Co-ordinate Interleaving over Parallel Gaussian Channels

Abstract

Co-ordinate Interleaving (CI) along with rotated constellations improves diversity gain over fading channels [1]. We propose employing CI over parallel Gaussian channels for improving the sum mutual information. I. INTRODUCTIONWhen the inputs to parallel Gaussian channels are Gaussian, waterfilling (WF) maximizes sum mutual information (MI). For given arbitrary input distributions, mercury/waterfilling (MWF) [2] was shown to maximize sum MI.Co-ordinate interleaving (CI), with input symbols from rotated QAM constellations, improves diversity gains over fading channels [1]. This paper studies CI from the mutual information perspective. With the insight gained from MWF and the precoding scheme of [3], we propose employing CI over parallel Gaussian channels for improving sum MI.