Edge Intelligence lies at the confluence of multiple technologies, particularly Artificial Intelligence, Machine Learning, Data Analytics, Internet of Things, and Cloud
Picture a scenario. On your way home, you see a fleet of delivery trucks driving towardstoward a popular supermarket in your neighborhood. You notice that except for the first vehicle the others are driverless. Yet, each is closely followed by the next and every move is seamlessly imitated across the entire fleet. The next day, you visit the supermarket to buy groceries. You swipe in with your smartphone and proceed to the aisles. You pick up everything you need and head straight for the exit – there is no checkout counter. On leaving the building, a notification on your phone informs you that the payment for your groceries has been debited from your bank account.
This episode could be an everyday experience in a few years. Thanks to recent technological developments, business operations, and customer experiences are more frictionless than ever. Today, driverless cars are in the works, and online purchases can be made with a voice command. The intelligent edge is set to drive such advances up a notch.
Pushing the digital envelope
Edge Intelligence lies at the confluence of multiple technologies, particularly Artificial Intelligence, Machine Learning, Data Analytics, the Internet of Things, and the Cloud. Simply put, it involves collecting, analyzing, and drawing insights from data close to the source. In the above scenario, the autonomous trucks could capture data on road conditions and vehicle speeds through sensors and relay it to the rest of the fleet. Similarly, data on supermarket purchases could be captured by sensors near the aisles. In both cases, real-time insights are generated by analyzing data through AI and ML near the source, or the ‘edge’.
Traditionally, data collected at the source is sent to a centralized server or cloud, where it is processed and relayed to the end-user. This presents bandwidth and latency issues in situations where real-time data is crucial. Delays could be costly for an autonomous fleet on a collision course or a drone monitoring a hazardous gas leak at a power plant.
Edge intelligence will be critical as the number of Internet-enabled devices continues to grow exponentially. By 2030, 125 billion devices are expected to be connected through IoT, pushing the need for more location-driven, decentralized, and distributed services. However, as a Deloitte report points out, edge intelligence will not replace the cloud or data centers. Rather, it will enable a more holistic cloud-to-edge architecture. Some parts will run at a data center, others at a centralized cloud, and many more on the edges.
What lies at the edge?
The world is becoming increasingly data-driven, fueled by rapid advances in next-gen technology. Customer expectations continue to grow in tandem. Bringing powerful computing capabilities closer to the origin of data can enhance operational efficiency and address new challenges in several ways. In less than a decade, 125 billion connected devices could generate an unprecedented amount of data. Transferring such huge volumes to data centers across geographies or a centralized cloud will cost enormous time and money. The intelligent edge allows near-instant responses at significantly lower costs through ultra-low latency and more efficient bandwidth use.
Consider its impact in different industries. Automobile plants could see machines seamlessly communicating with each other to build vehicles. Manufacturers can pre-empt technical failures and safety hazards right on the shop floor through predictive maintenance, enabled by IoT sensors closely monitoring machine conditions. In agriculture, autonomous tractors and robots can communicate with nearby sensors to access environment data. Systems can analyze weather patterns and alert farmers in the event of possible natural disasters. In entertainment, the delivery of content - be it music, video, or websites – can be made quicker and more customized by caching data at the edge.
The intelligent edge is a gamechanger. That said, it has certain disadvantages. Increased hardware costs are a notable case in point. Edge intelligence heavily relies on sophisticated hardware to capture and analyze data close to the source. This is not the case with a purely Cloud-based architecture, where local hardware is more basic. It also presents security and privacy issues. Not every device has the same degree of protection. Further, the greater number of storage points makes it more difficult to monitor. On the other hand, hardware prices are reducing, and data distribution over multiple edges can make it more secure than storing it in a distant data center. Therefore, businesses must weigh the pros and cons accordingly.
As adoption grows, edge intelligence could enable a wave of cutting-edge innovation that will define Web 3.0 and Industry 4.0. According to Gartner, 75% of enterprise-generated data is expected to be processed at the edge by 2025. The age of autonomous fleets and retail stores does not appear to be too far.
The author is President and Global CIO, HGS