Nonuniform Constellation Shaping Schemes for Short Distance and Mid-reach Fiber-optics Link Transmissions

Abstract

Among various industries including IT, finance, and health care, the vast demand of big data and analytics have led to requirements in high efficient transmission. Coherent optical transmission has become the standard approach, and recently researchers also have paid attention to the PAM signaling in data center because of its simplicity. For either modulation method, the traditional uniform distributions have a large gap to the Shannon limit, which restricts the efficient channel utilization. Nonuniform constellation shaping is an energy efficiency enhancement method. Using probabilistic shaping and geometric shaping scheme we can reduce the gap to Shannon limit by mimicking a Gaussian-like shape of constellation. FEC coding has been demonstrated to be an excellent method to correct the bit errors and increase the channel reliability; in particular, the LDPC code represents a good FEC candidate for the high information rate demands. This dissertation demonstrates the nonuniform constellation shaping schemes application for different modulation formats, along with LDPC coding and pulse shaping schemes, to satisfy never ending data demands in the various fields. In this dissertation, we investigate the application of non-uniform constellation shaping schemes into different modulation formats for both IM/DD and coherent detection. The proposed non-uniform shaping scheme is based on both probabilistic shaping and geometric shaping, the modulation formats under investigation include M-ary PAM, M-ary QAM, and 4-dimentional (4D) modulation formats. The main novelty of our work is the joint use of constellation shaping and pulse shaping schemes, together with LDPC coding.