Tuesday, 31 May 2016: 14:50
Aqua 307 (Hilton San Diego Bayfront)
M. Fujishima (Hiroshima University)
Over hundred years have passed since Marconi sent the first Atlantic wireless transmission in 1901. Over the years, frequency used for wireless communication has been increasing by ten folds every 20 years. According to the trend, frequency used for wireless communication will reach at terahertz band in 2020. In the research on terahertz wireless communication, transceivers with compound semiconductor were reported[1-2]. However, integration of large-scale digital circuits in a compound semiconductor is so difficult that complex modulation and demodulation for high-speed communication and frequency-domain equalization including propagation environment are unrealistic. On the other hand, although CMOS integrated circuits can realize complex modulation and frequency domain equalization, terahertz transceivers with CMOS are challenging since typical maximum operation frequency in advanced process is still around 300GHz. To overcome relatively low operation frequency with CMOS transistors, we have adopted a tripler in a transmitter, where the tripler utilizes cubic nonlinearity in a transistor[3]. Generally, when modulated signal is given to the tripler, its carrier frequency triples but original constellation is distorted. Therefore, in [4], the tripler was used only for QPSK (quadrature phase shift keying) where the amplitude of symbols is constant and the shape of constellation is not affected by tripler’s nonlinearity. Nevertheless, in order to realize higher communication speed with more bits per symbol such as 16 QAM or higher, how should the tripler be utilized? To solve the nonlinearity issue in the tripler, local oscillation signal (LO) as well as the modulated intermediate-frequency signal (IF) is given to the tripler simultaneously. When the two types of spectral components are given to the tripler, intermodulation between two signals occurs. As a result, spectral components of appear at the output. Since the second term maintains linearity for the IF, modulation information in the IF is preserved when only the second term is properly filtered. Adopting this technique, 300GHz-band CMOS transmitter is fabricated with 40nm process. Measured results show that 17.5Gbps per channel is realized over 6 channels with 32 QAM. As a result, total bit rate reaches at 105Gbps when all the 6 channels can be bonded.
[1] H-J. Song et al., "50-Gb/s Direct Conversion QPSK Modulator and Demodulator MMICs for Terahertz Communications at 300 GHz," IEEE Trans. Microwave Theory & Tech., vol. 62, no. 3, pp. 600 - 609, March 2014.
[2] C. Wang et al., "0.34-THz Wireless Link Based on High-Order Modulation for Future Wireless Local Area Network Applications ," IEEE Trans. Terahertz Science & Tech., vol. 4, no. 1, pp. 75 - 85, Jan. 2014.
[3] K. Katayama et al., "A 300GHz 40nm CMOS Transmitter with 32-QAM 17.5Gb/s/ch Capability over 6 Channels," 2016 IEEE International Solid-State Circuits Conf. Dig. (in press).
[4] S. Kang et al., "A 240GHz wideband QPSK transmitter in 65nm CMOS," 2014 IEEE Radio Frequency Integrated Circuits Symp., pp. 353 - 356, 1-3 June 2014.