IoT, edge computing, and AI: Experts discuss the intersection of pathbreaking technologies

Businesses face a multitude of challenges in a world of dynamic big data, where volume, quality, and speed can change by the second. Systems are often overwhelmed with large volumes and velocity of data, struggling to process them in real time, unable to keep up with the customer demands. 

Fast growing connected world, the Internet of Things (IoT) is growing to be critical to connect devices, systems, and applications to share data in real time. Ranging from simple smart home devices to complex industry machinery, IoT devices have a range of applications that can power everything from smart homes to smart cities. Moreover, the applications of IoT span a wide range of industries including manufacturing, agriculture, healthcare and transportation. A McKinsey report titled ‘What is the Internet of Things’ estimates that the total potential value for the IoT ecosystem could reach $12.6 trillion by 2030. 

However, as the use of IoT becomes more prolific, it’s worth looking at the challenges that dog this promising technology, including latency, security, scalability, and more. 

Volt Active Data recently brought together experts from a variety of fields for a closed roundtable discussion on ‘Internet of Things (IoT) for the digitally connected world: Understanding how to maximize value out of IOT’. Volt, a real-time data platform, helps businesses across the world process billions of transactions per day, while offering low latency, consistency, and availability at scale. 

Participants at the roundtable brought fresh insights and perspectives on IoT, edge computing, artificial intelligence integration (AI) from a range of different industries. 

Joining the moderator Rishabh Mansur, Head - Community, YourStory Media, were Jayraj Vyas, CTO, Ola; Ranjitha R, Director of Engineering, Myntra; Anurag Nandwana, Manager - Digital Twins, Incedo Inc.; Aayush Agrawal, Co-founder and CTO, WorkOnGrid; Sachin Gupta, Chief Information Officer, Amagi Corporation; Shobhit Mandloi, Director of Engineering, Wyb Social; Shravan A, Head of Technology, Bytebeam; Abhilash PK, Chief Product Officer, Gloify; Balamurali, Head Cloud and DevOps CoE, Tata Elxsi; Suresh Gokarakonda, CTO, Dataken; Kashi KS, Chief AI Officer, Alfahive; Sainadh Duvvuru, Co-founder and Chief Business Officer, HappyLocate; Santosh Hedge, Technical Head, WOW Skin Science; Rohan Babu A M, Founder and CTO, Renderpub; Sanjay Madhva, Director Engineering, Happiest Minds; Sarath Kummamuru, Senior Vice President of Engineering, Razorpay; and Vyom Rawat, Director of Technology, Blubirch.

The discussion was supported by the expertise of Fahad Khan, Sales Director - APAC, Volt Active Data; and Biplab Banerjee, Principal Solution Architect, Volt Active Data.

Speakers shared their unique use cases, challenges, and solutions to manage the vast volumes of data generated every day. They dove deep into the data challenges they were facing. Of particular concern was the management of real-time data - keeping it secure, managing petabytes of clickstream data, providing real-time analytics and data latency.

Speakers also addressed the dichotomy of data across various industries - where the data collection of some industries (finance, ecommerce, etc.) is high, while the data captured in certain legacy industries is extremely low. They revealed that many businesses are unable to understand their data, or struggle with establishing robust data governance strategies in their organisation. 

The challenge of network latency

As a general rule of thumb, the lower the network latency, the higher the speed and performance of the network. Latency can affect the performance of IoT devices, leading to a delay in data transmission and declining operational efficiency. In real-time applications, these delays are critical.

Many panellists discussed their concerns over data latency. Live streaming events often have to deal with latency, impacting the customer experience. In the gaming industry, companies have to determine the network speed of the user, the user’s device, and the load that the GPU and cloud can bear. Based on these factors, they then have to understand how they can reduce latency and offer a seamless and uninterrupted experience to the user.

Participants also pointed out the inevitability of network latency, stating that it was a problem for hardware vendors to resolve. With the large volumes of data generated today, latency becomes an inevitability. Some companies opt for data lakes, cloud native tools and internal teams to manage data. However, there are decisions that can be made to optimise the user experience - whether it is 10,000 users or a million. Pannellists pointed to the launch of the Google chip Willow - a state-of-the-art, quantum computing chip that significantly reduces errors as it scales up, concluding that the world is slowly moving towards solutions that will help improve latency and performance.

The rising tide of return to order in ecommerce

One of the issues discussed by panellists was the issue of fraudulent returns, particularly when it comes to fashion. For instance, if a customer were to order a red shirt, and then return it a blue one, how would the e-commerce company know until it was returned to the warehouse for checking?

Panellists discussed how they can stop the problem at the doorstep, at the time of delivery by potentially using edge devices, IoT or blockchain to combat the problem. Another solution discussed were tag loops with barcodes to identify the right product. While companies such as Razorpay offer insurance to D2C companies for returns, it is restricted to certain regions. Panellists also put forth suggestions such as sharing data so companies can assess customers on the basis of their CIBIL scores to determine if they are high risk. Ultimately, there needs to be a centralised hub of data concerning customers, that empowers businesses to take split second decisions on whether they should allow a certain purchase or transaction for a product. Panellists speculated that in the future, once devices such as washing machines, fridges and other devices and appliances are connected and enabled with AI, fraud could potentially be higher. Or it could birth new and innovative ways to combat fraud.

Is edge computing an inevitability in India?

Edge computing and IoT have powerful reasons to intersect with each other. IoT devices benefit from having compute power close to physical devices and data sources. Data analysis at the edge allows IoT devices to react faster. Together, the two technologies can reduce latency, improve network bandwidth, continue operations when a network connection is lost, and carry out localised data processing via analytics algorithms and machine learning (ML).

The discussion focused on edge computing, with participants sharing that though edge had fizzled out when it was initially introduced (having been overshadowed by cloud computing), it could be beneficial in India. In a densely populated country, edge could be a gamechanger. Unlike Europe, where the population is more spread out, India’s cities are tightly packed, with a variety of data generated from every neighbourhood. This is where edge computing can make a difference.

Speakers cited its potential efficacy in quick commerce, where localised insights and swift decision-making is critical. Operations could take a hit if quick commerce companies were to depend on the cloud. However, if companies were to equip dark stores with edge computing and streaming and complex processing capabilities, it could offer an even faster and more efficient shopping experience for customers.

The dilemma of data sharing 

The sheer volume of data generated by IoT devices is staggering. However, a bigger challenge arises when it comes to storing and sharing this data for real-time monitoring and analytics. While real-time sharing with third-party analytics tools and other systems is crucial to uncovering unique insights, organisations are often hesitant to do so, due to security concerns, latency and the competitive advantage of keeping valuable data for themselves.

Panellists discussed data sharing when it comes to smart cities. The amount of data generated by a small airport, for example, is extremely heavy to transmit to the cloud in order to carry out analytics. In this case, speakers concluded that a mix of edge computing, IoT devices AI, and a centralised hub for analytics would help when it comes to data sharing.

Encountering AI at the edge

The discussion delved into the integration of artificial intelligence with edge. Speakers cited how in healthcare, a combination of edge and AI could help ramp up decision making and effective diagnoses. They discussed how AI and edge computing could assist doctors with MRIs, accelerating the processing of medical images and swiftly spotting anomalies in different parts of the video. Quicker diagnoses can lead to greater decision making, improved patient outcomes and better healthcare.

The panel also discussed the development of targeted small language models (SLMs), with reduced memory requirements, that could enhance analysis at the edge. 

At the end, the panellists examined the capabilities of Volt Active Data. With increased connectivity of devices, they shared that the value of real-time data does not come from storing it and processing it, but from processing it live. Volt’s Active Data platform enables this convergence of data, guaranteeing data accuracy, scalability through distributed data partitioning and synchronous replication. Real-time analytics and machine learning integrations allow businesses to handle high frequency transactions.