A novel millimeter-wave on-wafer CAT(Computer-Aided-Testing ) system has been developed for measurement of S-parameters and NF ( Noise figure ). For the S-parameter test system, we have developed a holder setup and installed it in a semi-automatic wafer prober so that the waveguide-based T/R module can be directly connected to a probe-head through fixed waveguides, which feature low insertion loss of less than 2 dB, from 75 GHz to 98 GHz. The accuracy of the developed test system was confirmed by measuring, with this system, a co-planar offset short pattern then comparing measured and simulated results. A good agreement between the measured and calculated, in both return loss and return phase successfully demonstrated the superiority of the system. A W-band NF test system with a system noise of less than 8 dB has been also developed to provide an on-wafer NF measurement capability with an accuracy of 0.3 dB. These S-parameter and NF test systems possess great advantages to achieve high-speed automatic MMIC testing up to W-band.

This paper describes numerical analyses of resistive mixer operation, followed by measured performances of a V-band (50 - 75 GHz) monolithic InP HEMT resistive mixer operable with a very low LO power. Our model assumes that the channel conductance of the InP HEMT can be described by three linear functions according to the applied gate voltage. The calculated results obtained with the model have shown that the LO power level required for mixer operation is determined by the gate bias voltage and that a device with abrupt conductance shifts is suited to low LO power operation for a resistive mixer. It is also shown that conversion loss saturation of a resistive mixer is caused by its channel conductance saturation. A V-band monolithic resistive mixer has been designed and fabricated using Coplanar Waveguides (CPW) and a 0.15 mm InP HEMT with abrupt channel shifts. Good agreement between measured and simulated conversion losses are obtained. A minimum conversion loss of 8.4 dB is achieved at the 55 GHz RF-frequency and the -2 dBm LO power. It also exhibits an excellent IF output linearity to allow the 1 dB compression RF input level to be comparable with LO power, indicating good intermodulation performance. It is demonstrated that the proposed simple model of the channel conductance can easily calculate conversion characteristics of a resistive mixer with high accuracy.

Several grammars of which generative power is between context-free grammar and context-sensitive grammar were proposed. Among them are macro grammar and tree adjoining grammar. Multiple context-free grammar is also a natural extension of context-free grammars, and is known to be stronger in its generative power than tree adjoining grammar and yet to be recognizable in polynomial time. In this paper, the generative power of several subclasses of variable-linear macro grammars and that of multiple context-free grammars are compared in details.

This paper describes 2.4-GHz-band front-end building circuits--a down conversion mixer (DCM), a dual-modulus divide-by-4/5 prescaler, a transmit/receive antenna switch (SW), a power amplifier (PA), and a low noise amplifier (LNA). They are fabricated using a standard bulk 0.18 µm CMOS process with a lower current consumption than bipolar circuits, and can operate at the low supply voltage of 1.8 V. Meshed-shielded pads are adopted for lower receiver circuit noise. Pads shielded by metals become cracked when they are bounded, therefore silicided active areas are used as shields instead of metals to avoid these cracks. The meshed shields achieve lower parasitic pad capacitors without parasitic resistors, and also act as dummy active areas. The proposed DCM has a high IP3 characteristic. The DCM has a cascode FET configuration and LO power is injected into the lower FET. By keeping the drain-source voltage of the upper transistor large, the nonlinearity of the drain-source transconductance is reduced and a low distortion DCM is realized. It achieves a higher input referred IP3 with a higher conversion gain for almost the same current consumption of a conventional single-balanced mixer. The output referred IP3 is higher 5.0 dB than the single-balanced mixer. The proposed dual-modulus prescaler employs a fully-differential technique to achieve stable operation. In order to avoid errors, the fully-differential circuit gives the logic voltage swing margins. In addition, the differential technique also reduces the noise effect from the supply voltage line because of the common-mode signal rejection. The maximum operating frequency is 3.0 GHz, and the one flip-flop power consumption normalized by the maximum operating frequency is 180 µW/GHz.

Several grammars have been proposed for representing RNA secondary structure including pseudoknots such as simple linear tree adjoining grammar (sl-tag), extended sl-tag (esl-tag) and RNA pseudoknot grammar (rpg). The main purpose of this paper is to compare the generative power of these grammars by identifying them as subclasses of multiple context-free grammars (mcfg). Specifically, it is shown that the class of languages generated by esl-tag (ESL-TAL) properly includes the class of languages generated by sl-tag (SL-TAL) and the class of languages generated by cfg. Also, we show that the class of languages generated by rpg coincides with the class of languages generated by mcfg with dimension one or two and rank one or two. Furthermore, it is shown that SL-TAL is a full trio and ESL-TAL is a substitution closed full AFL.

Structured documents often contain character strings of which semantics can be naturally stored as database values or has direct correspondence with database values. By building bilateral logical links between character strings in documents and corresponding database values, semantically rich queries are made expressible. We have introduced a new ADT, named "paratext," to model text which has links with database values. Paratexts are logically viewed as consisting of two parallel layers; on the "appearance" layer, ordinary text (i. e. a linear sequence of character strings) is placed, while the "reference" layer holds an array of OIDs and literals. Each OID or literal on the reference layer is associated with a contiguous substring of the appearance layer text, and represents the semantics of the associated substring. We have also designed domain-specific functions for this document model. Using the functions, we can express queries which go back and forth between the two layers. In structured documents, such character strings can appear in the whole content of logical elements, or as phrases inside logical elements. We also present frameworks for the implementation of the paratext ADT, and discuss how traditional full-text indexing techniques can be extended to support paratext.

Multi-band WDM transmission beyond the C+L-band is a promising technology for achieving larger capacity transmission by a limited number of installed fibers. In addition to the C- and L-band, we can expect to use the S-band as the next band. Although the development of optical components for new bands, particularly transceivers, entails resource dispersion, which is one of the barriers to the realization of multi-band systems, wavelength conversion by transparent all-optical signal processing enables new wavelength bandtransmission using existing components. Therefore, we proposed a transmission system including a new wavelength band such as the S-band and made it possible to use a transceiver for the existing band by performing the whole-band wavelength conversion without using a transceiver for the new band. As a preliminary verification to demonstrate multi-band WDM transmission including S-band, we investigated the application of a novel wavelength converter between C-band and S-band, which consists of periodically poled lithium niobate waveguide, to the proposed system. We first characterized the conversion efficiency and noise figure of the wavelength converter and estimated the transmission performance of the system through the wavelength converter. Using the evaluated wavelength converters and test signals of 64 channels arranged in the C-band at 75-GHz intervals, we constructed an experimental setup for S-band transmission through an 80-km standard single-mode fiber. We then demonstrated error-free transmission of real-time 400-Gb/s DP-16QAM signals after forward error correction decoding. From the experimental results, it was clarified that the wavelength converter which realizes the uniform lossless conversion covering the whole C-band effectively achieves the S-band WDM transmission, and it was verified that the capacity improvement of the multi-band WDM system including the S-band can be expected by applying it in combination with the C+L-band WDM system.

One of cost-effective ways to increase the transmission capacity of current standard wavelength division multiplexing (WDM) transmission systems is to use a wavelength band other than the C-band to transmit in multi-band. We proposed the concept of multi-band system using wavelength conversion, which can simultaneously process signals over a wide wavelength range. All-optical wavelength conversion could be used to convert C-band WDM signals into other bands in a highly nonlinear fiber (HNLF) by four-wave mixing and allow to simultaneously transmit multiple WDM signals including other than the C-band, with only C-band transceivers. Wavelength conversion has been reported for various nonlinear waveguide materials other than HNLF. In such nonlinear materials, we noticed the possibility of wideband transmission by dispersion-tailored silicon-on-insulator (SOI) waveguides. Based on the CMOS process has high accuracy, it is expected that the chromatic dispersion fluctuation could be reduced in mass production. As a first step in the investigation of the broadness of wavelength conversion using SOI-based waveguides, we designed and fabricated dispersion-tailored 12 strip waveguides provided with an edge coupler at both ends. Each of the 12 waveguides having different widths and lengths and is connected to fibers via lensed fibers or by lenses. In order to characterize each waveguide, the pump-probe experimental setup was constructed using a tunable light source as pump and an unmodulated 96-ch C-band WDM test signal. Using this setup, we evaluate insertion loss, input power dependence, conversion bandwidth and conversion efficiency. We confirmed C-band test signal was converted to the S-band and the L-band using the same silicon waveguide with 3dB conversion bandwidth over 100-nm. Furthermore, an increased design tolerance of at least 90nm was confirmed for C-to-S conversion by shortening the waveguide length. It is confirmed that the wavelength converters using the nonlinear waveguide has sufficiently wide conversion bandwidth to enhance the multi-band WDM transmission system.

Recently, the Doherty amplifier technique has been the focus of attention not only for base stations but also for mobile terminals because of its high power-added efficiency in the large back-off region. In this paper, we present a miniaturized Doherty power amplifier (PA) module for W-CDMA mobile terminals. The developed Doherty PA module consists of a 4-mm-square ceramic substrate (4.0 mm4.0 mm1.5 mm, alumina, dielectric constant = 8.8), a 1-mm-square GaAs MMIC (1.0 mm1.0 mm0.1 mm), and 0603-size SMD passive components. To miniaturize the module size, the optimal designed quarter-wavelength transmission lines, which are used for impedance conversion for carrier amplifier output and phase compensation for peak amplifier input, are embedded in the ceramic module substrate. Two GaAs HBTs for a carrier amplifier and a peak amplifier and base bias circuits for each amplifier are integrated onto a single-chip GaAs MMIC. Measurement results at 1950 MHz in a W-CDMA uplink signal indicate that 27 dBm of the maximum output power, 45% of the power-added efficiency (PAE), 11 dB of power gain, and 43% of PAE at 6 dB back-off, i.e. 24 dBm output power, are obtained with the developed Doherty PA. In other words, the PAE is improved from the theoretical PAE of a conventional class B amplifier, namely, from 23% to 43%. This is the smallest Doherty amplifier developed in the form of a module for mobile terminals.

This paper presents a miniaturized dual-mode Doherty PA module applicable for an HPSK signal and an OFDM 64-QAM signal. Dual-mode operation with identical hardware is realized by introducing a bias switching technique, which changes bias conditions of amplifiers according to transmission signals, and employing dual-mode matching circuits, which are designed based on the results of load-pull measurements using an HPSK signal and an OFDM 64-QAM signal. The Doherty PA module consists of a Doherty stage and a gain stage. Two GaAs-HBTs for a Doherty stage and one GaAs-HBT for a gain stage are integrated onto a 1 mm-square single GaAs-MMIC. In the HPSK mode, maximum output power of 26.7 dBm, power added efficiency (PAE) of 41%, and power gain of 27 dB are obtained in the condition that adjacent channel leakage power ratio (ACLR) is under -38 dBc. In the OFDM 64-QAM mode, maximum output power of 21.0 dBm, PAE of 27%, and power gain of 28 dB are obtained under EVM < 3.0%. This is the first multi-mode Doherty PA module suitable for multi peak to average power ratio (PAPR) signals.

Recently, dynamic power supply voltage techniques, such as an Envelope Elimination and Restoration power amplifier (EER-PA) or Envelope-Tracking Power amplifier (ET-PA), have been attracting much attention because they can maintain high efficiency in large back-off region [1]-[6]. The dynamic power supply voltage techniques cause strong nonlinearity compared to a conventional power amplifier, hence a memoryless Digital Predistortion (DPD) technique is indispensable for these efficiency enhancement techniques. However, the performance of the memoryless DPD is degraded due to the frequency response of the envelope amplifier in the dynamic power supply voltage techniques [7]-[9]. In this paper, we clarify the degradation mechanisms of the memoryless DPD for the EER-PA due to the frequency response of the envelope amplifier based on the results of two-tone tests, and propose an analytical model for improving the performance of the memoryless DPD developed for the EER-PA. In addition, a prototype EER-PA is developed and we demonstrate that the residual distortion of the developed EER-PA with conventional memoryless DPD algorithm is compensated by the new algorithm based on the proposed analytical model. In the two-tone test, third-order intermodulation distortion (IMD3) with a tone spacing from 100 kHz to 4 MHz is improvement by up to 25 dB by the memoryless DPD algorithm based on the proposed model. Measured adjacent channel leakage power ratio (ACPR) of the developed EER-PA is improved from -22.5 dBc to -42.5 dBc in the OFDM signal test with 1.08 MHz bandwidth.

A 0.15 µm T-shaped gate AlInAs/InGaAs high electron mobility transistor (HEMT) with an excellent RF performance has been developed using selective wet gate recess etching. The gate recess is formed by a pH-adjusted citric acid/NH4OH/H2O2 mixture with an etching selectivity of more than 30 for InGaAs over AlInAs. The standard deviation of saturation drain current (Idss) is as small as 3.2 mA for an average Idss of 47 mA on a 3 inch diameter InP wafer. The etching time for recess formation is optimized and an ft of 130 GHz and an MSG of 10 dB at 60 GHz are obtained. The extremely low minimum noise figure (Fmin) of 0.9 dB with an associated gain (Ga) of 7.0 dB has been achieved at 60 GHz for a SiON-passivated device. This noise performance is comparable to the lowest value of Fmin ever reported for an AlInAs/InGaAs HEMT with a passivation film.

A Q-band high gain and low noise Variable Gain Amplifier (VGA) module using dual gate AlGaAs/InGaAs pseudomorphic HEMTs has been developed. The dual gate HEMT can be fabricated by the same process of the single gate HEMT which has the gate length of 0.15 µm. The Q-band VGA module consists of a 1-stage low noise amplifier (LNA) MMIC using a single gate HEMT and a 2-stage VGA MMIC using dual gate HEMTs. During the design, an accurate noise modeling is introduced to achieve low noise performance. A fully passivated film is employed to achieve reliability. The VGA module has a gain of more than 20 dB from 41 GHz to 52 GHz and a maximum gain of 24.5 dB at 50 GHz. A gain control range of more than 30 dB is achieved in the same frequency range. A phase deviation is less than 10 degrees in 10 dB gain control range. A minimum noise figure of 1.8 dB with an associated gain of 22 dB is achieved at 43 GHz and the noise figure is less than 2.5 dB with associated gain of more than 20 dB from 41 GHz to 46 GHz when biased for low noise figure. This performance is comparable with the best data ever reported for LNAs at Q-band including both GaAs based HEMTs and InP based HEMTs.

We developed a flexible-logic-gate single-electron device (SED) in which logic functions can be selected by changing the voltage applied to the control gate. It consists of an array of nanodots with multiple inputs and multiple outputs. Since the gate electrodes couple capacitively to the many dots underneath, complicated characteristics depending on the combination of the gate voltages yield a selectable logic gate when some of the gate electrodes are used as control gates. One of the important issues is how to design the arrangement of nanodots and gate electrodes. In this study, we fabricated a Si nanodot array with two simple input gates and two output terminals, in which each gate was coupled to half of the nanodot array. Even though the device had a very simple input-gate arrangement and just one control gate, we could create a half-adder function through the use of current maps as functions of the input gate voltages. We found that the nanodots evenly coupled capacitively to both input gates played an important role in getting a basic set of logic functions. Moreover, these results guarantee that a more complicated input-gate structure, in which each gate evenly couples many nanodots, will yield more complicated functions.

An ultra low noise 50-GHz-Band amplifier (LNA) MMIC has been developed using an AlGaAs/InGaAs pseudomorphic HEMT. A noise figure of 1.8 dB with an associated gain of 8.1 dB is achieved at 50 GHz. The noise figure is less than 2.0 dB from 50 GHz to 52.5 GHz. This is the state-of-the-art noise figure for low noise amplifiers around 50 GHz. The success of this LNA development came from the excellent HEMT and MMIC technologies and the accurate modeling of active and passive elements. Good agreement between measured and simulated data over the band from 40 GHz to 60 GHz is obtained.

Millimeter-wave monolithic low noise amplifier modules using 0.15 µm AlGaAs/InGaAs/GaAs pseudomorphic HEMTs have been developed at V- and W-bands for the Advanced Microwave Scanning Radiometer. To achieve low noise and high gain of V-band single-stage and W-band two-stage monolithic amplifiers, a reactive matching method is employed in the design of input noise matching and output gain matching circuits based on the results of on-carrier S-parameter measurements up to 50 GHz and noise parameter measurements at 60 and 90 GHz. A V-band four-stage monolithic amplifier module has been mounted on a hermetically-sealed package with microstrip interface and has achieved a noise figure of 3 dB with a gain of 42.2 dB at 51 GHz. A W-band six-stage amplifier module has been mounted on a hermetically-sealed package with waveguide interface and has achieved a noise figure of 4.3 dB with a gain of 28.1 dB at 91 GHz. These results represent the best noise figure performance ever achieved by multi-stage monolithic low-noise amplifier modules.

This paper describes two kinds of on-chip matched low-noise/driver MMIC amplifiers (LN/D-As) suitable for 2.4-GHz and 5.2-GHz short-range wireless applications. The ICs are fabricated in a 0.18 µm bulk CMOS which has no extra processing steps for enhancing the RF performance. The successful use of the current-reuse topology and interdigitated capacitors (IDCs) enables sufficiently low-noise and high output power operations with low current dissipation despite the chip fabrication in the bulk CMOS leading to large RF substrate and conductor losses. The main measurement results of the two LN/D-As are as follows: a 3.8-dB noise figure (NF) and a 10.1-dB gain under the conditions of 1.8 V and 6 mA, a 3.4-dBm 1-dB gain compressed output power (P1dB) for a 2.4-V voltage supply and a 13-mA operating current for the 2.4-GHz LN/D-A, and a 4.9-dB NF and an 11.1-dB gain with a 1.8 V and 10 mA supply condition, a 2.3-dBm P1dB at 2.4 V and 16 mA for the 5.2-GHz LN/D-A. Both MMICs are suited for low-noise amplifiers and driver amplifiers in 2.4-GHz and 5.2-GHz low-cost, low-power wireless systems such as Bluetooth and hiperLAN.

A new stuffing-pulling fusion technique for the mass fusion splicing of single-mode fibers is presented, where the fiber spliced-region is pulled after the fiber end faces are fused and stuffed to each other. The characteristic feature of this technique is that the stuffing and pulling strokes are varied, based on the observation of fiber end face gaps for every splice. This technique can reduce splice loss to a half in comparison with the conventional method.

An image type resonator method is proposed as a method to evaluate precisely the temperature dependence of dielectric material. At first, the temperature coefficients of the resonant frequencies, TCf are measured separately using the shielded dielectric resonators of three types; that is a parallel plate type, and an image type, and a MIC type resonator. Secondly, an intrinsic temperature coefficient of the resonant frequency TCf0, which is defined as the temperature coefficient of a resonant frequency when all the stored energy is confined inside a dielectric, is estimated from these measured TCf. Actually, the TCf0 values of a sapphire and (ZrSn) TiO4 rod are estimated from the TCf values measured for the resonators of three types. As a result, for the parallel plate type, the precision of TCf0 is about 0.1 ppm/. For the image and MIC types, the errors of about 0.5 ppm/ in the TCf0 values arise from the errors in the linear expansion coefficients of the resonators, rather than from the experimental errors in TCf. Then, another image type resonator is designed to estimate TCf0 within error of 0.1 ppm/. In this design, dimensions of the shielding cavity is determined to reduce the influence of the errors in the linear expansion coefficients on precision of the TCf0 estimation. Finally, for a (ZrSn) TiO4 ceramic rod, a TCf0 value estimated from TCf measured for the image type resonator is obtained with accuracy of about0.1 ppm/.