Cooperative Spectrum Sensing using DQN in CRN

Abstract

Abstract —It is imperative to address the problem of spectrum under usage and inefficiency because of the increasing spectrum demand and slender spectrum resources. One of the salient functions of cognitive radio is spectrum sensing which is used to avoid the interference of the unlicensed secondary users with licensed primary users and spot the available spectrum for enhancing the spectrum usage. The frequency band that a secondary user can utilize without interfering with any licensed primary users are called spectrum holes. Cooperative sensing is a remedy to improve the sensing performance, in which secondary users (SUs) cooperate among themselves to sense the spectrum and find the spectrum holes. Here we propose a deep reinforcement learning based spectrum sensing to discover the spectrum holes. We implement a deep reinforcement learning based method called Deep Q-Network (DQN) to find the spectrum holes. The secondary users (SU) uses the DQN to find the vacant channels in the spectrum effectively. The secondary user (SU) senses the spectrum associated with a single primary user (PU). The spectrum is sensed and the spectrum holes are detected to satisfy the requirement of the secondary user (SU).It is imperative to address the problem of spectrum under usage and inefficiency because of the increasing spectrum demand and slender spectrum resources. One of the salient functions of cognitive radio is spectrum sensing which is used to avoid the interference of the unlicensed secondary users with licensed primary users and spot the available spectrum for enhancing the spectrum usage. The frequency band that a secondary user can utilize without interfering with any licensed primary users are called spectrum holes. Cooperative sensing is a remedy to improve the sensing performance, in which secondary users (SUs) cooperate among themselves to sense the spectrum and find the spectrum holes. Here we propose a deep reinforcement learning based spectrum sensing to discover the spectrum holes. We implement a deep reinforcement learning based method called Deep Q-Network (DQN) to find the spectrum holes. The secondary users (SU) uses the DQN to find the vacant channels in the spectrum effectively. The secondary user (SU) senses the spectrum associated with a single primary user (PU). The spectrum is sensed and the spectrum holes are detected to satisfy the requirement of the secondary user (SU).