Abstract

Cognitive radio (CR) technology enables a secondary user (SU) to make the most use of the licensed spectrum when the primary user (PU) is inactive, hence increasing spectrum efficiency. SUs are required to complete the spectrum sensing process to identify the spectrum utilization. It is required to effectively detect PU signal to SU for using the idle licensed spectrum bands. Even though various spectrum sensing techniques are designed in CR networks, designing test statistics still results in a complex task. To solve spectrum scarcity issues and to increase spectrum utilization, Honey Badger Remora Optimization-based AlexNet (HBRO-based AlexNet) is developed in this research and the test statistics model is modeled with a deep learning approach with signal parameters, like signal energy, and Eigen statistics. However, SU can use AlexNet to boost the spectrum efficiency of PU's licensed spectrum, which will then permit an increase in the chance of detection in the CR network (CRN). Though considering signal-to-noise ratio (SNR) at 5 dB, the proposed spectrum sensing approach yields the probability of detection and probability of false alarm for the Rician channel as 1 and 0.999, respectively.

摘要

认知无线电 (CR) 技术使次要用户 (SU) 能够在主要用户 (PU) 处于非活动状态时充分利用许可频谱，从而提高频谱效率。SU需要完成频谱感知过程以识别频谱利用率。需要有效地检测到 SU 的 PU 信号以使用空闲的许可频谱带。尽管在 CR 网络中设计了各种频谱感知技术，但设计测试统计仍然是一项复杂的任务。为了解决频谱稀缺问题并提高频谱利用率，本研究开发了基于 Honey Badger Remora 优化的 AlexNet（基于 HBRO 的 AlexNet），测试统计模型采用深度学习方法对信号参数（如信号能量）进行建模，并且本征统计。然而，SU 可以使用 AlexNet 提高 PU 许可频谱的频谱效率，从而增加 CR 网络 (CRN) 中的检测机会。虽然考虑了 5 dB 的信噪比 (SNR)，但所提出的频谱感测方法产生莱斯信道的检测概率和误报概率分别为 1 和 0.999。