Embedded Systems

Measure how TinyML audio classifiers change over time on microcontrollers, then analyze latency, energy use, and accuracy drift.

Build and test compressed-sensing signal reconstruction for ECG and EMG, then measure accuracy, battery savings, and tradeoffs.

Design an energy-harvested people-counting node and learn intermittent computing, checkpointing, and low-power sensing for real buildings.

Build and test a free-space optical link, then use control data to improve beam pointing, signal stability, and link reliability.

Build a low-cost water sensor system and learn calibration, signal fusion, and neural network prediction for drinking-water safety.

Build a router-side detector for DNS and ICMP tunneling, then test stream features and a small transformer on public network traffic data.

Build and test an ultrasonic data link, then measure how OFDM and learned equalizers affect nearby-device key exchange reliability.

Test how SRAM startup noise can build device keys, then measure bit stability across boards, temperatures, and voltage changes.

Build an IoT fridge model that combines thermal imaging, door events, and prediction models to estimate spoilage risk by shelf zone.

Build a ripple-monitoring classifier that spots failing USB charger capacitors early and turns noisy waveforms into useful predictive signals.