Leveraging waveform structure for sub-Nyquist signal processing in mmWave applications

With increasing size of antenna arrays, energy efficient array processing is critical to viable solutions for mmWave 5G applications. Transmitters and receivers use antenna arrays to enable directional communication, high speed data transfer, and massive MIMO applications in 5G technologies. In this dissertation, we introduce a new concept of scaling power consumption using waveform structures. We apply this concept to the angle of arrival (AoA) estimation as an an array processing example. Directional communication with antenna arrays is necessary to mitigate the high propagation loss in a mmWave channel. Therefore, AoA is a useful parameter to characterize a mmWave channel. In a mmWave mobile device, power consumption by analog-to-digital converters (ADCs) is a primary concern for array processing at receivers. Power consumption by ADCs scales linearly with sub-GHz sampling speeds and quadratically with above GHz sampling speeds. In our solution, we leverage the waveform structure of the systematically and cooperatively designed pilot signals transmitted by multiple sources. We develop AoA estimation solutions from subsampled received signals at each RF chain. The subsampling frequencies can be scaled with a resolution that is an integer fraction of the signal’s Nyquist rate. A receiver can scale down the sampling frequencies when higher performance is not required and thus reduce a significant amount of energy consumption. In our first contribution, we present a variable rate preprocessing solution for array processing. The receiver decouples the superimposed source signals using the known set of designed pilot sequences. The variable rate sub-Nyquist decoupling prior to array processing has the potential to reduce computational complexity, improve estimation performance and increase the number of detectable sources. In our second contribution, we develop a variable rate two step AoA estimation solution. The first step decouples multiple sources at sub-Nyquist rate, similar to the first contribution but without the knowledge of the pilots’ parameters. A cooperative pilot design with special waveform structure makes computationally efficient decoupling and estimation of pilot parameters possible. This opens the possibility to a low overhead flexible access to mmWave networks where useful information about the sources is embedded in the pilot parameters. The second step performs maximum likelihood AoA estimation of the decoupled sources using a flexible antenna array design. Due to the temporal decoupling of the source signals, the antenna array has a simpler task of estimating a single AoA and can be flexibly designed to address other implementation challenges.