【正文】 onjugate reflector and the sensitivity is improved by the nonreciprocal nature of Faraday effect. Ⅱ. THEORETICAL BACKGROUND The basic structure of a polarimetric fiber optic current sensor is shown in Fig. 1. The linear polarization light produced by the polarizer is launched into the sensing fiber coil and rotated by the magnetic field induced by the current to be measured, and then the outgoing light wave is divided by the Wollaston prism into two orthogonal parts, the x ponent and the y ponent. The relative intensity of the two parts represents the Faraday rotation, with which the current can be acquired. The sensing fiber can be described by Jones matrix as [5]:where δ and θ denote, respectively, the linear birefringence of the fiber and the Faraday rotation angle. If the input light vector is oriented at 45176。 with respect to the x and y axes, thelight vector from the sensing fiber isthen the sensor response can be written as It is showed that the sensor response is not only related to Faraday rotation θ but also related to linear birefringenceδ. When δ is extremely small, the sensor response can be written as According to the above relation, we can get θ, and then the current to be measured can be obtained through θ. Hence δ is the major cause affecting the sensor’s accuracy. Because both δ and θ increase with the fiber length, it is not possible to increase the sensitivity by increasing the fiber length. This is the socalled quenching effect of the linear birefringence on Faraday rotation. Secondly, the linear birefringence itself is environmentally sensitive。 therefore the stability of the sensor will bee a serious problem. This problem can be alleviated to some extent by decreasing the intrinsic birefringence of the sensing coil. At the present time, the major approach is to anneal the fiber coil to eliminate its birefringence. But the residual birefringence and environmental factors are still critical to sensor response. Ⅲ. PROPOSED REFLECTION INTERFEROMETER CONFIGURATION The proposed OCT configuration is showed in Fig. 2. Light from a laser is directed to a polarizer. The resulting polarized light travels through the sensing fiber coil and then reflected by the conjugate reflector. The backward light goes through the fiber coil once more and arrives at the polarizer P2. The two orthogonal vectors along x and y axis, as illustrated in Fig. 3, are interfered by the output polarizer P2 and finally coupled to the detector, which produces the sensor’s response. Optical phase conjugation is a nonlinear optical phenomenon, which has been studied extensively. The phase of the output wavefrom a conjugate mirror is reversed instantaneously relative to the incident wave [6]. With this feature, the linear birefringence δ, which is reciprocal, can be pensated by the forward and backward propagations. But the Faraday rotation, which is nonreciprocal, will be doubled by the two propagations. In contrast, if an ordinary mirror is used, the structure is only equal to doubling the length of the fiber coil, and do not increase the sensor’s sensitivity.Fig. 2. Interferometric optical fiber current sensorFig. 3. The schematic diagram of the output interference wavesIf the incident polarized light to the sensing fiber isthen the outgoing light vector from the sensing fiber is The light vector reflected by the conjugate mirror is given by After traveling backward through the sensing fiber, the final output light wave can be written as The x and y ponents of Eo interferes on the axis of P2, and the bined vector isThe output intensity is given bywith the two casesandthe sensor’s outputs can be calculated respectively as If the linear birefringence can be ignored, that is δ = 0 and Δ=θ, from the two above equations we can get Faraday effect is usually very weak in optical fiber, and θ is very small, so the sensibility of I1 is larger than that of I2. Hence the error caused by δ can be given by If utilizing an ordinary mirror as the reflector, the final output light wave can be obtained as The x and y ponents of Eo interferes on the axis of P2, and the bined vector isThe output intensity is still I = E , with the two casesThe sensor’s outputs can be calculated respectively as If the linear birefringence can be ignored, . δ = 0 andΔ=θ, from the two above equations we can get For the same reason, the sensitivity of I1 is larger than that of I2. The error caused by δ can still be given by (2). Ⅳ. NUMERICAL SIMULATION AND ANALYSIS The proposed structure illustrated in Fig. 2 has been numerically simulated. Fig. 4 is the response curves as functions of linear birefringence determined by (1) and (4). Fig. 5 is the response curves as functions of Faraday rotation angle determined by (1) and (4). Fig. 6(a) is the relative error curves as functions of linear birefringence determined by (1), (2), and (3) with a conjugate mirror as the reflector. Fig. 6(b) is the relative error curves as functions of linear birefringence determined by (4), (5), and (3) with an ordinary mirror as the reflector. The parison between Fig. 4(a) and (b) shows that the response of the configuration with a conjugate reflector is relatively flat, and the response of the configuration with anordinary mirror is relatively abrupt. For the response as function of Faraday rotation angle, only the monotone increasing part of the curve can be useful. Comparing the curves in Fig. 5(a) with that in (b) with the same δ, the curves in (a) increase faster than that in (b). It states that the conjugate mirror can alleviate the effect of δ pared with an ordinary mirror. From Fig. 6 we can find that the error with conjugate mirror is considerably decreased pared with the ordinary mirror configuration at the same θ and δ. Calculation shows that, for 177。% measuring error, the resid