Periodic Reporting for period 2 - EPEAS (Leading the way to energy autonomous edge computing)
Reporting period: 2024-02-01 to 2025-07-31
IoT (Internet of Things) devices are deployed in volume to sense, measure, monitor, track… Many studies have tried to estimate the number of connected devices to be deployed in the coming years, and all end up with numbers above the billion range. Those devices are not connected to the grid and therefore rely on a battery as the source of energy. However, battery replacement causes maintenance headaches to consumers and industries. Indeed, for the consumer market the need to replace a battery lead to a poor customer experience, while for industrial applications, the number of batteries to be replaced causes high labour costs. Further the manufacturing and disposal of those batteries have a significant environmental impact: where and how to extract the row materials and how to recycle the batteries when they reach their end-of-life.
With this project we offer to solve the battery issue. An IoT device is built around four different functions: A sensor that gather information about the object environment, a microcontroller that process this information, a communication mean that send the results of the processing to the cloud and a power management that delivers energy to the object. At e-peas we already offer power management circuits that can be combined with tiny energy harvesters to supply the object from the energy available from its environment. The goal of this project is to combine it with a microcontroller and a sensor (namely an image sensor) that will require a minimum amount of power to operate. This combination will provide energy autonomy to the objects, removing the burden of battery replacement.
With this project we offer to solve the battery issue. An IoT device is built around four different functions: A sensor that gather information about the object environment, a microcontroller that process this information, a communication mean that send the results of the processing to the cloud and a power management that delivers energy to the object. At e-peas we already offer power management circuits that can be combined with tiny energy harvesters to supply the object from the energy available from its environment. The goal of this project is to combine it with a microcontroller and a sensor (namely an image sensor) that will require a minimum amount of power to operate. This combination will provide energy autonomy to the objects, removing the burden of battery replacement.
During the second reporting period, two groups of activities have been conducted:
Regarding the microcontroller,
• A proprietary communication interface optimized for interaction with our power management integrated circuit (PMIC) portfolio has been developed and tested.
• A module featuring a combination of a PMIC and our microcontroller has been developed. Its goal is to act as the master of a system relying on several circuits.
• The firmware toolset required for proper usage of our microcontroller has been updated (bug corrections, integration of the VS Code IDE, addition of new drivers...).
• We have also set the ground for upcoming developments on an advanced technology node with updates to the design and verification flows, and automation of the software development flow in order to leverage the work already done for the former development.
• The MCU has been experimentally validated and qualified.
• Finally, examples of code have been written and an evaluation platform has been developed to support the commercial development of our product.
Outcomes are that: we have all the material and collaterals required for the commercial development of this product line. Patents have been submitted to protect our innovation. Our company is now also ready to extend the MCU family with new developments toward products able to perform more complex tasks and featuring AI in advanced technology nodes.
Regarding the image sensor:
• Power consumption has been reduced thanks to the rework of power hungry IPs.
• Hardware accelerators have been developed to support on-chip image pre-processing, enabling a reduction of the amount of data transferred outside of the sensor.
• We have investigated and applied corrections to the photodiode to improve the black level of the image.
• Two new prototypes have been manufactured and experimentally validated.
• Based on those measurements, a new model of the photodiode has been developed for future development.
• A Graphical User Interface has been developed to ease the evaluation of the sensor (by e-peas or by its customers).
Outcomes are that: we have all the material required for the commercial development of the image sensor. Patents have been submitted to protect our innovation as well for this product.
Regarding the microcontroller,
• A proprietary communication interface optimized for interaction with our power management integrated circuit (PMIC) portfolio has been developed and tested.
• A module featuring a combination of a PMIC and our microcontroller has been developed. Its goal is to act as the master of a system relying on several circuits.
• The firmware toolset required for proper usage of our microcontroller has been updated (bug corrections, integration of the VS Code IDE, addition of new drivers...).
• We have also set the ground for upcoming developments on an advanced technology node with updates to the design and verification flows, and automation of the software development flow in order to leverage the work already done for the former development.
• The MCU has been experimentally validated and qualified.
• Finally, examples of code have been written and an evaluation platform has been developed to support the commercial development of our product.
Outcomes are that: we have all the material and collaterals required for the commercial development of this product line. Patents have been submitted to protect our innovation. Our company is now also ready to extend the MCU family with new developments toward products able to perform more complex tasks and featuring AI in advanced technology nodes.
Regarding the image sensor:
• Power consumption has been reduced thanks to the rework of power hungry IPs.
• Hardware accelerators have been developed to support on-chip image pre-processing, enabling a reduction of the amount of data transferred outside of the sensor.
• We have investigated and applied corrections to the photodiode to improve the black level of the image.
• Two new prototypes have been manufactured and experimentally validated.
• Based on those measurements, a new model of the photodiode has been developed for future development.
• A Graphical User Interface has been developed to ease the evaluation of the sensor (by e-peas or by its customers).
Outcomes are that: we have all the material required for the commercial development of the image sensor. Patents have been submitted to protect our innovation as well for this product.
The MCU has been experimentally validated and qualified. Documentation and software tools required for customer use have been developed.
The microcontroller's active power consumption is only 18 µA/MHz, and its quiescent current in deep sleep drops to less than 350 nA. A module featuring a proprietary communication interface between our microcontroller and one of our PMICs has been successfully developed.
The next step towards developing a product family is the inclusion of hardware accelerators for AI and the creation of products capable of handling more complex processing tasks.
The image sensor has also been experimentally validated and qualified. Software and documentation for customers are ready. The image sensor achieves a power consumption of only 300 pJ/frame/pixel. Thanks to corrections addressing the black level issue, we have achieved a signal-to-noise ratio above 40 dB and a dynamic range above 50 dB, in line with our initial goals.
The next step for commercial success is to establish strong partnerships with other semiconductor companies, such as solution providers for local image processing at the edge.
The microcontroller's active power consumption is only 18 µA/MHz, and its quiescent current in deep sleep drops to less than 350 nA. A module featuring a proprietary communication interface between our microcontroller and one of our PMICs has been successfully developed.
The next step towards developing a product family is the inclusion of hardware accelerators for AI and the creation of products capable of handling more complex processing tasks.
The image sensor has also been experimentally validated and qualified. Software and documentation for customers are ready. The image sensor achieves a power consumption of only 300 pJ/frame/pixel. Thanks to corrections addressing the black level issue, we have achieved a signal-to-noise ratio above 40 dB and a dynamic range above 50 dB, in line with our initial goals.
The next step for commercial success is to establish strong partnerships with other semiconductor companies, such as solution providers for local image processing at the edge.