Daniel Villamizar

Research Interests: Low-power circuit design, machine-human interface systems, machine learning hardware acceleration, sensor systems, analog analytics

Email: danvilla AT stanford DOT edu

One of the key enablers of the Internet of Everything (IoE) will be the ability of systems to interface with many sensors while minimizing power consumption and data payloads.

A common feature that will be present in many IoT systems will be that of voice command recognition. In addition, acoustic environment awareness could present significant benefits to cognitive systems which can make decisions based on what they are hearing. These are examples of “always-on” systems which will need to consume a minimal amount of power in order to maximize battery life or allow for energy harvesting.

We are researching analog front end (AFE) circuit topologies that will enable acoustic event/evironment sensing at sample rates significantly lower than the Nyquist rate of the signal. In order to achieve this, a combination of prior knowledge of the signal of interest along with signal classification techniques could be useful to lower the requirements of the AFE of the system. Furthermore, if the required A/D converter can be reconfigured, additional power savings can be realized through machine learning techniques.

Ultimately, the goal of our research is to propose a design framework that can be reconfigured to efficiently capture a wide range of sensor signals (e.g. pressure, light, touch, motion, gesture, etc.).