Research

Work in RF CMOS and optoelectronics established ways to model physical signal paths, identify the impairments that survive to a receiver, and place correction where it has practical value. Those same questions reappear in short-haul optical and copper interconnects, where drivers, channels, front ends, equalizers, clocks, and data converters must be understood as one coupled system.

Data conversion and adaptive analog interfaces draw directly on that foundation. Sampling, calibration, phase noise, timing, and converter architecture connect to the earlier work on oscillators and synchronization, while biomedical sensing and detector electronics extend the same concerns to weak physical signals, event-driven readout, and constrained front ends. The application changes, but the need to preserve information through imperfect hardware remains constant.

Neuromorphic, quantized, and AI-assisted hardware add computation to this lineage. They reuse ideas from mixed-signal modelling, approximation, timing, calibration, and observable-aware correction, while also providing new methods for automated design and adaptive control. The streams therefore reinforce one another: theory supplies limits and models; circuits expose real constraints; sensing and communication systems define useful observables; and intelligent hardware provides new ways to implement, verify, and adapt the resulting solutions.