In this paper, we propose a new method for the improvement of linearizing and adaptive convergence charateristics of the feedforward linear power amplifier. In this circuit, errors at the signal cancellation stage can be compensated at the error cancellation stage and the overall linearizaton and adaptation characteristics of the linear amplifier are improved. The broadband characteristics and linearizing capability are improved without increasing the complexity of circuits and the signal processing structure.

By noticing that most waveguide devices were connected to power source through coaxial line, three dimentional finite element analysis using TEM mode incident was tried. When tEM mode was used, being dissimilar to TE$_{10}$ mode incident, analysis model of TEM mode incident has almost the same geometry with real products. Therefore, the degree of trust about analysis result is improved. Reflection coefficients of coaxial line, W/G adaptor, magnetron obtained by simulation was compared with experimental results.

We have successfully fabricated high power (~20W) laser diode array, which are useful for pumping Nd:YAG lasers. The laser diode aray has 20 100.mu.m-wide cahnnels of which space was adjusted to 350.mu.m to improve thermal characteristics. And channel width is 100.mu.m. For an uncoated LD array, the output power of 15.66W has been obtained at 41A under quasi-CW operation, which results in about 0.42W/A slope efficiency. After aR(5%) and HR (95%) coatings on both facets, the output power was improved up to 21.18W at 40A under the same operation as above and the slope efficiency was 0.795W/A. On the other hand, by using a near field measurement system consisting of objective lens, eyepiece, CCD camera and image processing board, the typical near field patten of 1*20 LD array was observed.

AlGaAs/GaAs HBTs are developed well enough to be commercialized as an active device in optical transmission system, but there remains the unanswered questions about reliability. In this paper we applied the reverse constant current stress at the high voltage in avalanche region for a long time to find out a new degradation mechanism of junctrion I-V. The unction off-set voltage at which the current vanishes to zero was shifted to the negative direction of applied bias due to the increment of leakage current as the stress time increases. It was identified that the degradation was induced by the hot carriers which were generated at space charge region and trapped at the interface between GaAs base and the passivation nitride enhancing the electric field across the nesa edge.

Utilizing 2-dimensional device simulations incorporating lattic eheating models, we analyzed in detail the DC breakdown characterisics of NMOS trasistors with different structures, which are commonly used as ESD protection transistors. The mechanism leading to device failure resulting from electrostatic discharge was explained by analyzing the 1st and 2nd breakdown characteristics of LDD devices. Also a criteria for more robust designs of NMOS transistor structures against ESD was suggested by examining the characteristics changes with changes in structural parameters such as the LDD doping concentration, the drain junction depth, the distance between source/drain contacts, and the source junction area.

Volume hologram was recorded using reference beam form optical fiber taper. A singel mode fiber was chemically etced to make a taper structure, and we showed experimentally that the referencing by the irregular beam pattern from this taper structure could increase the storage density of photorefractive volume hologram. The spatial selectivity of the volume hologram with this method was increased by two times compared to the normal single mode fiber referencing case. A theoretical analysis with randomly phased plane model also confirmed the results.

The 1-D finite-element mehtod is applied for designing double-clad otpical fibers with dispersion-compensation capability. design parameters allowing only a $LP_{01}$ single mode are treated and macro-bending loss are taken into consideration. Design parameters are extracted to have the compensation ratio $(L_{SMF}/L_{DCF})$ of 6.2 for core structure with step-index profile, and of 5.2 for core with triangular-index profile.

We have shown that the Hough transform filter can be easily obtained by repeatedly recording one line segment pattern holographically by use of rotational multiplexing. With this filter, one can get resutls of the hough transform in the time of light propagation. To demonstrate our method experimentally, the hough transform filter for 18 different rottion angles was recorded, and experimental transform results were compared with simulated ones for a few input patterns. By introducing a compensation process for the non-uniformity of both the input plane wav eand the diffraction efficiency of each hologram. It is possible to get more precise transform results.

a wavelength division multiplexer based on a polymeric arrayed-waveguide grating has been designed and fabricated. A 4-channel multiplexer with a spacing of 3.2 nm is designe dby using te 2-dimensional beam propagation method. A UV-curable epoxy, NOA73, is used for the core layer, and a passive polymer, PMMA, for the cladding layer. The polymer waveguides are fabricated by the reactive ion etching method and their optical properties are characterized. The fabricted device has a center wavelength of 1548.3 nm, and the wavelength spacing between the channels is 3.2nm. The measured crosstalk is better than -18dB.

The characteristics of optical waeguides and IDT electrodes fabricated for acousto-optic tunable filters (AOTE) used for optical communications were analyzed. A $Ti:LiNbO_3$ in-diffusion method was employed for the formation of the optical waveguide with a dimension of width $8\mu\textrm{m}$, length $30000\mu\textrm{m}$, and thickness $1150{\AA}$. The diffusion was carried at $1050^{\circ}C$ for 8 houss to pattern the optical waveguide with Ti. The resulted waveguide exhibited a single mode at a 1550nm wavelength range and its propagation loss was less than 0.5dB/cm. The width of IDT, with 10 SAW periods, was $5000\mu\textrm{m}$, S11 reflection characteristics and impedances of th eelectrodes deposited with Au were analyzed using a network analyzer; $48.1\Omega$ at th ecenter frquency of 193MHz for Au deposition thickness of $1500{\AA}$ and $50.7\Omega$ at the center frequency of 192MHz for au deposition thickness of $1600{\AA}$.

An optimize delectroabsorption modulator structure is designed for high speed optical communication systems considering the extinction efficiency, operating bandwidth, polarization loss, and wavelength chirping. the operating wavelength region is $1.55\mu\textrm{m}$ and the deep ridge structure is adapted for th eminimum polarization loss. Simulations show that the absorption layer thickness larger than $0.25\mu\textrm{m}$, and the modulator length shorter than $200\mu\textrm{m}$ are required for the bandwidth over 10GHz. To obtain the modulatiron efficiency over 10dB/V, a wavelength detuning needs to be determined less than 40meV.

In this paper, we propose a new optical thresold generator as a key-stream genrator for stream cipher systems. The random key-bit stream is generated by a digital generator that is composed of LFSRs and nonlinear ligics. Digital implementatin of a key-stream generator requires large memory to implement programmable tapping points. This memory problem may be overcome easily by using the proposed optical system which has the proberty of 2D parallel processing.To implement hte threshold generator optically, we use conventional twisted nematic type SLMs (LCDs). This proposed system is based on the shadow casting technique for the AND operation between taps and sregister stages. It is also based on the proposed PMRS method for modulo 2 addition. The proposed PMRS method uses the property of light's polarization on LCD and can be implemented optically using one LCD and some mirrors. One of the major advantages of the proosed system is that there is no limitation of the number of the progarmmable tapping points. Therefore, the proposed system can be applied for the 2D encryption system which processes large amounts of data such as 2D images. We verify the proposed system with some simulation.