Fuzzy ELM-based optimal spectrum sensing in CR-IoT network

Abstract

In the Internet of Things (IoT), spectral efficiency is a major challenge since many sensors and devices are placed widely throughout the network. Spectrum scarcity issues have been a concern for the Internet of Things since its inception. The use of Cognitive Radio (CR) technology is thought to be a potential remedy. Two of the problems that CR presents in the IoT are quick response times and efficient spectrum detection even at low signal-to-noise ratios. In order to address the spectrum scarcity issue, spectrum prediction is a key task of CR. As a result of spectrum prediction, sensor and decision-making delays are reduced, resulting in fewer collisions between primary and secondary users. To overcome this problem, an innovative GSCI-FELM technique has been proposed. The ultimate aim of the proposed method is to identify the spectrum holes and allocate the best suitable channel to the SU-IoT device. The Global Channel State Information (GSCI) estimation is first used to identify the spectrum holes locally at the SU-IoT devices, which solves the PU identification issue by converting it into an Idle Channel State (IDC) or Busy Channel State (BCS). Eight different features are extracted and given to Fuzzy ELM for classification. The secondary users are given access to the available node once it has been identified. According to the findings, the proposed technique is effective in predicting the spectrum availability of the various nodes in the IoT spectrum network. The accuracy of the proposed FELM is 5.76%, 2.96%, 7.4%, and 1.85% better than LSTM-RNN, STFT-CNN, SVM and SHAE respectively.