NCS is building out artificial intelligence (AI) capabilities to maximise the potential of the Internet of Things (IoT) in Singapore, as businesses move from trial to deployment phases.
The technology provider aims to capitalise on a global market expected to reach $745 billion in 2019 – representing an increase of 15.4 per cent in 2018.
Today, more than seven billion IoT devices are in use with the figure expected to rise exponentially within the space of three years, reaching 29 billion by 2022 – of which 18 billion will be specifically related to IoT.
Such growth has been driven at business level as organisations move away from pilot projects into sizeable deployments, whether that be “monitoring pollution levels to the creation of smart carparks”.
According to Dr Tan Kar Han – head of product R&D at NCS – IoT applications have “solved numerous issues everywhere”, while also helping to streamline operations, collect data and understand customer behaviour.
“While the first wave of IoT brought connected sensors that digitalised the physical world and enabled the automation of mundane tasks, a key problem was that many existing solutions did not work well together,” said Kar Han.
Findings from Gartner predict that by 2022, more than 80 per cent of enterprise IoT projects will include an AI component, up from 10 per cent in 2018.
Designed to allow on-boarding of a large variety of IoT devices, Kar Han said such platforms deploy AI to “consolidate and analyse” the data collected, thereby helping to unlock the “full potential of new and existing IoT solutions”.
In spearheading research and development at NCS – a subsidiary of Singtel – Kar Han is tasked with overseeing a division building “world-changing products”, housing a SCALE@NTU joint lab, customer-focused product management and a full-stack product engineering team.
Kar Han cited an example of AI-IoT convergence through modern smart buildings deploying both technologies for advanced climate control.
“Conventional systems currently in use are reactive: simple thermostats would sense temperature fluctuations and proceed to switch the air conditioning on or off,” Kar Han explained. “However, there will be a slight delay with thermostats due to reaction time.
“For example, building cooling systems can be slow to react to sudden changes in conditions. If there is a sudden influx of occupants, temperatures can rise quickly and become uncomfortable before thermostats can pick up the change.”
On the flip side, Kar Han said a modern climate control system with both IoT and AI capabilities would be able to integrate people counters to instantly detect the increase of occupants.
“Taking into account complex patterns in air flow and occupant movement, the system can then pre-emptively activate the chillers at the right power to cool specific locations within the building,” he said.
Another area to benefit from AI-enabled IoT is in the realm of smart manufacturing in connected factories, Kar Han added.
“Many manufacturing processes are already adopting AI to tune numerous parameters,” he added. “The use of AI optimises the production line for customised targets, allowing factories to be more versatile and flexible. The increased efficiency results in lower inventories and waste, thereby reducing costs.”
Despite market maturity, Kar-Han acknowledged that implementing AI solutions with IoT at an enterprise-level can be challenging, as customers continue to struggle with balancing strategies with realities.
“Many companies have data science teams capable of designing and building AI models,” he said. “However, it is beyond their domain expertise to deploy these AI capabilities and integrate them into production.”
Furthermore, Kar Han said some companies have invested in IoT infrastructure and built applications to take advantage of IoT sensing, but these applications may not have the ability to deploy the latest AI models.
“These problems can be addressed with versatile and easy-to-use IoT integration platforms such as NCS’ unifAI,” Kar Han added. “The one-stop IoT portal makes it easy for data scientists to build AIpowered IoT applications, allowing them to directly access live IoT data streams for data collection and deploy AI models into production.”