Low-power computation using FPAA for wearable devices
Shah, Sahil Sandeep
MetadataShow full item record
This work investigates low-power mixed-signal computation techniques for real-time applications. The need for real-time processing, with rise of wearable devices, creates a strong drive for researching and developing methods and system architectures which reduces the power consumption. By performing the computation locally near the sensor node one can increase the energy efficiency of such devices by reducing the need for communication to the cloud. Analog computation has shown promising results in this space by significantly reducing power consumption by processing signals in analog without having to convert them into the digital domain. Further, by adding programmability and configurability to analog, the effects of process, voltage, and temperature variations could be reduced significantly.