Generally speaking, Edge Computing is the practice of physically executing computing activities near to target computers, rather than in the cloud or on the computer itself. We have seen various design trends for structures over the last few decades. It was planned as a centralized or decentralized device, depending on the device bottleneck. The growing amount of data (IoT) and the networking layer (and computation) limitations are currently leading to a decentralized framework such as Edge Computing. Edge computing is a distributed computing model that brings data storage and computation closer to where it is needed to maximize response times and save bandwidth. The origins of edge computing lie in content delivery networks that were developed in the late 1990s to support edge server web and video content deployed close users.
Such networks developed in the early 2000s to host software and application elements on the edge servers, resulting in the first commercial edge computing platforms hosting software such as dealer locators, shopping carts, real-time data aggregators, and ad insertion engines. Modern edge computing expands this approach significantly through virtualization technology which makes it easier to deploy and run a wider range of applications on the edge servers.
Advantages of Edge Computing
The primary reasons for Edge Computing are as mentioned:
- STRONGER HARDWARE: Most applications in today’s world depend mostly on very heavy, or specialized hardware. For example, modern machine learning algorithms work best with GPUs or Tensor Processing Units (TPUs). It’s usually not necessary to expand systems with such hardware. For such specialist hardware and hardware with more processing power in general, Edge Nodes are preferable.
- BETTER LATENCY: If applications depend on immediate feedback, sending data to the cloud can take too long to measure and return data to the system. Nonetheless, if the path is reduced to and back to the (much closer) edge node, it is possible to realize other use cases.
- SCALABILITY: The processing power of the devices is constrained iin most cases due to their small size. In addition, introducing a new use case needing better hardware would demand that all potential users or the network administrator upgrade the equipment, which reduces the acceptance rate of use cases. Edge nodes do not suffer from these problems, and can quickly and continuously be extended.
- RELIABILITY: If the link to the central cloud is interrupted, the key functionality of devices will still be accessible. It can be accomplished by depending on local contact with an Edge Node which would be less vulnerable to issues. If an Edge Node crash, the devices will be transferred to an alternative Edge Node.
- DATA THROUGHPUT: Appliances can generate significant volumes of data. For example, one single autonomous car might generate up to 4000 gigabytes of data a day. If every single car sends all the data, it generates a huge load on the network all the way to the central data centers. The maximum of the path can be replanted by performing the needed computations on Edge Nodes closer to the computer.
- ADAPTABILITY: Edge Nodes can be easily configured to deliver individual subsets of services, depending on the environment, after creating a base environment. Even if some use cases are only valuable in the cities, others in rural locations may be more valuable.
The Architecture of Edge Computing
The layers are defined as follows:
- CLOUD: Computing capacity and storage on this layer are practically unlimited, but latencies and data transfer cost to this layer can be very high. In an Edge Computing framework, the cloud may be used as deep-term storage, as an urgent lower level coordinator or as an efficient tool for irregular tasks.
- EDGE NODE: These nodes are positioned prior to the actual network’s last mile, often recognized as the “downstream”. Edge nodes are devices that possess high compute power and are also capable of routing network traffic. It ranges from base stations, routers, or switches up to small-scale data centers.
- EDGE GATEWAY: Edge Gateways are comparable to Edge Nodes though less effective. They are capable of speaking most common protocols and are capable of handling computations not needing specialized hardware such as GPU. Devices on this layer are also used to convert devices on lower levels or to act as a lower-level application interface. This means things like cell phones, vehicles, and sensors. E.g. Cameras and motion detectors.
- EDGE DEVICES: You’ll find small devices with very limited resources on this sheet, such as single sensors or embedded systems. Typically, such tools are configured for a specific form of processing, and therefore constrained in their interaction functionality. Devices may be smartwatches, traffic lights, or environmental sensors on this board.
Contact in an Edge Computing environment is not limited to inter-layer traffic but in peer-to-peer fashion may also occur within a layer. This allows for faster communication between devices in different mists which can communicate via their respective Edge Nodes. Through omitting contact over the Internet, it can also further reduce network load and costs.
Edge network services minimize the volumes of data to be transferred, the consequent traffic, and the distance to be covered by the data. That delivers lower latency and lowers transmission costs. Computation offloading for real-time applications, such as facial recognition algorithms, demonstrated substantial improvements in response times as seen in early research. More work has shown that using resource-rich machines called cloudlets near mobile users, providing services usually found in the cloud, offers gains in execution time when some of the tasks are offloaded into the edge node. On the other hand, unloading each task will lead to a slowdown due to the transfer times between system and nodes, so that an optimal configuration can be established based on the workload.
One use of the architecture is cloud gaming, where certain aspects of a game could run in the cloud, while the rendered video is transmitted to lightweight clients like smartphones, VR glasses, etc. This streaming form is also known as streaming by pixels.
Many popular technologies include wired, self-contained vehicles, smart cities, Industry 4.0 (smart industry), and home automation.
Use case Classes for Edge Computing
- CONTENT SCALING: Scaling or converting content to another medium to save network bandwidth. Consider a smart home with several surveillance cameras uploading their pictures to a server. Such cameras cannot be standard and have different formats/resolutions. Having one central instance that transforms all these streams into standard formats and resolutions is also useful so as not to waste any time on the server or the devices.
- LOCAL CONNECTIVITY: Locally connect the devices in a manner that is independent of external (often internet) links. Most home automation systems use a form of local contact independent of the connection to the internet. For example, the Philips hue ecosystem relies on ZigBee. The local networking feature can be used in these situations to allow users to monitor their home automation devices directly without depending on it.
- EDGE CONTENT DELIVERY: Deliver content, similar to a conventional CDN, over the Edge network. Imagine a huge, central-network apartment building. In such a program, content planned to be used by many residents (e.g. the new episode of Game of Thrones on its release day) may be loaded in advance to the central network, minimizing the demand on the network at night as the data can be loaded.
- AGGREGATION: Using, and combine, data from different sources of data placed on the Edge in every type. Typical data sensors in a smart home scenario include weather sensors, window and door touch sensors, plant moisture sensors as well as intelligent thermostats. The core unit will deduce valuable recommendations for the user, use the data from all these sensors mixed.
Such use case classes are not disjointed, and there are several instances that can fairly be put in case of classes with multiple uses.
Significance of Edge Computing
Edge computing is important in all working fields. As the data is closer to the storage it is being used in many companies. It can capture and analyze the company data on-premise to provide better business intelligence without compromising its security. It also saves costs from all sources because maintenance and managing of the edge applications are easier than other applications. In the upcoming time, the world would have better and increased computation power than the present one. It will create an opportunity to edge applications to boost the productivity of the companies by providing autonomous systems.