chance.tarver
Projects

2017–2021 · Co-author / RL & systems

CBRS & Dynamic Spectrum Sharing

Reinforcement-learning-aided spectrum sharing in the CBRS band. Use-or-share frameworks, channel selection for unlicensed LTE, and the measurement scaffolding behind them.

The 3.5 GHz CBRS band introduced a three-tier spectrum-sharing regime (incumbents, PAL licensees, GAA users) and immediately raised practical questions: how do PAL holders and GAA users actually coexist without throwing away spectrum? How do unlicensed LTE eNBs pick channels well when they share the band with Wi-Fi and other LTE operators?

This line of work, done across my Samsung internship years and into early full-time, used reinforcement learning to attack both questions. The “use-or-share” policy work showed RL-aided listen-before-talk can increase spatial reuse at lower-tier nodes while keeping interference at higher tiers bounded.

Selected publications