Interference mitigation using a ML-driven xApp in O-RAN architecture

Authors

  • DIANA LIN Department of Cybersecurity Engineering, George Mason University, Fairfax, VA
  • SAMARTH BHARGAV Department of Cybersecurity Engineering, George Mason University, Fairfax, VA
  • Nathaniel Kabigting Department of Cybersecurity Engineering, George Mason University, Fairfax, VA
  • Azuka Chiejina Department of Cybersecurity Engineering, George Mason University, Fairfax, VA
  • Vijay Shah Department of Cybersecurity Engineering, George Mason University, Fairfax, VA

DOI:

https://doi.org/10.13021/jssr2023.3964

Abstract

The proliferation of wireless devices and technologies has led to an increase in the number of bad actors in cellular networks. Consequently, the differentiation between signals of interest (SOI) and various types of interference has become crucial in mitigating performance degradation in the network. We leverage the flexibility of Open Radio Access Network (O-RAN) architecture to implement an interference-classifying microservice capable of controlling communication between user and base station. By deploying a trained convolutional neural network (CNN) in a RAN Intelligent Controller (RIC), we are able to classify and mitigate different types of interference with up to 98.9% accuracy. Our deployment of the microservice as an extended application (xApp) dynamically adjusts the modulation coding scheme (MCS), resulting in a more stable bitrate and a lower block error rate under conditions with interference as opposed to using a fixed MCS.

Published

2023-10-27

Issue

Section

College of Engineering and Computing: Department of Cybersecurity Engineering

Categories