Power allocation and signal labelling on physical layer security

Abstract

Secure communications between legitimate users have received considerable attention recently. Transmission cryptography, which introduces secrecy on the network layer, is heavily relied on conventionally to secure communications. However, it is theoretically possible to break the encryption if unlimited computational resource is provided. As a result, physical layer security becomes a hot topic as it provides perfect secrecy from an information theory perspective. The study of physical layer security on real communication system model is challenging and important, as the previous researches are mainly focusing on the Gaussian input model which is not practically implementable. In this thesis, the physical layer security of wireless networks employing finite-alphabet input schemes are studied. In particular, firstly, the secrecy capacity of the single-input single-output (SISO) wiretap channel model with coded modulation (CM) and bit-interleaved coded modulation (BICM) is derived in closed-form, while a fast, sub-optimal power control policy (PCP) is presented to maximize the secrecy capacity performance. Since finite-alphabet input schemes achieve maximum secrecy capacity at medium SNR range, the maximum amount of energy that the destination can harvest from the transmission while satisfying the secrecy rate constraint is computed. Secondly, the effects of mapping techniques on secrecy capacity of BICM scheme are investigated, the secrecy capacity performances of various known mappings are compared on 8PSK, 16QAM and (1,5,10) constellations, showing that Gray mapping obtains lowest secrecy capacity value at high SNRs. We propose a new mapping algorithm, called maximum error event (MEE), to optimize the secrecy capacity over a wide range of SNRs. At low SNR, MEE mapping achieves a lower secrecy rate than other well-known mappings, but at medium-to-high SNRs MEE mapping achieves a significantly higher secrecy rate over a wide range of SNRs. Finally, the secrecy capacity and power allocation algorithm (PA) of finite-alphabet input wiretap channels with decode-and-forward (DF) relays are proposed, the simulation results are compared with the equal power allocation algorithm.