If the fiber dispersion of soliton transmission line is optimized, the amplifiers output power becomes almost constant for different amplifier spacing and pulse width. Numerical simulations indicate the optimal dispersion can be determined, as the ratio of amplifier spacing to dispersion length is about 0. 8 for uniform dispersion line.

Polarization mode properties of a polarization-preserving fiber having elliptical stress-cladding with barrier layer around the core are clarified theoretically and experimentally. Normalized frequency dependence of modal birefringence and polarization mode dispersion were measured for the polarization-preserving fiber. It is found from theoretically and experimentally evaluated results that the elliptical stress-cladding fiber has polarization-mode properties similar to those of the polarization-preserving fiber having isolated stress-producing lobes. Stress profiles of the elliptical stress-cladding fiber were measured and calculated for confirming the polarization-mode properties of the fiber. It is known from the measured results that the similarities of polarization-mode properties between the elliptical stress-cladding fiber and the fiber will stress-producing lobes are caused by those of differential stress profiles around the core regions between them.

Experimental optical gain characteristics of an erbium-doped fiber amplifier have not been explained well by conventional laser schemes in the case of two-channel amplification. Modified simple laser schemes including cross relaxation among degenerate levels were valid for the explanation of the optical gain dependence on input signal power and on the erbium-doped fiber length.

A measurement technique of chromatic dispersion in single-mode fiber transmission lines is proposed. By using a local signal generator, an optical line to transmit a reference signal is unnecessary in the proposed measurement technique, even if the input and output end points of the fiber are separated by a large distance.

Mode power distribution of a 1 km long parabolic-index optical fiber is measured by two methods; the far-field pattern method and the pulse method. These results are in good agreement with each other and the validity of the far-field pattern method to the parabolic-index fiber is confirmed.

Splice loss arising from differences in fiber structural parameters such as outer diameter, core diameter, and refractive index difference is studied experimentally and theoretically. Practical formulation for splice loss is made as a function of parameter differences between fibers to be spliced. It is found that the experimental splice loss for conventional graded-index fibers is estimated within an error of 0.1 dB. Considering the fabrication yield rate of fibers as well as splice loss, tolerances of the fiber parameter deviations from the optimum values are determined.

Optical amplifier structures suitable for a 622Mbit/s repeater in an optical communication system containing one in-line amplifier have been investigated. Two wavelengths of 1.533µm and 1.549µm are considered for two cases, i.e., single-channel transmission and two-channel wavelength division multiplexing transmission. The basic amplifier structure is of a two-stage type where forward pumped and backward pumped erbium-doped fibers are connected with each other through intermediate optical filters and an optical isolator. First, the effect of the intermediate optical filters was clarified in optical gain and bit error rate characteristics. Then, the erbium-doped fiber length was optimized on the basis of the allowable optical loss of the optical system which was operated at a bit error rate of 10-9. As a result, the appropriate length of the forward pumped erbium-doped fiber was found to be about 20m for both cases of single-channel and two-channel wavelength multiplexing amplifiers. With the designed amplifier used in the system, the calculated allowable optical line loss was more than 90dB for both the cases.

A optical cable of eight step index fibers was installed in a duct. Prior to laying the cable, four subducts were laid in the duct to allow independent laying of four cables. A cable jointing test was implemented in a manhole. A 32 Mbit/s transmission system was tested on a 5.15 km fiber link.