We discuss and experimentally demonstrate a scheme to achieve photorefractive solitons of arbitrary linear polarization using the quadratic electro-optic effect and describe the observation of the self-trapping of a set of linear polarized beams in different positions of a paraelectric photorefractive crystal of potassium-lithium-tantalate-niobate (KLTN) biased by the inhomogeneous field produced by two miniaturized top electrodes. The polarization of the single solitons of the set is determined by the local electrostatic configuration and the underlying tunable anisotropy, which is detected through zero-field electro-activation.

~ 456 / CLEO'98 / THURSDAY AFTERNOON Matched Filter Receiver 4 "(1) AWGNNJ2 R2 CThUS Fig. 1. System configuration with R2 blowup.ber of regenerators; however, we feel that the region in which this analysis is valid provides a misleading and unfair comparison among R1, R2, and R3.For simplicity and tractability, R2 is modeled as a front-end filter h(t) and a timeinvariant thresholder f0. The filter is modeled as a running integrator of width T. Also shown is a 1 pulse of width T through R2. In the following, the losses L, and Lz are normalized, without loss of generality, to 1.An exact analysis of R2 requires the probability density function (pdf) pI(I) ofconditioned on a 1 or 0 being sent. We are unaware of a closed-form solution to pI( I ) .Reference 4 provides bounds on p I ( I ) for T = T. Reference 2 ignores the receiver filter (the integral in the above expression) and hence is applicable only for large receiver bandwidths (BWs) where the integrand can be modeled as constant for a bit time. In this case, I can be approximated as Bernoulli; however, this is an unfair comparison because the performances of R1, R2, and R3 systems can all be improved by reducing the receiver BW. We limit our discussion to b = T/T 2 1 so that s( t) * h( t)s( t). Note that because T must accommodate the maximum supported bit rate (recall that the system is meant to be bitrate transparent), our analysis applies to transmission of lower-rate signals.For large b, we approximate pI(I) by sampling the integrand at the Nyquist rate and assume n(t) is constant within a correlation time T. This provides b statistically independent samples and thus, pAI) is approximately binomial with mean Q(EdbN,) if 0 was sent and A[ 1 -Q(EdbN,)] if 1 was sent, where E, = AzT/2 is the average energy per bit. Simulations have indicated that the binomial is a more accurate approximation than a Gaussian. Figure 2 shows the approximate bit error rate for various ratios of noise before and after the regenerator. Each plot shows the effect of b = 5,10,20 on R2 and the exact results for R1 and R3.For No > > Mo, e.g., regeneration near the egress, R2 performs worse than R1, as expected since (1) the cascade of the regenerator and the receiver can be modeled as a single receiver, (2) a matched filter receiveris optimal, and (3) the cascade does not form a matched filter.For No << M,, e.g., regeneration near the ingress, R2 performs worse than R1 at small SNRs and better than R1 at high SNRs. The crossover SNR increases with b, and hence the required SNR into R2 (to perform at least as well as R1) increases with decreasing bit rates.However, for No << M,,, any reasonable regeneration device will have comparable performance.For No = M,, R2 performs uniformly worse than R1. Although this approximate analysis does not preclude the possibility of a crossover at high SNRs, it appears likely that no significant performance improvements can be achieved at SNRs of interest. Hence, R2 probably does not provide significant distance extension when used as a midspan repeater....

The strong asymmetry in charge distribution supporting a single non-interacting spatial needle soliton in a paraelectric photorefractive is directly observed by means of electroholographic readout. Whereas in trapping conditions a quasi-circular wave is supported, the underlying double-dipolar structure can be made to support two distinct propagation modes.

No abstract

We investigate the effect of non-paraxiality in the dynamics of dispersive shock waves in the defocusing nonlinear Schrödinger equation. We show that the problem can be described in terms of a relativistic particle moving in a potential. Lowest order corrections enhance the wave-breaking and impose a limit to the highest achievable spectrum in an amount experimentally testable.