Low-power technology for front-end GaAs ICs and hybrid IC (HIC) for a mobile communication equipment will be presented. For low-power operation of GaAs front-end ICs, new techniques of the intermediate tuned circuits, the single-ended mixer, dualgate MESFETs, and the asymmetric self-aligned LDD process were investigated. The designed down-converter IC showed conversion gain of 21 dB, noise figure of 3.5 dB, 3rd-order intercept point in output level (IP3out) of 4.0 dBm, image-rejection ratio of 20 dB at 880 MHz, operating at 3.0 V of supply voltage and 5.0 mA of dissipation current. The down-converter IC was also designed for 1.9 GHz to obtain conversion gain of 20 dB, noise figure of 4.0 dB, IP3out of 4.0 dBm, image-rejection ratio of 20 dB at 3.0 V, 5.0 mA. The up-converter IC was designed for 1.9 GHz using the same topology of circuit and showed conversion gain of 15 dB, IP3out of 7.5 dBm, and 1 dB compression level of -8 dBm with -20 dBm of LO input power, operating at 3.0 V, 8.0 mA. Another approach to the low-power operation was carried out by HIC using the GaAs down-converter IC chip. The HIC was designed for 880 MHz to show conversion gain of 27 dB, noise figure of 3.3 dB, IP3out of 3.0 dBm, image-rejection ratio of 12 dB, at 2.7 V, 4.5 mA. The HIC measures only 8.0 mm6.0 mm1.2 mm.

A three-dimensional micromagnetic simulation using the Landau-Lifshitz-Gilbert equation was performed for thin-film magnetic recording media and magnetoresistive (MR) heads with soft adjacent layers (SAL). For recording media the simulation results for magnetization curves and media noise were compared with the results of experiments. Although the media model needs to be improved, the qualitative agreement between simulation results and experimental results shows that this micromagnetic simulation can be a useful tool for analyzing and predicting magnetic properties and recording characteristics. This work also showed that media noise is influenced by magnetostatic interaction, and that the decrease of the magnetostatic interaction is favorable for obtaining a high signal-to-noise ratio. For an MR head the output obtained with a nonuniform sense current distribution is similar to the output obtained with uniform sense current distribution for both low and high anisotropy fields (Hk=2 Oe and 10 Oe) SAL. With the low Hk SAL, however, the asymmetry of the output obtained for nonuniform sense current differs from the asymmetry obtained for uniform sense current; the difference is due to a magnetization vortex in a biased state in the SAL. With the high Hk SAL, the difference between the asymmetry obtained for nonuniform sense current and the one obtained for uniform sense current is not large; no vortices are found in the SAL at the biased state.

Neuromorphic computing with a spiking neural network (SNN) is expected to provide a complement or alternative to deep learning in the future. The challenge is to develop optimal SNN models, algorithms, and engineering technologies for real use cases. As a potential use cases for neuromorphic computing, we have investigated a person monitoring and worker support with a video surveillance system, given its status as a proven deep neural network (DNN) use case. In the future, to increase the number of cameras in such a system, we will need a scalable approach that embeds only a few neuromorphic devices in a camera. Specifically, this will require a shallow SNN model that can be implemented in a few neuromorphic devices while providing a high recognition accuracy comparable to a DNN with the same configuration. A shallow SNN was built by converting ResNet, a proven DNN for image recognition, and a new configuration of the shallow SNN model was developed to improve its accuracy. The proposed shallow SNN model was evaluated with a few neuromorphic devices, and it achieved a recognition accuracy of more than 80% with about 1/130 less energy consumption than that of a GPU with the same configuration of DNN as that of SNN.