Im-ocdm híbrido para isac: uma abordagem baseada em aprendizado profundo para comunicação e sensoriamento em redes 6g

Abstract

This dissertation proposes and evaluates an integrated waveform and receiver architecture for Integrated Sensing and Communication (ISAC) systems, based on index modulation over Orthogonal Chirp Division Multiplexing (IM-OCDM). The proposed waveform, referred to as ZC-IM-OCDM, assigns communication symbols to the active subchirps and uses the complementary support to map radar coefficients obtained by applying a local Discrete Fresnel Transform (DFnT) to a temporal Zadoff–Chu sequence. This construction enables a structured radar component to be incorporated into the transmitted block while preserving the support structure of index modulation. The power balance between the communication and radar components is controlled by the power ratio (PR). The analysis is organized into two complementary parts. In Part I, the radar sensing performance is investigated through the autocorrelation of the transmitted waveform, using the Expected Integrated Sidelobe Level (EISL) and the Peak Sidelobe Level Ratio (PSLR) as the main metrics. The results show that the proposed hybrid waveform reduces the sidelobe energy with respect to both Orthogonal Frequency Division Multiplexing - Payload Matched (OFDM-PM) and zero-padded IM-OCDM references, especially when the ZC-derived radar coefficients are favored by the PR setting. In Part II, the communication performance is evaluated by means of a Convolutional Neural Network (CNN)–Transformer neural receiver, formulated as a per-subchirp five-class classifier with auxiliary communication/radar supervision and Maximum a Posteriori (MAP)-restricted decoding. The results over Rayleigh and Rician channels show that performance depends systematically on the structural regime k, on PR, and on the propagation channel. The final formulation preserves the structural constraints of index modulation and achieves competitive communication performance while supporting the dual communication–radar operation of the proposed waveform.