A Recurrent Neural Network MAC Protocol Towards to Opportunistic Communication in Wireless Networks

One of the major challenges in opportunistic networks is the correct identification of transmission opportunities. In this work, a new cognitive MAC protocol, called Makiuchi is proposed. The Makiuchi is builtin using a recurrent neural network to model the channel occupation and to detect the exact moment of transmission opportunity. The training process was performed based on data from a real world experiment using Software Defined Radio (SDR) for monitoring a Wi-Fi channel. The Makiuchi MAC protocol was implemented using the discrete event simulator OMNeT++ and INET networking framework. Preliminary simulation results demonstrate the expected behavior of the Makiuchi protocol in the process of identification and allocation of Secondary Users' transmission opportunities. When compared with IEEE 802.11, the Makiuchi algorithm is able to improve in 22% the network performance, embracing a bigger number of transmissions opportunities, while it reduces in ≈50% the number of collisions with the Primary User.