Without AI and ML, IoT is Just as Good as M2M


Thingify Pvt. Ltd. is a Delhi-based IoT start-up catering to segments like Industry 4.0 and Smart Cities with its IoT platform and analytics-as-a-service solution. In an interview with EFY bureau, Rishi Gargi, Founder, Thingify talks about key technologies, tech ecosystem in India, deployment of IoT, security of IoT solutions and more.


Q. In your opinion, which are the top 3 technologies that are shaping the future of IoT? How are they influencing the future of IoT?

The three technologies that I would bet on are:

  1. Artificial Intelligence (AI): AI will completely reshape how we use our devices and help achieve the most optimal way of using the connected devices.
  2. Edge: Edge is another aspect of making devices smarter and have intelligence at the local. It will also significantly reduce the data to be transmitted to the cloud.
  3. LPWAN: LPWANs will help connected devices operate at a lower cost with greater power efficiency than traditional GSM & WiFi based systems.

Q. How’s the development of standards (or lack of it) affecting the adoption of IoT?

Rishi Gargi, Founder, Thingify

The disparate devices and applications offered by an array of original equipment manufacturers (OEMs) and solution providers make adoption of IoT tougher for business operations and usage of IoT devices in day-to-day life. Also, the competing standards such as LoRa vs NB-IoT in terms of connectivity, Industry 4.0 (Germany) vs IIC (USA) for IIoT, OneM2M alliances vs OIC for specifications are making it harder for systems to be interoperable. Development of standards will reduce the interfacing cost and help bring these solutions onto common platform and therefore build correlation between the device data.

Q. How ready is India’s tech eco-system to develop and deploy IoT solutions?

India’s tech eco-system is evolving for IoT solutions but it needs support to build hardware in house and investment to support R&D, only then it will become economically viable for Indian IoT ecosystem to prosper. There are limited incubators for IoT and minimal support for start-ups. As per NASSCOM, the IoT market in India is poised to reach USD 15 billion by 2020. Government bodies like NITI Aayog should introduce more initiatives to help Indian tech eco-system excel.

Q. Do you foresee India’s tech industry developing its own IP and branded products/solutions in the IoT arena?

Yes, India is slowly picking up on IoT. The number of start-ups in this space are increasing gradually. There have been few unique and innovative IPs recently launched by some Indian start-ups but disruption is yet to be seen.

Q. Do you see the Open Source phenomenon play an important role in the IoT arena?

Open source can play a crucial rule. Having solutions built on open source hardware, software and protocols make it easier to be integrated with other systems. True sense of IoT is achieved only when all the systems work in sync. If a system is made completely isolated, unless required, it cannot gel with the rest in the IoT ecosystem.

Q. How do you see the role, technologies like AI/ML will play in the evolution of IoT solutions?

Without AI and ML, IoT is just as good as machine-to-machine (M2M). Having decisions taken on the basis of best usage/most optimal solutions is what AI will offer. Business rules tend to get outdated with time, but AI models keep evolving with the user feedback and help drive better decision making and system adaption.

Q. What’s your opinion on the state of security available for IoT solutions? How do you see the evolution from hereon, w.r.t threats and counter-measures?

Security is one of the most important aspects to be taken care of for an IoT solution. An IoT solution is prone to vulnerabilities at various points such as local connectivity from device to hub, Wi-Fi connectivity hub to router, TCP/IP or UDP mechanisms from router to cloud, backdoor vulnerability at cloud, etc. The current solutions on the software side are IAM policies and authentication tokens. On hardware side, Microchip is working with big players like Google, Microsoft & Amazon to achieve hardware security solutions and has already released its 5 cent security IC that is compatible with most of the popular boards available. Security will evolve multi-fold as conventional methods won’t be feasible for all the layers and constraints. AI can be the driving factor for identifying unusual patterns and anomalies to identify hacks.

Q. W.r.t. edge vs cloud – where do you think will we see faster development in the next year or two?

According to me, it is Cloud. Intelligence over cloud is growing rapidly. It will take time to move it to edge. Although companies like Dell, HPE, Lenovo, Adlink, ARM and Intel are working aggressively on innovating hardware that can support computation at the edge. Lenovo recently came up with Tiny EPC that consumes lesser battery and providing high computation power. The intelligence over edge looks more promising but it will take time for such hardware to get mature in terms of processor rating, memory rating, circuits and thresholds.

Q. What are the key technologies missing, which when made available, will accelerate adoption of IoT across the globe?

The adoption of IoT is affected by the value proposition provided by the connected devices. If the cost of print circuit board (PCB) fabrication and assembly can be reduced, cost of connected devices will significantly drop. On the part of accessibility, if there can be an easier way of accessing the IoT devices (common platform and protocols), it would be more user-friendly rather than having all systems work in silos. Also, having standardisation would help make better decisions for the devices, using AI.

Q. Any thoughts on where to go edge versus where to go cloud?

I opt for edge when the decisions are to be made in real-time or when the system is to be deployed at a remote location (the connectivity to cloud is constrained). For non-real-time systems and when the computation power is a constraint, cloud is a better option. Also, cloud gives the flexibility of updating the models whereas in case of edge, it has to be done using OTA.


Please enter your comment!
Please enter your name here