A universal coding approach for superposition mapping

Superposition mapping (SM) is a modulation technique that uses linear superposition to produce Gaussian-like data symbols. Communication systems employing SM have a theoretical potential to approach the Gaussian channel capacity without using active signal shaping. This paper tackles several critical issues of SM and provides corresponding solutions. We point out that repetition coding is often an indispensable part for SM systems, a topic which has been obscured for a long time. More importantly, the bandwidth efficiency limit of SM systems with equal power allocation of about 2 bits per symbol per dimension can be eliminated by using irregular repetition codes. Following this cognition, we propose a universal coding approach, called low-density hybrid-check (LDHC) coding, for SM systems with arbitrary power allocation.