Wireless Spiking Neural Network for Internet of Things <Abstract> The Spiking Neural Network (SNN) is a neural network, whose behavior is close to natural brain activities. It is possible to achieve intelligent signal processing like object recognition.Because the SNN expresses the neuron infor- mation by spike signals, the energy consumption for in- formation processing is much smaller than the Neuman- architecture computing. The Wireless Sensor Network (WSN), which are linked by wireless communications col- lects fruitful environment data continuously and automati- cally by sensor nodes and transmits the data to the network server. It is, however, too difficult to transmit all the envi- ronment data to the data server because of communication capacity and data explosion. Additionally, the power con- sumption is also a problem. The concept of gWireless Brain-Inspired Computingh (WiBIC) was originally put forward by authors [1]. The WiBIC concept is that the SNN and the WSN are merged in term of eNetworkf. Because of the ewirelessf charac- teristic, it is possible to install the WiBIC anywhere. The WiBIC has no data server in the network and can acquire intelligence in the WSN itself. Additionally, low power consumption due to the spike signal processing was a ma- jor advantage. The SNN, whose neurons are connecting by infrared(IR) communication, was proposed and imple- mented [1]. However, the system had some problems in terms of the extensibility and the scalability. This paper proposes and implements the SNN with wire- less communications for Internet of Things (IoT). The im- plemented network could obtain the XOR function through reinforcement learning. By applying the wireless commu- nication for IoT, the SNN works with sufficient commu- nication distance and low power consumptions. It is easy to comprehend the fired neurons by broadcasting Medium Access Control header and Carrier Sense Multiple Access mechanism. |