An Adaptive Recurrent Neural Network Model Dedicated to Opportunistic Communication in Wireless Networks
Silas S. Fernandes, Mariana Rodrigues Makiuchi, Marcus Vinicius Lamar, Jacir Luiz BORDIM · 2018
One of the major challenges in opportunistic networks is the correct identification of a transmission opportunity and its corresponding duration. In this work, a new adaptive model for opportunity forecast is proposed. The system is based on in-channel spectrum sensing and the use of recurrent neural network to model the occupation of the channel and detect the correct moment of transmission opportunity. The results, based on realistic experiments using a Software Defined Radio for monitoring a Wi-Fi channel, are presented. The proposed model reached a precision of 82.52% for noisy environment and 96.78% for mild environment, decreasing significantly the false positive rate comparing to non-adaptive recurrent neural equivalent, which is an important aspect in opportunistic use of a channel.