Edge computing works on the principle of decentralization of processing load from the cloud. It brings computational activities closer to the source to reduce latency and bandwidth issues. It achieves this by moving computational activities from the cloud to local devices like the user's computer, mobile phone, or IoT devices. Data from sensors can be locally processed to monitor the condition of machinery, and efficiency of operations.
The cloud is usually where all the magic happens. Data gathered from various devices and sensors are sent to the cloud for processing. Given the remoteness of most cloud servers from the devices and the amount of data transmitted, this is always a possibility of bandwidth and latency hiccups. This can impact the real-time and agile systems.
In an IoT enabled or a connected environment, there are a lot of last mile devices that have computing power like edge servers, computers, and smart phones. An edge computing framework aims to share and alleviate the computational load of the cloud by using the power of these local devices. This can improve response times, balance the load on communication bandwidth, and widen the scope of application that was earlier constrained by lead times.
The network edge is the location of the device or local network containing the device that generates or uses the data.
Network edge defines the set of all applications and equipment that convert data to value for the user. It is also known as WAN edge. It consists of two groups of entities – devices and infrastructure. Devices could be machines, sensors, or computers, smartphones that generates or uses the data to take last mile actions. Infrastructure enables the devices to communicate and compute as required. They could be smartphones, modems, or ISPs amongst others. Edge computing uses the network edge to lighten the load on the cloud.
Edge Computing has been finding applications in diverse industries. But two of the fastest adopters have been automotive and industrial manufacturing.
A modern car is supposed to do more last mile computing than a fighter jet. When you add vehicle autonomy, the need for instantaneous response based on external stimuli increases manifold. ADAS (Advanced Driver Assistance Systems) gathers inputs from both outside the vehicle and inside the vehicle to proactively assist the driver. Edge Computing can enable the car to do a lot of these actions without depending on bandwidth, connectivity, or latency of a pureplay cloud system.
Manufacturing environments generate volumes of data every second. While some of this information can be processes in the cloud, there are stimuli that needs one-the-fly analysis and response. This helps automate actions that drive efficiency and operator safety. For instance, operator fatigue alerts or shaft deviations in a rotating equipment will need almost instantaneous actions. Edge computing can use the network edge computational bandwidth to predict and respond.
The primary benefits of edge computing are that it helps optimize communication bandwidth and cloud server utilization. Since both bandwidth and cloud involve cost of subscription, edge computing can help reduce these costs considerably.
Edge computing will also significantly reduce latency by decentralizing processes to the edge. This will translate to outcomes like agility in response.
When edge computing is combined with IoT, machine learning and artificial intelligence, it can evolve into what enthusiasts call as “Internet of Conscious Things”. This can help bring a whole new level of assisted environments empowering the manpower with the right recommendations and actions.
Statista predicts that by 2025 there will be over 75 billion IoT devices installed worldwide. With such exponential growth of IoT enabled devices, pricing policies of bandwidth, and cloud featured amenities may get competitive. Edge computing is likely to play a pivotal role in balancing cost efficiency.
Neither edge computing nor cloud computing are going to replace one another. They will always co-exist to deliver maximum value. The appropriate delegation of computational load between cloud, edge and client computing devices will be the key to success.
