Beamforming for impulse-radio UWB communication systems based on complex-valued spatio-temporal neural networks

Ultra-wideband (UWB) wireless communications based on impulse radio uses so short pulses that its occupied bandwidth is very large and the density of spectral power is low. Then, UWB has high confidentiality and low interference to other narrowband wireless communications. However, in the situation of multiple access and fixed transmission-speed, communication quality of UWB becomes low by interference from other UWB nodes. In high-speed UWB communications, where symbol period is short and pulse density is high, the communication quality of UWB is also low by multipath interference. In this paper, we reduce the degradation of communication quality by using a complex-valued spatio-temporal neural network beamformer (CVST-NN-BF). We evaluate bit error rate (BER) of UWB wireless communications based on impulse radio in multipath environment.