The integration of AI and the connected devices of the IoT - having been anticipated for years to solve social issues such as in the form of smart cities - is now on its way to making inroads in multiple sectors and creating opportunities for genuine digital transformation. Such transformation will be crucial in the shift towards truly smart cities, which has been given further impetus by the Covid-19 pandemic.
But among the technological bottlenecks to full deployment are delays due to connecting to internet servers and the capacity limits of the microchips installed on devices and components.
Very short delays in the response times of microchips, known as latency, can have huge consequences. They can result in inefficiencies in IoT devices, make the use of AI simply impractical in fields such as autonomous vehicles, or even be fatal in the case of disaster prevention applications. For example, a very small delay in an autonomous driving can lead to severe accidents.
One answer to this is
Edge AI, which unlike
cloud-based AI,
doesn’t need to connect to remote servers to be functional. However, with Edge AI, there remains the challenge of having algorithms that can function independently on even the smallest chips without a loss of performance.
Seminal technology from
Tokyo start-up AISing is
one crucial piece in
solving such puzzles.
Its Edge AI Memory Saving Tree (MST) is an extremely fast, lightweight and accurate algorithm with memory requirements low enough that it can be deployed on tiny chips that can fit on the tip of a finger, and which is capable of prediction and learning on the chip itself.
Its ability to run on tiny
chips massively increases
MST’s potential
applications,
and its ultra-accuracy and self-updating capacity allow it to comfortably outperform peers such as the widely-used ‘random forest’ machine learning algorithms, according to AISing founder and CEO IDESAWA Jun-ichi.
“AISing’s advanced Edge AI algorithms are being deployed in more and more fields and industries including manufacturing, construction and drones. Most of the deep learning experts are from IT engineering. I came from a mechanical engineering background, so I know a lot about AI and also have the knowhow to undertake the complex task of applying it to machines. Our company is very good at these three intricately intertwined areas, and no one else in the world is doing this.”
- IDESAWA Jun-ichi
In one Japanese factory,
the deployment of this
one-of-a-kind MST
algorithm reduced
latency
and cut crucial seconds of delay to make the operation smarter, increasing efficiency and saving the company a huge amount of money annually.
The ability of MST to
continue to learn ‘on
the spot’ and update itself
contributes to AISing’s mission, which Idesawa states is, “to make each and every device a little smarter, which in turn will make a super-efficient society, such as smart cities, a reality.”
NEC Corporation (NEC) has been leveraging its ground-breaking smart city tech in projects across Asia and Europe. NISHIOKA Mitsuyo from its Super City Business Promotion division explains the company’s approach: “Our concept for smart cities essentially contributes to revitalizing local economies and improving the quality of life by our city operating system, while preserving the unique strengths, culture and characteristics of each community.”
The cornerstone of NEC’s smart city platform development was also involved in the deployment of FIWARE, a framework of open source platform components that originated in Europe and is designed to accelerate smart solutions. Meanwhile, the Japanese government has created a Smart City Reference Architecture framework to ensure smart cities develop in line with the principles of citizen-centricity, interoperability and sustainability. NEC has been deploying its own City OS platform, with its proprietary tech in biometrics, AI data-analytics, ID management systems and full-layered security, which in conjunction with FIWARE can deliver unique solutions to regional social issues.
When the city of Lisbon, Portugal decided to implement smart city policies, NEC bid for the project utilising its City OS alongside FIWARE, impressing the municipal government by having a trial system up and running in just six hours. NEC and the city worked together on developing and deploying the Lisbon Intelligent Management Platform. NEC was able to leverage its 120 years of experience in public-private partnerships to create optimal solutions for the city’s specific problems.
“Lisbon had a lot of issues, including as a capital city, very heavy traffic,” says TAKAHASHI Shunsuke of NEC’s Super City Business Promotion division, who explains that the integration of complex and public and private datasets enabled more proactive decision making and operational efficiency while ensuring public safety.
In neighbouring Spain, the province of Cordoba faced a range of issues, being made up of 77 municipalities, many with limited resources. But the flexibility of NEC’s City OS meant it was still able to deliver results via intelligent waste, energy and water management, as well as through a host of other measures.
“Pilot services could be tried out in one municipality and then transferred to towns with fewer resources,” says NISHIOKA.
The interoperability of the platform opens up the possibility of collaborations not only between domestic municipalities but also across national borders. Lessons learned in smart city technology and practices in Europe may be applicable in regional Japan, and vice-versa.





