Contributions to chirp-based waveform design and physical-layer security for plc systems

Abstract

This dissertation investigates the design of chirped waveforms for baseband transmission and the enhancement of Physical Layer Security (PLS) in wiretap channel scenarios, with a focus on Power Line Communication (PLC) systems. Specifically, the work proposes the Hermitian Symmetric Orthogonal Chirp-Division Multiplexing (HS-OCDM) technique and evaluates an Artificial Noise (AN) injection strategy exploiting the degrees of freedom provided by the Cyclic Prefix (CP). Regarding waveform design, the time-frequency structure of existing real-valued OCDM techniques is analyzed via group delay. It is demonstrated that this existing formulation yields chirp sequences that occupy only half of the available symbol block duration, thereby restricting their temporal spreading. To address this limitation, a modified Discrete Fresnel Transform (DFnT) kernel is proposed to introduce the HS-OCDM technique. Numerical analyses confirm that this formulation ensures the transmission of real-valued sequences that fully occupy the symbol block, offering superior robustness against synchronization impairments and CP length violations. Subsequently, in the context of PLS, the effectiveness of a CP-based AN injection strategy is evaluated using a unified formulation. This approach enables a comparative assessment of the HS-OCDM technique against other CP-based techniques across both narrowband and broadband PLC channels. Numerical results reveal that the HS-OCDM technique proves to be the most robust alternative under severe channel and synchronization impairments. Finally, the statistical behavior of the CP-based AN vector is investigated through first- and second-order statistical moment analyses. Numerical results show that the projection onto the legitimate channel’s null space introduces non-negligible correlation patterns. Although the AN energy is confined to the CP at the legitimate receiver, enabling easy removal, the persistence of these cyclic correlations at the eavesdropper represents a potential vulnerability that could be exploited.