Files

Abstract

While wireless communication has become an ubiquitous part of our daily life and the world around us, it has not been able yet to deliver the multi-gigabit throughput required for applications like high-definition video transmission or cellular backhaul communication. The throughput limitation of current wireless systems is mainly the result of a shortage of spectrum and the problem of congestion. Recent advancements in circuit design allow the realization of analog frontends for mmWave frequencies between 30GHz and 300GHz, making abundant unused spectrum accessible. However, the transition to mmWave carrier frequencies and GHz bandwidths comes with new challenges for wireless receiver design. Large variations of the channel conditions and high symbol rates require flexible but power-efficient receiver designs. This thesis investigates receiver algorithms and architectures that enable multi-gigabit mmWave communication. Using a system-level approach, the design options between low-power time-domain and power-hungry frequency-domain signal processing are explored. The system discussion is started with an analysis of the problem of parameter synchronization in mmWave systems and its impact on system design. The proposed synchronization architecture extends known synchronization techniques to provide greater flexibility regarding the operating environments and for system efficiency optimization. For frequency-selective environments, versatile single-carrier frequency domain equalization (SC-FDE) offers not only excellent channel equalization, but also the possibility to integrate additional baseband tasks without overhead. Hence, the high initial complexity of SC-FDE needs to be put in perspective to the complexity savings in the other parts of the baseband. Furthermore, an extension to the SC-FDE architecture is proposed that allows an adaptation of the equalization complexity by switching between a cyclic-prefix mode and a reduced block length overlap-save mode based on the delay spread. Approaching the problem of complexity adaptation from time-domain, a high-speed hardware architecture for the delayed decision feedback sequence estimation (DDFSE) algorithm is presented. DDFSE uses decision feedback to reduce the complexity of the sequence estimation and allows to set the system performance between the performance of full maximum-likelihood detection and pure decision feedback equalization. An implementation of the DDFSE architecture is demonstrated as part of an all-digital IEEE802.11ad baseband ASIC manufactured in 40nm CMOS. A flexible architecture for wideband mmWave receivers based on complex sub-sampling is presented. Complex sub-sampling combines the design advantages of sub-sampling receivers with the flexibility of direct-conversion receivers using a single passive component and a digital compensation scheme. Feasibility of the architecture is proven with a 16Gb/s hardware demonstrator. The demonstrator is used to explore the potential gain of non-equidistant constellations for high-throughput mmWave links. Specifically crafted amplitude phase-shift keying (APSK) modulation achieve 1dB average mutual information (AMI) advantage over quadrature amplitude modulation (QAM) in simulation and on the testbed hardware. The AMI advantage of APSK can be leveraged for a practical transmission using Polar codes which are trained specifically for the constellation.

Details

Actions

Preview