In statistical pattern recognition, it is important to estimate the distribution of patterns precisely to achieve high recognition accuracy. In general, precise estimation of the parameters of the distribution requires a great number of sample patterns, especially when the feature vector obtained from the pattern is high-dimensional. For some pattern recognition problems, such as face recognition or character recognition, very high-dimensional feature vectors are necessary and there are always not enough sample patterns for estimating the parameters. In this paper, we focus on estimating the distribution of high-dimensional feature vectors with small number of sample patterns. First, we define a function, called simplified quadratic discriminant function (SQDF). SQDF can be estimated with small number of sample patterns and approximates the quadratic discriminant function (QDF). SQDF has fewer parameters and requires less computational time than QDF. The effectiveness of SQDF is confirmed by three types of experiments. Next, as an application of SQDF, we propose an algorithm for estimating the parameters of the normal mixture. The proposed algorithm is applied to face recognition and character recognition problems which require high-dimensional feature vectors.

The linearity and noise required by GSM and UMTS receivers make the design of a CMOS mixer for these applications so challenging. A new technique for IP2 improvement in Zero-IF active mixers is presented in this paper. This inductor-less technique is based on canceling the parasitic capacitor of common source node of the switching transistors and synthesizing resistive impedance. Using this technique, a reconfigurable down converter mixer works from 900 MHz to 2.4 GHz is designed supporting GSM, DCS, PCS, UMTS and IEEE802.11 b-g standards. The mixer IIP2 is higher than 71 dBm in GSM and UMTS bands. The mixer conversion gain is higher that 12 dB in all frequency bands. The design is done in 65 nm CMOS technology and consumes 10 mA from a 1.2 V supply. The design meets the performance required for all mentioned standards, while its area and power is comparable with high performance single band mixers.

Quality-of-service (QoS) is essential for multimedia applications, such as video-conferencing and voice over IP (VoIP) services, in wireless mesh networks (WMNs). A consequence of many clients accessing the Internet via the same backhaul is that throughput depends on the number of hops from the backhaul. This spatial bias problem is formulated as a mixed-integer nonlinear programming problem that considers end-to-end delay in terms of gateway selection, least-hop and load-balanced routing, and link capacity constraints. In this paper, we propose a routing algorithm for the network layer and a bandwidth allocation scheme for the medium access control (MAC) layer. The latter achieves fairness in both throughput and end-to-end delay in orthogonal mesh backbone networks with a distributed scheme, thereby minimizing the objective function. Our experiment results show that the proposed algorithm achieves throughput fairness, reduces end-to-end delay, and outperforms other general schemes and algorithms by at least 10.19%.

A fully integrated broadband up-conversion mixer with low power consumption is demonstrated on 90 nm CMOS technology in this paper. This mixer has a single-ended input and a multi-layer stacked Marchand balun is used for converting the differential output of the single-balanced mixer topology to a single-ended output. This balun employs inductive coupling between two metal layers and includes slotted shields to reduce substrate losses. The circuit size is 650 µm570 µm. At 22.1 GHz, the integrated mixer achieves a conversion gain of 2 dB with a maximum power dissipation of only 11.1 mW from a 1.2 V dc power supply at LO power of 5 dBm. Input referred 1-dB compression point is -14.8 dBm. The LO and RF return loss are better than 10 dB for frequencies between 20-26 GHz.

Development of ultrafast optical interfaces that can operate in sub-terahertz region is important to apply superconducting electronic devices to the high-end systems. We have performed several fundamental researches to realize the ultrafast optical input interface for superconducting electronic devices. Firstly, we observed optical response of amorphous Ge thin films, and the results indicated that an amorphous Ge photoconductive switch could stably operate in a terahertz frequency range as an optical-to-electrical signal converter in the low-temperature region below Tc of YBCO. Next, we have fabricated optical-to-electrical signal conversion system with photomixing technique, and we have demonstrated the generation and the detection of high frequency signals over 50 GHz. Finally, we have observed optical responses of a Josephson vortex flow transistor under irradiation of femtosecond laser pulses, and the results suggeste that the device has high potential as an optical interface.

A new approach for solution of the Tensor-Volume Integral Equation (TVIE) using Galerkin-based moment method (MoM) for three-dimensional dielectric bodies is proposed. Two problems of plane wave scattering by a dielectric sphere and a thin-wire antenna in close proximity to a dielectric body are investigated. In both cases, cubic modeling is applied and a combination of entire-domain and sub-domain basis functions, including three-dimensional polynomial functions with different degrees is utilized for field expansion inside dielectric bodies. Power polynomial is adopted for this purpose and its property is discussed over the proposed mixed-domain MoM formulation. Numerical examples show that based on the proposed method, a relative fast algorithm and suitable accuracy are achieved compared with conventional MoM. The accuracy of the proposed method is verified by comparing it with the Mie theory, conventional MoM and the FDTD method.

An effective way to boost power gain without noise figure degradation in a cascode low noise amplifier (LNA) is demonstrated at 4 GHz using 0.35 µm SiGe HBT technology. This approach maintains the same current consumption because a low-pass π-type LC matching network is inserted in the inter-stage of a conventional cascode LNA. 5 dB gain enhancement with no noise figure degradation at 4 GHz is observed in the SiGe HBT LNA with inter-stage matching.

In this paper, we have derived the new solution for the medium-frequency and the high-frequency ground wave propagation in a surface duct over mixed-paths. We have shown newly that the solution for the ground wave propagation in a standard atmosphere can be obtained directly from the solution for the surface duct problem by applying the analytic continuation from the negative equivalent radius of curvature of the earth to the positive one. Through the theoretical and experimental studies, it is confirmed that the radio wave propagating over the sea in the land-to-sea mixed-paths is enhanced by the recovery effect. It is clarified that the ground wave is also enhanced in the surface duct in a long range propagation. It is shown that the unexpected attenuation and the anomalous variation with distance are appeared in the propagation in the urban area due to the emergence of the slow-wave type trapped surface wave.

Simultaneous all-optical frequency up/downconversion technique utilizing a single semiconductor optical amplifier Mach-Zehnder interferometer (SOA-MZI) for full-duplex WDM radio over fiber (RoF) applications is presented. Using this technique, error-free simultaneous upconversion and downconversion of RoF signals with a finite-length single mode fiber were experimentally demonstrated. The results show the potential of the proposed scheme for use in a cost-effective full-duplex WDM RoF link.

A microwave photonic mixer utilizing an InGaAs photoconductor for radio over fiber applications is proposed and fabricated. Static and dynamic characteristics of the fabricated microwave photonic mixer were investigated. The microwave photonic mixer showed an optical bandwidth of approximately 300 MHz and a uniform conversion loss characteristic for the electrical input frequency up to 20 GHz.

A 10-GHz sub-harmonic Gilbert mixer is demonstrated in this paper using the 0.35 µm SiGe BiCMOS technology. The time-delay when the sub-harmonic LO (Local Oscillator) stage generates sub-harmonic LO signals is compensated by using fully symmetrical multiplier pairs. High RF-to-IF isolation and sub-harmonic LO Gilbert cell with excellent frequency response can be achieved by the elimination of the time-delay. The SiGe BiCMOS sub-harmonic micromixer exhibits 17 dB conversion gain, -74 dB 2LO-to-RF isolation, IP1 dB of -20 dBm, and IIP3 of -10 dBm. The measured double sideband noise figure is 16 dB from 100-kHz to 100-MHz because the SiGe bipolar device has very low 1/f noise corner.

A 5.2 GHz 1 dB conversion gain, IP1 dB = -19 dBm and IIP3= -9 dBm double quadrature Gilbert downconversion mixer with polyphase filters is demonstrated by using 0.35 µm SiGe HBT technology. The image rejection ratio is better than 47 dB when LO=5.17 GHz and IF is in the range of 15 MHz to 45 MHz. The Gilbert downconverter has four-stage RC-CR IF polyphase filters for the image rejection. Polyphase filters are also used to generate LO and RF quadrature signals around 5 GHz in the double quadrature downconverter.

Band connections employed in active cancellation circuits for effective reduction of digital substrate noise are proposed. An almost-odd-symmetrical noise characteristic is utilized for canceling out noises. Advancing this idea, interlaced connections of four bands are also proposed. Excess cancellation by those bands is more effective for noise reduction in a guard ring than a cancellation by two bands. Use of L-shaped bands on the basis of the interlaced connection suppresses the noise more. Simulation and experimental results show that the proposed band connections reduce the noise.

We propose here the first efficient publicly verifiable hybrid mix-net. Previous publicly verifiable mix-net was only efficient for short ciphertexts and was not suitable for mixing long messages. Previous hybrid mix-net can mix long messages but did not have public verifiability. The proposed scheme is efficient enough to treat large scale electronic questionnaires of long messages as well as voting with write-ins, and offers public verifiability of the correctness of the tally. The scheme is provably secure if we assume random oracles, semantic security of a one-time symmetric-key cryptosystem, and intractability of decision Diffie-Hellman problem. This paper is the full version of the extended abstract appeared in FC 2006 [10].

An even harmonic quadrature mixer (EH-QMIX) with a balanced configuration is proposed for a direct conversion receiver. The unit even harmonic mixer (EHMIX) used for I/Q paths consists of two anti parallel diode pairs (APDPs) and a pair of diplexers. When the second harmonic of LO (2LO) from the LO section is applied to the LO port as a spurious component, a conventional single-ended EHMIX using APDP converts the 2LO leakage from the LO section into the baseband and the d.c. offset and the self-detected LO noise arise at the baseband degrade the sensitivity. This proposed balanced EHMIX configuration can cancel out the 2LO leakage in itself. Therefore, the d.c. offset and the LO noise are significantly suppressed and the degradation of the sensitivity can be avoided. The suppression characteristic of the d.c. offset and the LO noise are verified by the simulation and the measurements. By using this balanced configuration, the fabricated EH-QMIX achieves wider frequency band characteristic than that of the single-ended EH-QMIX, and it shows 20% relative bandwidth at L-band.

This paper presents novel phase-continuous frequency hopping (FH) control for a direct frequency synthesizer (DFS) using a quadrature mixer driven by two direct digital synthesizers (DDSs). To achieve wideband FH in both of the lower and the upper sidebands of a local frequency in a quadrature mixer, the proposed DFS decreases or increases the phase of DDS output signals corresponding to frequency offset from a local frequency of the quadrature mixer. To realize phase decrement, the proposed method adds a complement number in a phase accumulator of a DDS, while a conventional DDS does not use phase decrement but uses a switchable combiner. In addition, as the phase accumulator output changes continuously by summing phase increment, the proposed method always assures phase continuity of a DFS output signal, which ends up suppressing sidelobe level of frequency hopped signals. The calculation and measurement results indicate that a sidelobe of a signal spectrum using the proposed phase continuous method is approximately 10 dB better than that using a conventional phase discontinuous method.

This paper proposes a wideband speech coder in which a G.711 bitstream is embedded. This coder has an advantage over conventional coders in that it has a high interoperability with existing terminals so costly transcoding involving decoding and re-encoding can be avoided. We also propose a partial mixing method that effectively reduces the mixing complexity in multiple-point remote conferences. To reduce the complexity, we take advantage of the scalable structure of the bitstream and mix only the lower band of the signal. For the higher band, the main speaker location is selected among remote locations and is redistributed with the mixed lower-band signal. By subjective evaluations, we show that the speech quality can be maintained even when the speech signals are partially mixed.

We extend the sliding block code in symbolic dynamics to transform J (≥2) sequences of Markov chains with time delays. Under the assumption that the chains are irreducible and aperiodic, we prove the central limit theorem (CLT) for the normalized sums of extended sliding block codes from J sequences of Markov chains. We apply the theorem to the system analysis of asynchronous spread spectrum multiple access (SSMA) communication systems using spreading sequences of Markov chains. We find that the standard Gaussian approximation (SGA) for estimations of bit error probabilities in such systems is the 0-th order approximation of the evaluation based on the CLT. We also provide a simple theoretical evaluation of bit error probabilities in such systems, which agrees properly with the experimental results even for the systems with small number of users and low length of spreading sequences.

We performed numerical analyses on structure sensitive field emission properties of our proposing plasmon resonant photomixer (PRX) in the terahertz range. The photomixer incorporates doubly interdigitated grating strips for gate electrodes and a vertical resonator structure for realizing highly efficient terahertz emission even at room temperature. We investigated the dependence of total field emission properties of PRX's on their material and dimension parameters. Introduction of low-conductive gate electrodes and ac-coupled 2D periodic plasmon gratings with depleted connecting portions are effective for expanding its lower cutoff frequency. The cutoff frequency, which is around 1.0 THz in standard metal-gates configuration, is expanded to less than 500 GHz. The output intensity could also be amplified more than double. On the other hand, a shorter vertical cavity is effective for expanding its upper cutoff frequency, which is expanded close to vertical resonant frequency, while maintaining the lower cutoff frequency. The combination of these design rules can realize much broader bandwidth operation.

Various mixed-signal very-high-speed integrated circuits have been developed using InP DHBTs. These circuits have been designed for fiber-optic 43 Gbit/s transmissions applications. They include: on the transmitting side, a clocked driver and an EAM driver, as well as a PSBT/DQPSK precoder; on the receiving side, a sensitive decision circuit, a limiting amplifier and an eye monitor. System experiments made possible by these circuits include a 6 Tbit/s transmission on >6000 km distance.