A method that predicts how much conflicts among O-RAN applications will degrade network performance, using profiling data and weighting each application by the frequency of its control actions.

A mathematical model of a real antenna tracking system, including motion latency and velocity limits, quantifies how mechanical pointing errors degrade a THz uplink during LEO satellite passes and yields design guidelines for high-frequency ground stations.

PACIFISTA profiles O-RAN applications in a sandbox and combines hierarchical graphs with statistical models to detect conflicts and gauge their severity before deployment, showing that even xApps with similar goals can cost users 16% of throughput.

A patented method that profiles network-control applications in simulation, predicts whether they will conflict and how severely, and decides which applications can safely be deployed together.

An automation framework built on OpenShift and GitOps that deploys an end-to-end, O-RAN-compliant 5G network in seconds and has run months of over-the-air tests without human intervention.

SeizNet fuses implantable and wearable sensor data in deep-learning models that run at the edge, predicting epileptic seizures with over 97% accuracy within the size and energy budget of an implantable device.

A patented network architecture in which an edge-computing orchestrator inside the Open RAN instantiates applications on demand to control autonomous vehicles beyond line of sight.

Field and simulated experiments with a drone collecting data from wake-up-radio sensor nodes, showing that adaptive route planning beats a fixed path on every metric and that wake-up radios extend network lifetime by orders of magnitude over duty cycling.

Ground sensors and satellite imagery are fused into a georeferenced map of vine water stress, giving farmers contact-free monitoring of crop health; validated with a prototype in a working vineyard.

A feasibility study of controlling drones beyond visual range over commercial cellular networks, showing that the Colosseum hardware-in-the-loop emulator reproduces real-network behavior closely enough to serve as a testbed for new UAV features.