Iljin Lee

A 220–320-GHz Vector-Sum Phase Shifter Using Single Gilbert-Cell Structure With Lossy Output Matching

This paper presents a wideband vector-sum phase shifter (VSPS) that operates over the entire WR-3 band (220-320 GHz). Compared to conventional VSPSs with double Gilbert cells, the proposed phase shifter employs a single Gilbert-cell structure for vector modulation. This reduces the output current combining ratio from 8:2 to 4:2, and boosts the impedance at the combining node, thus facilitating wideband output matching at upper millimeter-wave and terahertz bands. The simplified structure leads to a reduction in dc power consumption and chip area without sacrificing the 360 ° phase-shifting property. Lossy matching is applied at the Gilbert-cell output to further increase bandwidth and stability at the expense of relatively high loss. The phase shifter is implemented using a 250-nm InP DHBT technology that provides f T and f max exceeding 370 and 650 GHz, respectively. The measurements exhibit a wideband phase shift with continuous 360 ° coverage and average insertion loss ranging from 11.8 to 15.6 dB for the entire WR-3 band. The root mean square amplitude and phase error among different phase states are less than 1.2 dB and 10.2 °, respectively. The input-referred 1-dB compression is measured at 0.7 dBm on average. The dc power consumption is 21.8-42.0 mW at different phase states.

H-band down-conversion and up-conversion mixers with wide IF bandwidth

This paper presents H-band down-conversion and up-conversion mixers implemented in a 250-nm InP double heterojunction bipolar transistor technology. The mixers are aimed to achieve a wide IF bandwidth and high conversion gain for high-speed wireless communication. The- mixer core employs a Gilbert cell pumped by a fundamental LO signal, leading to high conversion gain and high isolation. To achieve a wide IF bandwidth, the inductive-peaking, Cherry-Hooper, and staggered-tuned techniques are used at IF output of the down-conversion mixer. The down- and up-conversion mixers exhibit measured conversion gain of 7.5 and -5.2 dB with 3-dB SSB IF bandwidth of 20 and 25 GHz at LO frequency of 270 and 280 GHz, respectively.

PAPR Analysis of the OFDMA and SC-FDMA in the Uplink of a Mobile Communication System

In recent years, OFDMA(orthogonal frequency division multiple access) and SC-FDMA(Single Carrier Frequency Division Multiple Access) have been widely studied for the uplink of a mobile communication system. In this paper, PAPR(Peak-to-Average Power Ratio) and BER(Bit Error Rate) performance of the OFDMA and SC-FDMA systems are studied in relation to the uplink of a mobile communication system. Three kinds of sub-carrier allocation methods in the OFDMA system and 2 kinds of sub-carrier allocation methods in SC-FDMA system are suggested to compare and improve system performance. Simulation results show that in the OFDMA system, the first sub-band allocation method has better PAPR reduction performance than the other methods. In the SC-FDMA system, the distributed allocation method offers similar PAPR, compared with the sub-band allocation method. PAPR can be further reduced by adding a spectrum shaping filter with an appropriate roll of factor. Furthermore, it is found that on average, SC-FDMA can reduce the PAPR by more than 5 dB compared to OFDMA, when the total sub-carrier number is 1,024 and the sub-carrier number allocated to each user changes from 8 to 512. Because of the frequency diversity and low PAPR characteristics, SC-FDMA system of the distributed sub-carrier allocation method can achieve better BER performance than the OFDMA system.

300 GHz InP rectangular cavity antenna

A rectangular cavity antenna design for terahertz integrated circuits is presented. The antenna is built in a 75 μm-thick InP substrate with a dielectric constant of 12.5. Thick substrates often cause unwanted substrate modes that severely hamper radiation efficiency. Our proposed cavity antenna design, however, overcomes the substrate mode problems and utilizes the dominant TE101 mode to maintain better than 70 percent of radiation efficiency across 41 GHz bandwidth at 300 GHz.

Performance Evaluation of FD-CI-OFDM System with PAPR Reduction and Frequency Diversity Effects

Orthogonal frequency division multiplexing(OFDM) that is very useful for the high-speed communication system has serious problem of high peak-to-average power ratio(PAPR) in time domain. Because of this, the non-linear distortion can be produced and system performance gets worse. CI-OFDM system can get the peak power lowered. In this CI-OFDM system, each parallel data is distributed into N all sub-carriers and conveyed by the orthogonal phase factor. Also, CI-OFDM shows frequency diversity effect. Therefore, CI-OFDM system is better than ordinary OFDM system in terms of BER performance and the PAPR reduction. When it is implemented, however, there is a serious problem whether it can separate and compensate the phase factor in order in the receiver. Because all bits are transmitted simultaneously over all subcarriers and each other phase factors in transmitter. In this paper, we propose and evaluate a FD-CI-OFDM system that is a version of improved CI-OFDM. This is designed by use of the Walsh Hadamard sequence. The FD-CI-OFDM shows better performance than ordinary OFDM and CI-OFDM system.

Design and Performance Analysis of Pre-Distorter Including HPA Memory Effect

OFDM(Orthogonal Frequency Division Multiplexing) signals suffer serious nonlinear distortion in the nonlinear HPA(High Power Amplifier) because of high PAPR(Peak Average Power Ratio). Nonlinear distortion can be improved by a pre-distorter, but this pre-distorter is insufficient when the PAPR is very high in an OPFDM system. In this paper, a DFT(Discrete Fourier Transform) transform technique is introduced for PAPR reduction. It is especially important to consider the memory effect of HPA for more precise predistortion. Therefore, in this paper, we consider two models, the TWTA(Traveling-Wave Tube Amplifier) model of Saleh without a memory effect and the HPA memory polynomial model that has a memory effect. We design a pre-distorter and an adaptive pre-distorter that uses the NLMS(Normalized Least Mean Square) algorithm for the compensation of this nonlinear distortion. Without the consideration of a memory effect, the system performance would be degraded, even if the pre-distorter is used for the compensation of the nonlinear distortion. From the simulation results, we can confirm that the proposed system shows an improvement in performance.