I am listing a set of projects that in a particular form would be useful for a topic for special problem class or for a senior capstone design student group. I am also very excited by the opportunity for other undergraduate research groups to make use of this technology. The particular details will be different in these two cases.
The large-scale Field Programmable Analog Array (FPAA) concept has the potential to revolutionize computing in many ways seen by the digital VLSI movement in the 1970s and 1980s. I invite potential undergraduate and graduate students into this opportunity.
I fully expect each of these projects can lead to publication(s) as well as eventual industrial interest.
For a special topics class, there are clear requirements in the first three weeks of the semester:
For capstone senior design class, I would highly recommend the using the FPAA and resulting interfacing concepts for most projects. I've seen the struggle with building Custom Off-The-Shelf hardware in the past, and I would like to see groups get past many of these issues going forward. The timing might need to be different depending on the particular interest.
As an aside, only in very rare cases would an incoming student be funded through my research funds the first semester. I almost never fund terminal MS students, and never coursework MS students (non-thesis option). I want to make expectations clear.
Ultra-Low Power Sensor Data Analysis and Classification: Embedded Machine Learning. Follow a structure like Fig 8 in SoC FPAA paper. Sensor: microphone (acoustic or ultrasound), MEMs ( motion, vibration)
Music synthesis : Our laboratory has published circuit(s) for music synthesis on an earlier FPAA device , and our group mentored a senior design team ( F2015-S2016) looking at music synthesis on an FPAA. We continue to have significant interest in many areas in this field. This area is excellent to learn IC design (MOSFETs) concepts if you enjoy discrete (e.g BJT + Resistors ) design. Special topics focus naturally looks to building core music synthesis blocks, particularly ones earlier developed to be widely used.
Vision: Vision classification. Recent results in image convolution focused towards subsample front end for convolution.
Communications . Analog beamforming. Spectrum sensing.
Analog for E and M simulation. No inductor, so GmC build.
Neuron model . Neuron WTA. VMM input and comparison. Synapse type inputs.
Improving Configurable Logic Block (CLB) efficiency in FPGA / FPAA devices: Because we have our own FPGA / FPAA devices, we know all of the details of the architecture. Typical FPGA companies (Xlinix, Altera, Cyprus) hide their architectural details from the user. As a result, we can explore circuit opportunities in these fabrics.