Simon Bicaïs defended his PhD thesis on Thursday, November 12 at the CEA-Leti. Part of the work was founded by the ANR-Brave project.

Slides are available here.

To deploy high-rate wireless services, future communication networks envisage the use of wide frequency bands. Still, the usual frequency bands in the sub-6 GHz spectrum are extremely limited and expensive. To expand its available spectrum, the forthcoming generation of mobile networks with 5G initiates the use of higher frequencies through the exploitation of millimeter-wave bands. In this search for frequency resources, the sub-THz spectrum from 90 to 300 GHz offers unprecedentedly large available bands, several tens of GHz. Wireless communications in sub-THz frequencies are therefore seen as a foremost solution to achieve Tbit/s data rates and meet the requirements of future wireless connectivity. Nevertheless, existing and mature wireless technologies cannot be directly transposed to the sub-THz bands as they do not consider the specific features of sub-THz communications. Additional research is hence required to design efficient communication systems adapted to the constraints of sub-THz frequencies. Some of the major technological challenges brought by using high carrier frequencies and large bandwidths include: the performance limitations caused by the strong phase impairments of high-frequency oscillators; and the problem of high sampling rates required by the analog-to-digital conversion. In this thesis, the conducted research focuses on the development of the physical layer for sub-THz communication systems and attempts to overcome these technological barriers. Our objective is twofold: to increase the communication data rate and to relax the constraints on radio-frequency architectures. To do so, our approach consists in jointly designing signal processing for the analog and digital domains.

 The two main contributions of this work are: the optimization of coherent transceivers for strong phase noise channels; and the proposal of dedicated communication systems with non-coherent and high-rate architectures. First, we have proposed optimized transmission schemes for strong phase noise channels including: the modulation, the demodulation, and the link adaptation. The proposed solutions achieve high spectral efficiency communications with relaxed constraints on radio-frequency oscillators. Our results show that the use of optimized transmission schemes greatly contributes to mitigate the impact of phase noise on coherent transceivers. Consequently, our work describes valuable technical solutions to the development of physical layers with high spectral efficiency for the sub-THz spectrum. Second, we have also targeted low-complexity physical layers readily implementable in sub-THz frequencies. We have studied the design of communication systems specifically dedicated to the sub-THz bands using non-coherent architectures. In order to implement high-rate communications with non-coherent architectures, we have considered the use of spatial multiplexing and wide frequency bands. Our work on spatial multiplexing in sub-THz frequencies demonstrates that high-rate communications can be implemented with low complexity and low power architectures using multi-antenna systems and energy detection receivers. Besides, the use of wide bands strongly constrains the analog-to-digital conversion. In order to reduce the required sampling frequencies of converters and to simplify practical implementations, we have proposed a new receiver for high-rate impulse radio systems. We have shown that the proposed receiver architecture, using parallel projections of the received signal in the analog domain, leads to near-optimal performance with significantly reduced sampling frequencies.