High-Speed Sensing of the Electromagnetic Environment for Cognitive Radio Receivers

Abstract

In this paper we demonstrate an EM environment aware (EMEA) radio called the Intelligent Transceiver Radio Node (ITRN) that is suitable for use in cognitive radio applications.
The ITRN is an end-to-end solution that can quickly find interferers and act upon them in a defensive manner such as filter, move the channel, move to different band, etc. While the ITRN is capable of finding interferers in both the spectral dimension, we present a framework that allows for future expandability into more measurement domains.
To break to sensing time tradeoff with spectral and angular resolution, we employ the use of compressed sensing (CS). By making a few assumptions on the local EM environment’s current state, we are able to perform spatial and spectral scans that are a factor of 10 times faster than the current state of the art. Information on the spectral locations of the interferers, along with a current QoS estimate is then sent to a machine learning based decision engine (MLBDE) where reinforcement learning is used to determine the optimal channel selection.
For the ITRN’s sensor, we use a custom 8 antenna RF-ASIC fabricated in TSMC 65nm CMOS called the Direct Space to Information Converter (DSIC). The output of the DSIC is sent to an Ettus X310 radio. A custom UHD interface was constructed in the field programmable gate array (FPGA) to speed the streaming data rate by using a variable data packet size. Custom UHD circuitry was also created to synchronize the DSIC with the clock on the X310.
In GNU Radio, we perform the baseband DSP and Orthogonal Matching Pursuit (OMP) which is used to recover the spectral locations of the interferers. Lastly, the output of OMP along with a QoS estimation is sent to the MLBDE which calculates the new optimal channel selection and retunes the ITRN.