Not cloud computing but edge computing is the way forward.
We are moving towards an interconnected world where billions of devices would offer uninterrupted services to people and industries with the help of Internet-of-Things (IoT) and communication advancements like 5G. They will generate mammoth size data at the network edge, however, cloud computing and on-device computing could be blockers to these technologies from reaching their full potential. They have high latency and limited computation capacity and cannot serve the need for instant data processing and analysis of the evolving space.
So what’s the solution? Edge computing.
What is Edge Computing?
Cloud computing is where most of the computing action happens these days. Here the IoT devices collect and deliver data to a remote cloud network that performs computation-intensive tasks and distribute the results back. The problem with this is that these cloud computing servers are usually stored at a long physical distance causing latency issues. To rely on such a network for applications like autonomous driving or medical surveillance becomes an issue.
To add to that, putting all the pressure on one central network increases the overhead cost required to keep it robust.
Edge computing is proving itself to be a promising alternative for all these shortcomings of cloud computing. It deploys computation capacity close to the data source for instant data processing rather than wasting time transmitting data. In this, a huge number of servers are deployed at the edge of the network so that the tasks at IoT end devices can be offloaded to the edge servers for instant processing. Edge computing can also pair up with deep learning technologies and process data at lightning speed.
Besides low latency, it also helps in better protection of data and application security while reducing the pressure on the central network.
Edge Computing with Deep Learning
Deep learning is one of the breakthrough technologies of recent times. It is being widely used in
computer vision, natural language processing, healthcare, supply chain management, and more. Combining the benefits of edge computing with the perception ability of deep learning algorithms will further expand the opportunities. It’s a perfect match!
Applications of edge computing can benefit from the processing capabilities of deep learning to handle complex situations such as personalized medical treatments, automating claims and damage analysis in insurance, and more. While edge computing can support deep learning with its specifically designed hardware foundations, e.g., the light-weighted Nvidia Jetson TX2 developer kit.
Some deep learning models that are already being used for edge computing applications include Restricted Boltzmann machine (RBM), AutoEncoder (AE), Deep Neural Network (DNN), Convolutional Neural Network (CNN), Recurrent Neural Network (RNN), and Deep Reinforcement Learning (DRL).
A few examples of how deep learning-powered with edge computing can work could be –
-
- Smart Multimedia – For instance, because of the limited processing capabilities of cameras, traditional video analytics usually heavily rely on cloud computing for content processing which is unable to satisfy latency-sensitive applications like camera-based surveillance. Edge computing pushes the video analytics from the remote cloud to the local edge, allowing the video content to be processed near the data source to enable quick or even real-time response.
Amazon also released the world’s first deep-learning-enabled video camera, where the locally executed deep learning function enables real-time object identification without the involvement of the cloud.
- Smart Transportation– Vehicle information can now be accessed anytime anywhere with the help of 5G and mobile edge computing which uses cellular networks. It will enable intelligent transportation management like autonomous driving, traffic prediction, traffic signal control, etc.
- Smart City– It is yet another important application for deep-learning-enabled edge computing. Geographically distributed big data requires a distributed computing model for local processing and management. The integration of edge computing and deep learning enables the deep penetration of computing intelligence into every corner of a city, forming a smart city that provides more efficient, economic, energy-saving, and convenient services.
- Smart Multimedia – For instance, because of the limited processing capabilities of cameras, traditional video analytics usually heavily rely on cloud computing for content processing which is unable to satisfy latency-sensitive applications like camera-based surveillance. Edge computing pushes the video analytics from the remote cloud to the local edge, allowing the video content to be processed near the data source to enable quick or even real-time response.
Edge Computing for IoT
Edge computing helps move intelligence to the edge in IoT applications. However, a different variation of edge computing called Mobile Edge Computing (MEC) is used in IoT which uses cellular networks for primary connectivity to reduce latency and offer massive bandwidth for applications to scale.
By bringing cloud computing capabilities to remote locations, MEC can offer local processing and storage and make IoT data actionable at scale. This together with 5G can further improve the speed at which data processing is done, making it happen nearly in real-time.
A Future with Edge Computing
A review by Gartner says that in 2018, just 10% of all data was processed at the edge. However, it expects the number to bump up to 75% by 2025.
This shift from cloud to edge computing is going to be massive and will only be possible with increasingly powerful hardware and smart AI systems that can process information, communicate across networks and make decisions locally at lightning speed.
The opportunities for developing and deploying high-speed, low-latency applications that require data transfer in fractions of a second are abundant and we have just started to realize it.