ON BIT ERROR RATE CALCULATION FOR OPTICAL FIBRE SYSTEMS

Abstract

In this thesis, two methods of error rate calculation for a digital optical fibre system are studied and compared. The first method makes use of the Gauss Quadrature Rule (GQR) to evaluate the error probability integral. This method has been widely applied by many investigators. The second method uses a saddlepoint approximation to compute the error probability integral. This second technique is computationally simpler. In order to ascertain the accuracy of these two techniques, an experimental system has been constructed and error rates measured for comparison with the computed results. The effect of different receiver filter bandwidths on the error performance, which has not been studied before, is also studied and the bandwidth which yield the minimum error probability is identified. Finally, a gamma probability distribution is used to model the received optical power fluctuation in a digital fibre optic system affected by 'modal' noise. New recursive formulae for the moments of the receiver output current are obtained which are used in the GQR approximation of the error probability. New saddlepoint equations are also derived which are used to evaluate the error probability in the second method. Error rates are then calculated for varying degrees of modal noise disturbance.