Enhanced Channel Estimation Using Dilated Convolutional LSTM in CRN-IoT Systems

Abstract

Cognitive Radio Networks (CRN) inside the Internet of Things (IoT) provide dynamic spectrum management, improving communication efficiency through the utilization of underutilized frequency segments. Current deep learning models for channel estimation in cognitive radio networks encounter issues including elevated computing complexity, sluggish adaptability to swiftly changing settings, and limitations in managing the varied characteristics of IoT devices. The Adaptive Skip-based Convolutional Deep-Skip Long Short-Term Memory (AdSk-based ConvD-SkipLSTM) model effectively resolves these challenges by delivering expedited and precise spectrum sensing and channel estimation, hence enhancing overall network performance. The identification of unused spectrum bands is conducted by the energy detection method. Subsequently, channel estimation is executed utilizing the proposed AdSk-based ConvD-SkipLSTM model. The suggested model improves the precision and efficacy of channel estimation, guaranteeing dependable communication. The proposed channel estimation model is evaluated using metrics such as normalized mean square error (NMSE), outage probability, and bit error rate (BER), demonstrating superior performance compared to traditional channel estimation techniques. The proposed method achieved a minimal BER of 1.62E-06 in comparison to current channel estimating techniques.