STDP Training Design and Performances of a SNN for Sequence Detection as a Wake Up Radio in a IoT Network

Wake up radio receivers are an efficient approach to reduce energy consumption in Internet of Things (IoT) nodes. They monitor the channel to identify when a specific activation sequence is transmitted. However, up to now, they rely on classical processors which still consume too much. In this paper, to reduce this energy consumption, we propose to exploit Spiking Neural Network (SNN) architecture. In order to be able to detect the sequence in various channel conditions, we exploit the Spike-Timing Dependent Plasticity (STDP) learning rule to train the SNN. We show that the SNN succeeds in performing the sequence detection accurately, and delve into the training set design to improve the accuracy.