An architecture for faster than Nyquist Turbo broadcasting

Recently [16], we have proposed using faster than Nyquist (FTN) signaling to achieve transmission over continuous-time broadcast channels and shown that it can achieve the capacity region of the two-user Gaussian broadcast channel. Benefits of FTN broadcasting include separate encoding and explicit transmission of all users' data, i.e., unlike the previously proposed broadcast coding schemes, no joint encoding is needed. This paper presents a design of a Turbo-coded broadcast transmitter based on the FTN signaling. The proposed receiver architecture has a low implementation complexity and is based on Turbo decoding and successive cancellation of FTN-induced intersymbol interference. The presented simulation results indicate that the designed FTN broadcast architecture can be superior to the time-sharing broadcasting used in practice, and shows the potential to perform close to the capacity boundaries of the Gaussian broadcast channel.

I. Introduction

Broadcast channel coding has been shown to allow higher data rates than the traditional time-sharing broadcasting [1]. In particular, the discrete-time Gaussian broadcast channel has been studied extensively due to its practical relevance and the superposition coding and dirty paper coding were shown to achieve the capacity region of this channel [1]–[9]. The main problem in coding for broadcast channel is how to reliably multiplex more than one message into one channel symbol, while keeping the total number of channel symbols the same as in the single user channel. Previously known broadcast coding techniques multiplex the messages at the discrete-time symbol domain and the connection with the continuous-time channel is made with the Nyquist rate transmission and sampling.