Receiver Design for Digital Fiber Optic Communication Systems, I

Abstract

This paper is concerned with a systematic approach to the design of the “linear channel” of a repeater for a digital fiber optic communication system. In particular, it is concerned with how one properly chooses the front-end preamplifier and biasing circuitry for the photodetector; and how the required power to achieve a desired error rate varies with the bit rate, the received optical pulse shape, and the desired baseband-equalized output pulse shape. It is shown that a proper front-end design incorporates a high-impedance preamplifier which tends to integrate the detector output. This must be followed by proper equalization in the later stages of the linear channel. The baseband signal-to-noise ratio is calculated as a function of the preamplifier parameters. Such a design provides significant reduction in the required optical power and/or required avalanche gain when compared to a design which does not integrate initially. It is shown that, when the received optical pulses overlap and when the optical channel is behaving linearly in power, <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sup> baseband equalization can be used to separate the pulses with a practical but significant increase in required optical power. This required power penalty is calculated as a function of the input and equalized pulse shapes.