Over the years, 5G and Artificial Intelligence (AI) have both proven to be transformative technologies in their own ways. Both these technologies have one thing in common – both need to deal with massive amounts of data. We are on the cusp of an era where they will complement and enhance each other and bring the power of Machine Learning to our -devices, enabling richer, smoother, and more personalized experiences than ever before.
Most of the initial implementations of AI/ML applications involved offline machine learning with large amounts of historical data and the inferencing engine resided in the cloud. As end devices became more powerful, some of the inferencing schemas moved to end devices. That helped with quicker decision making.
Today there are billions of connected devices in the world, and the number is increasing rapidly. The volume of the data generated by these devices is so enormous that it is difficult to transmit it to the Cloud for processing and managing in real-time. With 5G, new use cases are being imagined that will push the data growth exponentially higher. As such, a more scalable approach is needed to balance the resources as well as the latency and accuracy of decision making. A decentralized model can help overcome this challenge as well as potentially address privacy and security concerns. In a decentralized approach, intelligence is not just limited to the Cloud but is distributed along the cloud edge and, more importantly, the end devices. A combination of edge cloud AI and on-device intelligence can help with many new experiences with 5G. Coincidentally, for 5G we are looking at disaggregated networks with number of network elements deployed in data centers at the cloud edge. This architecture can also be leveraged for distributing the cloud AI from centrally located data centers to to the cloud edge. This helps in localizing lot of traffic to appropriate set of servers and make it manageable to handle load while keeping latencies reasonable.
While this is happening, the capabilities of connected devices such as smartphones, cars, sensors, and other connected devices are increasing many folds. And they now have in-built AI capabilities that enable them to perceive, reason, and act on their own, while transmitting only the most relevant data to the edge-cloud/Cloud. This shift to on-device intelligence can yield significant benefits in many socioeconomic spheres of life and a wide range of human activities. It can enable more personalized virtual assistants, enhanced photography, better security, connected healthcare, and robot-guided industrial operations – to name but a few of the countless possible applications.
Let us take the example of road safety. Automobiles – whether of the autonomous kind or equipped with advanced driver assistance systems sensors – will be able to communicate directly with each other and even warn the driver of approaching vehicles or pedestrians around a blind corner. Processing data close to the source, using on-device AI, is highly beneficial in such cases as it enables the car and its driver to react in real time, as opposed to relying on the Cloud. While a centralized Cloud will continue to be important for Big Data training, the application of AI will increasingly shift to power-efficient devices at the Wireless Edge.
By applying AI to both 5G networks and devices, we can have more efficient wireless communication, longer battery life, and better user experiences. The capabilities of AI can prove very effective in addressing issues in areas such as service quality, deployment, network efficiency, anomaly detection, and network security, which can be difficult to solve by conventional methods. AI can also improve the 5G end-to-end system. Radio awareness is at the heart of how AI enhances 5G. Increased radio awareness improves device experience, system performance, and radio security. AI and Machine Learning arethe perfect toolsfor making sense out of the complex RF signals around devices.
Meanwhile, the low latency and high capacity of 5G allows AI processing to be distributed across the device, the Edge Cloud, and the Central Cloud – which opens the doors for a variety of newer, richer experiences. The Wireless Edge architecture is adaptable and allows performance and economic tradeoffs to determine the best distribution of workloads for achieving the required latency or the compute requirement for any application. The impact of 5G in AI-enhanced applications will be felt in not only industrial operations but also in everyday experiences. The retail shopping experience, for instance, can be completely personalized for each customer with the help of boundless extended reality (XR), virtual assistants, and vastly improved voice interfaces. Customers, when they window-shop, will see everything that is of interest to them, tailored to their preferences and likings and even virtually try it on.
On-device AI and on-device Machine Learning will maximize personalization, ensure greater privacy
To truly maximize the possibilities of the 5G-AI combination, we need to think beyond Cloud-centric AI. Today, we already have partially distributed AI, which allows devices to refine data before passingit on to the Cloud for aggregated analysis. As the next step, on-device AI will go beyond merely inference and enable Machine Learning on the device itself.
This holds the promise of some very important benefits for users, which, in turn, will help in driving wider adoption of AI. The first is that of scale; spreading processing over millions of devices will enable us to harness huge amounts of computational power. Secondly, as the user’s own data is used for their device training, the AI model learning will be customized to each user. And thirdly, the use of on-device data for on-device training will ensure privacy while extracting the value of the data.
We are in for exciting times in the years ahead. The convergence of powerful technologies that have evolved parallelly till now will provide a base for technology innovations and applications that will leverage the best of both worlds and make the one we live in safer, easier and more personal.
By Sachin Kalantri – Senior Director, Product Marketing, Qualcomm India Pvt. Ltd
Solana based venture platform SolRazr has unveiled the first decentralized developer ecosystem for the Solana blockchain, comprising…
Delivering fresh insights into Indian enterprises’ ability to create business value from data analytics, Alteryx, Inc,…
Shoppers Stop, a retailer of fashion and beauty brands, is collaborating with Accenture to accelerate…
Your email address will not be published. Required fields are marked *
Asha Jadeja Motwani is a multi hyphenate. She is the founder of…
Equinix announced last week that it has extended the Platform Equinix into…
Are Students Who Have Graduated Online Able to Secure Jobs With IT Companies?