The project addresses optical data transmission in long-haul fiber optical links based on the nonlinear Fourier transform (NFT). In NFT-based transmission, the Kerr effect-based nonlinearities can be treated as a usable property, thus increasing the fiber capacity at high signal power levels. The focus of the first project phase was on integrating a transmitter supporting optical transmission using the discrete nonlinear Fourier domain spectrum only, which describes the solitonic parts of the signal. We showed transmission of four closely packed QPSK-modulated solitons over a distance of 3800 km, with a pulse-to-pulse separation of 250 ps in the time domain and 15 GHz in the frequency domain, enabling a data rate of 8 Gb/s (single polarization) in this experiment in a 60 GHz window with an integrated silicon photonics transmitter. Closer multiplexing can increase the data rate to more than 16 Gb/s using the integrated transmitter.The overarching objective of the second project phase is to concomitantly exploit both the continuous and discrete nonlinear spectra, building on the techniques developed in the first phase in the transmitter and developing a novel integrated receiver enabling this. Using both spectra will allow us to increase the spectral efficiency significantly. The continuous spectrum as yielded by the NFT is analogous to the spectrum of the conventional Fourier transform and can thus carry a similar spectral efficiency. The discrete spectrum then adds additional capacity, as well as potential equalization possibilities. In order to further advance the state of the art of nonlinear Fourier transform based optical transmission, it will be necessary to develop an integrated Rx able to receive the entire spectrum (consisting of both continuous and discrete nonlinear spectra), without dead bands, so that the NFT can be applied to the entire signal. In the first phase, the best signal-to-noise ratio was achieved by slicing the spectrum at the receiver and analyzing each soliton subchannel independently, as joint reception was in particular hindered by excessive signal dynamical ranges. This, however, leads to dead bands between the subbands incompatible with an NFT analysis of the continuous spectrum. The solution pursued in the second phase will be to receive overlapping spectral slices in an integrated receiver and to stitch the overall spectrum back together without dead bands, after which the full spectrum can be analyzed with the NFT and the transmitted information recovered. Apart from investigating the robustness of the algorithms and investigating novel equalizer concepts numerically, the functionality of the NFT-based transmission system comprising both Tx- and Rx-side photonic integrated circuits will be verified by transmission experiments.