Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Hardware-in-the-Loop Evaluation of Goodness of Fit (GoF) Testing for Dynamic Spectrum Sharing

Mir Lodro, Simon Armour, Mark A. Beach

TL;DR

This work tackles robust primary-user detection for dynamic spectrum sharing using nonparametric goodness-of-fit tests. It develops a hardware-in-the-loop evaluation with an RFSoC 4x2 transceiver and a Keysight F8 channel emulator to implement KS, Cramer-von-Mises, and Anderson-Darling GoF tests, complemented by eigenvalue-based detectors. The results show noise-only data conform to theoretical noise distributions while PU presence induces detectable deviations; signal data are best described by a lognormal fit with strong supporting statistics, confirming the viability of GoF-based sensing on real hardware. Overall, the study demonstrates the practicality and effectiveness of nonparametric GoF spectrum sensing for reliable dynamic spectrum sharing in real-world hardware settings.

Abstract

In contrast to parametric spectrum sensing, non-parametric spectrum sensing can effectively detect the primary user's presence or absence without prior information about the primary user. Particularly, non-parametric spectrum sensing can be useful in dynamic spectrum sharing. The secondary user must detect incumbents and peer secondary users in dynamic spectrum sharing. The secondary user can use the licensed spectrum if the primary user is not detected using its band. The primary user detection problem is the goodness-of-fit testing problem. In this work, we performed a hardware-in-the-loop evaluation of goodness-of-fit tests such as Cramer-von-Mises (CM), Anderson-Darling (AD) and Kolmogorov-Smirnov (KS) tests. We used a wideband radio transceiver RFSoC 4x2 from AMD and an F8 radio channel emulator to perform GoF tests.

Hardware-in-the-Loop Evaluation of Goodness of Fit (GoF) Testing for Dynamic Spectrum Sharing

TL;DR

This work tackles robust primary-user detection for dynamic spectrum sharing using nonparametric goodness-of-fit tests. It develops a hardware-in-the-loop evaluation with an RFSoC 4x2 transceiver and a Keysight F8 channel emulator to implement KS, Cramer-von-Mises, and Anderson-Darling GoF tests, complemented by eigenvalue-based detectors. The results show noise-only data conform to theoretical noise distributions while PU presence induces detectable deviations; signal data are best described by a lognormal fit with strong supporting statistics, confirming the viability of GoF-based sensing on real hardware. Overall, the study demonstrates the practicality and effectiveness of nonparametric GoF spectrum sensing for reliable dynamic spectrum sharing in real-world hardware settings.

Abstract

In contrast to parametric spectrum sensing, non-parametric spectrum sensing can effectively detect the primary user's presence or absence without prior information about the primary user. Particularly, non-parametric spectrum sensing can be useful in dynamic spectrum sharing. The secondary user must detect incumbents and peer secondary users in dynamic spectrum sharing. The secondary user can use the licensed spectrum if the primary user is not detected using its band. The primary user detection problem is the goodness-of-fit testing problem. In this work, we performed a hardware-in-the-loop evaluation of goodness-of-fit tests such as Cramer-von-Mises (CM), Anderson-Darling (AD) and Kolmogorov-Smirnov (KS) tests. We used a wideband radio transceiver RFSoC 4x2 from AMD and an F8 radio channel emulator to perform GoF tests.
Paper Structure (8 sections, 7 equations, 7 figures, 1 table)

This paper contains 8 sections, 7 equations, 7 figures, 1 table.

Table of Contents

  1. Introduction
  2. Spectrum Sensing Scenario
  3. Goodness-of-Fit (GoF) Tests
  4. Kolmogorov-Smirnov
  5. Cramer-von-Mises and Anderson-Darling
  6. Hardware Setup
  7. Results and Discussions
  8. Conclusion

Figures (7)

  • Figure 1: Schematics of spectrum sensing and the SU coexistence with incumbents and peer SU network.
  • Figure 2: Measurement setup.
  • Figure 3: Distribution fitting and the CDF plots of the noise data at SU 1. The test statistics of noise data are measured using MME, ME-AM, ME-GM, and AM-GM.
  • Figure 4: CDF plots and distribution fitting of MME test statistics of the noise samples and signal samples.
  • Figure 5: CDF plot and distribution fitting of the signal data of SU 1 using AM-GM eigenvalue detection. Candidate distributions applied are lognormal, Chi-square, gamma, and exponential distribution.
  • ...and 2 more figures