially causing latency issues.

Also read: This IoT solution can predict power outages in your building and also help you save 30% energy

For that reason, edge analytics and edge computing will become increasingly important in large-scale IoT deployments. In edge computing, computing capabilities are decentralized and performed closer to the “edge” of the network, where data from sensors, devices and equipment originate. Processing or analyzing data at the edge, i.e., at the device or gateway level, instead of in a centralized server or the cloud would be advantageous for multiple reasons.

First, it would enable real-time applications. Second, it would be less burdensome on the infrastructure as only the necessary data needs to be sent to the cloud for further processing and storage.

This does not imply the displacement of traditional data centers or cloud computing as we know it. Instead, it is more likely that edge computing would co-exist with cloud computing as computing workload is distributed to where it makes most sense. Sensor data will be collected and processed on edge gateways where edge analytics can be applied using rule-based algorithms and where action can be implemented in real-time. Then the filtered data could be sent to the cloud for enrichment, aggregation with data from other sources such as CRM databases, and then fed into analytics engines in the cloud to generate models that could be send back to the edge for edge analytics.

Machine Learning

Another key IoT enabling technology complementing the rise of edge computing is machine learning. Often used interchangeably with Artificial Intelligence (AI) although not the same, machine learning in computing terminology refers to the method of generating automated analytical models that do not need to be explicitly programmed by humans. While the knowledge to enable machine learning were available decades before, it wasn’t until recently that computing power became powerful enough to perform machine learning.

As computing power becomes better and cheaper, machine learning would increasingly be incorporated in IT architectures not only in the cloud but at the edge as well.. Machine learning would not only power analytics in the centralized cloud but also greatly increase the efficiency of edge analytics.

Edge computing and machine learning will play key roles in future IoT architectures to handle the implosion of data as adoption of IoT grows. The big three cloud providers — Amazon Web Services (AWS), Microsoft and Google — are incorporating these technologies into their offerings. All three already offer machine learning-as-a-service on their respective cloud platforms. Perhaps realizing the need for a comprehensive strategy, Amazon Web Services (AWS) and Microsoft have each launched their own edge computing software that can run on system-on-a-chips (SOCs) or at an edge gateway despite potential cannibalization to its core cloud services.

As with all new technologies, only time will tell how these developments will play out, but one thing is for certain, the age of the Internet of Things has arrived, and we should embrace ourselves for the future it will bring.

—-

Featured Image Credits: 123RF Stock Photography

The post What role does edge computing and machine learning have in IoT? appeared first on e27.