An Oscillator Based Energy Efficient Computing Architecture for Smart Sensors

In the past decade, there is a fast-growing need for sensing technology applications. However, the implementation of sensing and proceeding creates a large burden on data processing. The idea for designing stable cluster state for coupled phase oscillator in pattern recognition is receiving significant attention. This paper gives an overview of an oscillator neural network (ONN) based hierarchical associative memory (AM) architecture using oscillator synchronization and stable cluster state for pattern recognition and how such architecture can be efficiently used in local processing units. The ONN based AM architecture can be easily achieved using CMOS technology on local hardware units. Unlike cloud computing system, our architecture provides an approach when it is under an offline mode which couldn't get responses from the web server. The proposed architecture can help perform sensing and data processing efficiently on the local device without connecting to the Internet.