Adaptive Closed-Box Interference Susceptibility Testing

Summary

Modern communication systems are designed to adaptively operate in complex, time-varying radio frequency (RF) environments.  However, current state-of-the art methods for assessing interference susceptibility in controlled laboratory settings rely on steady-state equilibrium responses that may not adequately reflect real-world use cases.  With the goal of moving beyond traditional steady-state test methods, this project seeks to develop new approaches to assess interference susceptibility of closed-box, commercial off-the-shelf (COTS) communication systems in complex, dynamic spectrum environments.  The work will result in more relevant and efficient laboratory testing to support spectrum policy decision making and technology deployment. 

Description

Our approach combines state-of-the-art testbed automation, RF metrology, data analytics, and machine learning to develop flexible, broad-purpose methods that will enable efficient, rigorous testing.  Specifically, we plan to establish methods for experimental characterization of closed-box communication links as dynamical systems, including assessment of system dynamics (system identification) and interference impacts.  Next, we aim to demonstrate adaptive test methodologies built on dynamical systems models and machine learning, e.g., reinforcement learning.