Neuron-inspired signal propagation is proposed for communication in networks of nanodevices. Nanodevices should be able to interpret and forward signals inside the network in order to transport the information between two endpoints. Applications at the nano level demand processing systems that are very power efficient and simple. To achieve that, a brain inspired spiking neural network with pattern recognition and relaying capabilities is presented. The neural network learns the desired features using STDP, a power efficient and biologically plausible learning method. Finally, several nanonetworks are simulated, communicating using OWC. The results obtained show that signal similarity between the emitted and received signal highly depends on the design space of the neurons. It is possible to create networks with NDs capable of transporting information between two endpoints.