Physical Layer Security of Short Packet Communications

This dissertation aims to conduct research on security issues of 5G wireless networks, which are vulnerable to external security threats while supporting services for a massive number of users and devices. In practical wireless communication systems, the communication is subject to overhearing by external eavesdroppers due to the broadcast nature of the wireless medium. Physical layer security (PLS) shows promise as a viable option for securing future communication systems because it utilizes channel characteristics to hide transmitted messages from possible adversaries without depending on traditional cryptographic solutions. However, 5G systems are expected to support various traffic types, including short packet transmission, which results in new challenges in terms of security. Particularly, short packet transmission introduces a penalty on the secrecy capacity, which is the rate of secure communication between authorized parties in the presence of an adversary. It is well-known that PLS is based on the assumption that transmission happens with a maximum rate reliably and securely when the blocklengths are sufficiently large. In the literature, limited studies focus on PLS for short packet communications (SPC) and the performance analysis of secure SPC remains an open problem. Our goal is to study large-scale networks, but first, as a simple case, secure communication of a wiretap channel under the attack of an active eavesdropper, with two capabilities, namely half-duplex and full-duplex, is investigated. It appears that an active eavesdropper is more harmful to the secrecy throughput than a passive one, and the full-duplex eavesdropper (Eve) is more dangerous than a half-duplex Eve. Indeed, the performance is measured in terms of average secrecy throughput and theoretical approximations are validated through Monte Carlo simulations throughout all the contributions of the dissertation. Second, the wiretap channel model with multiple passive eavesdroppers is explored to shed light on a more realistic scenario in large-scale wireless networks. Although an increased number of antennas can lead to higher average secrecy throughput, achieving higher secrecy throughput is more effectively accomplished by increasing the transmission rates. As a final contribution, the previous wiretap channel setting is extended by adding multiple receivers. The security performance against colluding and non-colluding attackers is thoroughly examined. According to our results, it is more advantageous for eavesdroppers to collude and they are more powerful when their number increases. And we conclude the dissertation with a discussion of future work.