Seminar “Internet-of-Things: Ultra Low Power Applications”
- Time: 14:30, 26/12/2022
- Venue: The Information Technology Institute, Vietnam National University, Hanoi
Invited talk #1: Beat Sensor with LoRa powered by Energy Harvesting – Technology to realize IoT sensors for long range communication
Abstract: The use of IoT (Internet of Things), in which sensors are placed everywhere and measured data is connected to the Internet, is expanding. In the near future, as many as one trillion sensors will be produced, called TSU (Trillion Sensor Universe), and will be placed at agriculture and aquaculture farms, factories, offices, homes, infrastructure, etc. The collected data become Big Data and it is processed into useful information by AI. By effectively using this information, it is possible to improve the productivity of agriculture, fisheries, and factories, to save energy in offices and homes. Then IoT will be common technology for various applications soon.
Since IoT sensors are placed everywhere, the way to send data is digital signal with wireless RF transmission. DC power supply by batteries is usually used, nowadays, however, it is difficult to replace 1 trillion batteries. Instead, energy harvesting technology that collects environmental energy is used as a power source of IoT sensors. In modern society, all kinds of radio frequency waves (RF) are flying around everywhere where people live at any time. If the energy of these radio waves can be collected as the power source for IoT sensors, both wireless power supply and wireless data transmission will be possible to realize an IoT sensor that does not require cable or battery.
In this key note speech, I will introduce our approach to realize IoT systems utilizing RF energy harvesting and low power IoT sensors. Our developing IoT sensor called as Beat Sensor with LoRa consumes 50uW power and achieves long range communication of more than 500m. The power is generated by RF energy harvesting without battery. This Beat Sensor with LoRa powered by RF energy harvesting could realize real IoT era.
Invited talk #2: Algorithms and Hardware Architectures for Spiking Neural Networks
Abstract: Spiking Neural Network (SNN) – the third generation of Artificial Neural Network (ANN) has gained many research interest o-ver the years. Due to its bio-plausible nature, it has been shown that SNN could reach higher energy efficiency, with comparable computing capabilities when compared to ANN. However, deploying SNN onto many embedded platforms with limited resources is currently facing many difficulties, due to the size of many SNN networks. Furthermore, current state-of-the-art network SNN with convolutional topology is having issues with finding the optimal solution for the Max-Pooling Network. In this talk, we would like to introduce our latest research works which aims to solve the mentioned challenges, from designing new training algorithms for SNN and proposing efficient hardware architectures for SNN.
