A 0.82 µW CIS-based Action Recognition SoC with Self-Adjustable Frame Resolution for Always-on IoT Devices
Abstract : An always-on video-based human action recognition (HAR) system on chip (SoC) integrated with a CMOS image sensor (CIS) is proposed for the Internet of Things (IoT) devices. The proposed SoC is the first always-on integrated circuit (IC) performing the full process of HAR in a single chip. To resolve large power consumption from vision sensor and compute-intensive DNN operation, the proposed SoC operates in two different modes; 1) in adaptive frame resolution based human action recognition (AFR-HAR) mode, CIS resolution prediction algorithm and self-adjustable CIS reduce 42.9–91.8% of readout power by adaptively adjusting frame resolution. 2) In motion event detection (MED) mode, the motion event detection unit (MEDU) skips unnecessary imaging and DNN computation by monitoring motion events and leads to over 99% power saving. The proposed HAR SoC is simulated in 65-nm CMOS technology and occupies 8.56 mm 2 . It consumes only 0.82 ${\mu }\text{W}$ when no motion is detected and 0.31–8.52 mW for evaluating human actions on the ActivityNet dataset.
? The recent developments in digital technologies have provided a driving force to apply smart, IoT based solutions for the existing problems in a smart city context.
? This paper reviews the existing literature on the application of IoT in in energy systems, in general, and in the context of smart grids particularly.
? As there are variety of sensor and actuator devices, communication technologies, and data computing approached, in this section, we explain the existing technologies which enable IoT.
? The main contribution of this paper is to extend the existing body of literature by providing energy policy-makers, economists, energy experts, and managers with a general overview of the opportunities and challenges of applying IoT in different parts of the entire energy sector.
? The aerospace vehicle may stay away from the sun for months, that is why solar energy can not be a proper solution for this problem.
? Several issues involving batteries, such as costs, high maintenance, limited lifetime, and environmental issues, make them a poor solution for IoT .
? The energy harvesting networks expose themselves to prominent security issues putting the secrecy of the system to risk.
? Renewable energy source harvesters possess the advantage of low volume, low weight, limited environmental impact, and long life.
? They discovered that one impact can result in energy below 90 nJ with 2.5 lW of power for one harvester unit.
• The work proposes a platform by integrating multiple systems, such as air-conditioning, lighting, and energy monitoring to perform building energy optimization.
• Other solutions propose designing co-simulation models for energy systems to integrate the system and minimize synchronization delay error between the subsystems.
• The research in proposes a wireless sensor and actuator network to provide an IoT-based automatic intelligent system.
• Whereas, by optimizing the operation of devices and machines in the IoT, the proposed system achieves reduction in their overall energy consumption at a given time.
? To overcome this limitation on WSN, there is a requirement to explore an efficient harvesting system with high performance for the WSN environment.
? At high strain amplitude with 0.5 million cycles of testing, it was observed that there was a severe loss in the performance of the PEH.
? They faced low output performance because of the small device bandwidth, and device inherent frequency and common sound frequency mismatch.
? This limits the WSN’s performance, its lifetime, and its capacity because of the regular replacement on the depletion of the battery.
? Energy harvesting for the low-power devices from sources like thermal, solar, wind, radiofrequency, sound, etc. has been performed efficiently over the last decade.
|