Firstly, IoT is increasingly finding diverse applications. Secondly, many IoT devices are remotely connected. Both cases call for power management. The source of power becomes a challenge in situations when sensors are integrated into concrete infrastructure. “Energy savings come from power efficiency. What’s required is a multi-day battery life for products. The beginning must be made by strengthening the ecosystem, then building around the devices and ensuring an extended battery life,” said Aditya Bedi, Arm, speaking at the Arm Tech Symposia India 2018. 

At a basic level, many power-related issues need to be resolved. Leakages in the power supply circuit or even interrupted power supply can pose a problem. A drop in battery voltage is another stumbling block. “For power optimisation to happen, the focus should be on multi-year lifetime devices that run on a single battery charge,” said Bedi.

Power has many dimensions that open out channels of innovation. For instance, startups and entrepreneurs can chalk out storage solutions that include renewable forms of energy. The quality of power supply and its affordability are other aspects that need to be considered, along with metering, billing and monitoring. Power-saving solutions need to be instilled into large and small devices. The reason for including small devices is that most of them are in large numbers. Further, many devices are spread on a large area and are single-charge versions. Being widespread, charging these devices may not be very easy.

Consequently, devices should be conceptualised as multi-lifespan options. Devices with longer battery life lower the power consumption. However, devices like smart locks, smart bandage, medical nano-robots and asset metering don’t require charging. “Design optimisation results in energy efficiency. Ultra-low voltage compute seeks to maximize energy efficiency through active power,” highlighted Vasant Sivashanmugam, Arm.  

Along with power management, IoT devices require edge computing because of the quantum of data generated and the speed at which it happens. Edge computing elicits different definitions. Simply put, it is computing data which is close to edge of the network and not located at a central point. Computing on the edge offers many advantages, the most prominent being security. While securing sensitive data, it can be retained at source and needn’t be stored in a centralised cloud.  

Another noteworthy dimension at the Symposia was the discussion on machine learning. Described as a shift in computing, machine learning enables intelligence at the edge using a single traditional performance. Machine learning and computer vision markets are evolving rapidly. “Solutions, algorithms and networks are constantly changing. Machine-learning processing requires static scheduling, efficient convolutions, bandwidth reduction mechanisms and programmability,” explained Tanuj Arora, Arm. 

Future trends point to a need for machine learning-equipped devices and processors, along with object detection features. However, many of the existing solutions in the market are somewhat closed and hence offer little flexibility. “We want to provide choice and flexibility to customers. For us, machine learning and AI is top priority. We have built optimised ground-up architecture for machine learning processing. The software is open source (OS), so it enables easy deployment,” added Arora. The soon-to-be-released offering is targeted at mobiles and embedded solutions.

At the Arm symposia, several companies showcased their products. One of the most interesting ones was from KLE Technological University (KLE Tech) in Hubli. KLE Tech has forged industry collaborations and joint research programmes. Its board of studies of every programme has at least two senior members from leading tech companies. The latest initiative is a project on smart transportation, in which the students used Arm products to showcase various dimensions of autonomous vehicles.

Firstly, IoT is increasingly finding diverse applications. Secondly, many IoT devices are remotely connected. Both cases call for power management. The source of power becomes a challenge in situations when sensors are integrated into concrete infrastructure. “Energy savings come from power efficiency. What’s required is a multi-day battery life for products. The beginning must be made by strengthening the ecosystem, then building around the devices and ensuring an extended battery life,” said Aditya Bedi, Arm, speaking at the Arm Tech Symposia India 2018. 

At a basic level, many power-related issues need to be resolved. Leakages in the power supply circuit or even interrupted power supply can pose a problem. A drop in battery voltage is another stumbling block. “For power optimisation to happen, the focus should be on multi-year lifetime devices that run on a single battery charge,” said Bedi.

Power has many dimensions that open out channels of innovation. For instance, startups and entrepreneurs can chalk out storage solutions that include renewable forms of energy. The quality of power supply and its affordability are other aspects that need to be considered, along with metering, billing and monitoring. Power-saving solutions need to be instilled into large and small devices. The reason for including small devices is that most of them are in large numbers. Further, many devices are spread on a large area and are single-charge versions. Being widespread, charging these devices may not be very easy.

Consequently, devices should be conceptualised as multi-lifespan options. Devices with longer battery life lower the power consumption. However, devices like smart locks, smart bandage, medical nano-robots and asset metering don’t require charging. “Design optimisation results in energy efficiency. Ultra-low voltage compute seeks to maximize energy efficiency through active power,” highlighted Vasant Sivashanmugam, Arm.  

Along with power management, IoT devices require edge computing because of the quantum of data generated and the speed at which it happens. Edge computing elicits different definitions. Simply put, it is computing data which is close to edge of the network and not located at a central point. Computing on the edge offers many advantages, the most prominent being security. While securing sensitive data, it can be retained at source and needn’t be stored in a centralised cloud.  

Another noteworthy dimension at the Symposia was the discussion on machine learning. Described as a shift in computing, machine learning enables intelligence at the edge using a single traditional performance. Machine learning and computer vision markets are evolving rapidly. “Solutions, algorithms and networks are constantly changing. Machine-learning processing requires static scheduling, efficient convolutions, bandwidth reduction mechanisms and programmability,” explained Tanuj Arora, Arm. 

Future trends point to a need for machine learning-equipped devices and processors, along with object detection features. However, many of the existing solutions in the market are somewhat closed and hence offer little flexibility. “We want to provide choice and flexibility to customers. For us, machine learning and AI is top priority. We have built optimised ground-up architecture for machine learning processing. The software is open source (OS), so it enables easy deployment,” added Arora. The soon-to-be-released offering is targeted at mobiles and embedded solutions.

At the Arm symposia, several companies showcased their products. One of the most interesting ones was from KLE Technological University (KLE Tech) in Hubli. KLE Tech has forged industry collaborations and joint research programmes. Its board of studies of every programme has at least two senior members from leading tech companies. The latest initiative is a project on smart transportation, in which the students used Arm products to showcase various dimensions of autonomous vehicles.