Title: Spectral Efficient Communications employing 1-Bit Quantization and Oversampling at the Receiver

Abstract: To relax power consumption requirements in multigigabit/s communications systems low resolution quantization can be used. Information-theoretic results have shown that systems employing 1-bit quantization and oversampling are a viable option for this. This work investigates such a structure under the influence of additive Gaussian noise and two matched pulse shaping filters. It is described how a BCJR algorithm, based on a finite-state channel assumption, can be used to reconstruct symbols of higher order modulation schemes that allow the transmission of more than one bit per symbol. Furthermore, it is shown how symbol sources that are fitted to the 1-bit constraint can significantly improve the error rate performance of the system.