EECS technical reports

Conditions of Use

Archive Home Page

A Scalable and Cost Effective Architecture for High Gain Beamforming Antennas

Bakr, Omar
Technical Report Identifier: EECS-2010-178
December 23, 2010

Abstract: Many state-of-the-art wireless systems, such as long distance networks (point-to-point, point-to-multipoint, and mesh) and high bandwidth networks using mm-wave frequencies, require high gain antennas to overcome adverse channel conditions. These networks could be greatly aided by adaptive beamforming antenna arrays, which can significantly simplify the installation and maintenance costs (e.g., by enabling automatic beam alignment), and improve the capacity of these networks. Such networks typically require gains ranging from 20-30dBi with wide scanning range in both dimensions. To achieve this, arrays with hundreds or even thousands of antennas are required, which cannot be done with existing techniques that do not scale very well beyond 10-20 antennas.

In this dissertation, we examine and address the main challenges presented by large arrays, starting from electromagnetic/antenna and radio circuit design and proceeding to the signal processing and algorithms domain. We propose 3-dimensional antenna array structures that realize large gains and scan angles at a much reduced size and form factor compared with conventional planar antennas. At the circuit level, we propose a hybrid RF/digital beamforming radio architecture that takes advantage of low cost silicon integration to reduce the overall component count and power consumption levels of the system without limiting the capacity. We consider different techniques for implementing compact beamformers reliably at high radio frequencies, and present signal processing techniques based on adaptive filtering methods for optimizing those beamformers. The performance implications of low precision analog beamformers and implementation errors are also analyzed and quantified, and computationally efficient vector quantization techniques that take advantage of the size and scale of the arrays to compensate for low precision are proposed. We validate our approach with mathematical proofs and computer simulations.