Bit-Wise Decoders for Coded Modulation and Broadcast Coded Slotted ALOHA

Doctoral thesis, 2016

This thesis deals with two aspects of wireless communications. The first aspect is about efficient point-to-point data transmission. To achieve high spectral efficiency, coded modulation, which is a concatenation of higher order modulation with error correction coding, is used. Bit-interleaved coded modulation (BICM) is a pragmatic approach to coded modulation, where soft information on encoded bits is calculated at the receiver and passed to a bit-wise decoder. Soft information is usually obtained in the form of log-likelihood ratios (also known as L-values), calculated using the max-log approximation. In this thesis, we analyze bit-wise decoders for pulse-amplitude modulation (PAM) constellations over the additive white Gaussian noise (AWGN) channel when the max-log approximation is used for calculating L-values. First, we analyze BICM systems from an information theoretic perspective. We prove that the max-log approximation causes information loss for all PAM constellations and labelings with the exception of a symmetric 4-PAM constellation labeled with a Gray code. We then analyze how the max-log approximation affects the generalized mutual information (GMI), which is an achievable rate for a standard BICM decoder. Second, we compare the performance of the standard BICM decoder with that of the ML decoder. We show that, when the signal-to-noise ratio (SNR) goes to infinity, the loss in terms of pairwise error probability is bounded by 1.25 dB for any two codewords. The analysis further shows that the loss is zero for a wide range of linear codes. The second aspect of wireless communications treated in this thesis is multiple channel access. Our main objective here is to provide reliable message exchange between nodes in a wireless ad hoc network with stringent delay constraints. To that end, we propose an uncoordinated medium access control protocol, termed all-to-all broadcast coded slotted ALOHA (B-CSA), that exploits coding over packets at the transmitter side and successive interference cancellation at the receiver side. The protocol resembles low-density parity-check codes and can be analyzed using the theory of codes on graphs. The packet loss rate performance of the protocol exhibits a threshold behavior with distinct error floor and waterfall regions. We derive a tight error floor approximation that is used for the optimization of the protocol. We also show how the error floor approximation can be used to design protocols for networks, where users have different reliability requirements. We use B-CSA in vehicular networks and show that it outperforms carrier sense multiple access currently adopted as the MAC protocol for vehicular communications. Finally, we investigate the possibility of establishing a handshake in vehicular networks by means of B-CSA.