COST 241 measurements of the nonlinear refractive index, n/sub 2/, exhibit a large scatter depending on the specific measurement technique. This is largely due to the electrostrictive contribution to the Kerr nonlinearity, as is revealed by the resonant behaviour of n/sub 2/ (with peak values up to 3.9 10/sup -20/ m/sup 2/ W/sup -1/) observed with signal modulation frequencies in the 0.11 GHz range.

We demonstrate experimentally a versatile fibre ring laser. When all-optically mode-locked by a data signal the fibre laser can generate either a copy of a repetitive data signal, a recovered optical clock, or a recovered frame synchronization signal. We propose to use the recovered frame synchronization signal in a novel true all-optical demultiplexer.

All-optical frame synchronisation recovery is demonstrated experimentally. A frame encoded input data signal forces a fibre laser to modelock. The fixed header is copied onto the laser output, whereas in the payload area the clock is recovered. The authors propose to use this frame synchronisation signal in an all-optical demultiplexer that unambiguously identifies each individual channel.

Distortions induced by erbium doped fiber amplifiers on optical AM signals are shown, experimentally and theoretically, to originate from the so called locked-inversion gain tilt of the amplifier. The locked-inversion gain tilt may be measured by simple methods, or it may be calculated from a standard steady-state numerical model. It is demonstrated by experiment and calculation that the gain-tilt, and hence the harmonic distortion, may be reduced to acceptable levels by proper design of the fiber amplifier without the need for separate equalization.

A theoretical analysis describing an interpretation of the fundamental processes, responsible for the second-order distortion of AM signals in erbium-doped fiber amplifiers (EDFAs) is presented. It is shown that the locked-inversion gain tilt rather than the widely used continuous wave gain tilt is responsible for the distortion. The theoretical analysis is confirmed by experiments showing that the two types of gain tilt may differ by factors of up to 60. In particular, it is demonstrated that the locked-inversion gain-tilt may be significant even at the CW gain peak. Experimental results indicate that the relevant gain tilt can be reduced to acceptable levels by proper design of the EDFA.< >

Measurement of composite second-order distortion (CSO) for a 28-channel AM-TV system verifies a recent theory on erbium-doped fiber amplifier (EDFA) gain-tilt distortion. The fundamental difference between locked-inversion and continuous-wave (CW) gain-till is explained experimentally and theoretically. Good agreement between simulated and measured locked-inversion gain curves are found. The results confirm that gain-tilt distortion in optical AM-TV systems can be reduced to acceptable levels by proper EDFA design. The measurements indicate that EDFA-induced CSO cannot be used to compensate CSO generated by other mechanisms over the whole frequency band in multichannel systems.< >

The existence of discrete cladding modes in single-mode fibers is illustrated by inducing periodically repeated microbends along the fiber axis and performing spectral measurements of the loss are performed. In order to explain the results of the measurements, it is necessary to apply a microbending theory in which coupling between the guided mode and a number of discrete cladding modes is considered. Very good agreement between theory and measurement is achieved. The consequences of the existence of discrete cladding modes with regard to the proper choice of artificial microbending spectrum used for fiber characterization are also discussed.< >

Microbending loss calculations for (single-coated) singlemode fibres are performed using the theory of linear load distributions on semi-infinite spaces to include the buffer effect of the coating. Together with the inclusion of photoelasticity this method significantly improves the agreement between calculated and measured results. Practical consequences for measurement methods are discussed.

Using a multilayer approximation of the index distribution, analytical solutions of the microbending losses according to Petermann&#39;s approach for single-mode fibers with arbitrary index profiles can be derived. It is shown that less than 20 layers are required to ensure satisfying accuracy for graded-index single-mode fibers. For single-cladded fibers, simple profile-independent formulas for the ω...

A new bending loss formula for single-mode fibres is presented. Comparison with measurements and conventional theory shows good agreement. Reported bends on the loss curves for matched- and depressed-clad fibres are explained.

Independent of the actual index profile of a single-cladded dispersion-shifted single-mode fiber, it is shown that the loss components given in the headline at the zero dispersion wavelength λ 0 accurately can be obtained by simple expressions that only depend on the Laplace spot size value at λ 0 . With \lambda_{0} = 1.55 \mu m, these relations are used to estimate and minimize the total loss in single-cladded dispersion-shifted fibers with arbitrary core-index profiles.

Based on profile-independent expressions, optimum values of the dispersion slope at the zero-dispersion wavelength for single-mode fibers are given, as a function of production tolerances. This optimum value minimizes the spread in dispersion or maximizes the usable wavelength interval of the fiber. Maximum allowable tolerances for dispersion-flattened fibers are found.