Modeling and Compensation of Nonlinear Distortion in Multi-Antenna RF Transmitters
Multi-antenna systems are utilized as a way to increase spectral efficiency in wireless communications. In a transmitter, the use of several parallel transmit paths and antennas increases system complexity and cost. Cost-efficient solutions, which employ active antenna arrays and avoid expensive isolators, are therefore preferred. However, such solutions are vulnerable to crosstalk due to mutual coupling between the antennas, and impedance mismatches between amplifiers and antennas. Combined with the nonlinear behavior of the power amplifiers, these effects cause nonlinear distortion, which deteriorates the quality of the transmitted signals and can prevent the transmitter from meeting standard requirements and fulfilling spectrum regulations. Analysis, assessment and, if necessary, compensation of nonlinear distortion are therefore essential for the design of multi-antenna transmitters.
In this thesis, a technique for modeling and predicting nonlinear distortion in multi-antenna transmitters is presented. With this technique, the output of every individual transmit path, as well as the radiated far-field of the transmitter can be predicted with low computational effort. The technique connects models of the individually characterized transmitter components. It can be used to investigate and compare the effects of different power amplifier and antenna array designs at early design stages without complicated and expensive measurements.
Furthermore, a digital predistortion technique for compensating nonlinear distortion in multi-antenna transmitters is presented. Digital predistortion is commonly used in transmitters to compensate for undesired nonlinear hardware effects. The proposed solution combines a linear function block with dual-input predistorters. The complexity is reduced compared to existing techniques, which require highly complex multivariate predistorter functions.
Finally, a technique for identifying multi-antenna transmitter models and predistorters from over-the-air measurements using only a small set of observation receivers is presented. Conventional techniques require a dedicated observation receiver in every transmitter path, or one or more observation receivers that are shared by several paths in a time-interleaved manner. With the proposed technique, each receiver is used to observe several transmitter paths simultaneously. Compared to conventional techniques, hardware cost and complexity can be reduced with this approach.
In summary, the signal processing techniques presented in this thesis enable a simplified, low-cost design process of multi-antenna transmitters. The proposed algorithms allow for feasible, low-complexity implementations of both digital and analog hardware even for systems with many antennas, thereby facilitating the development of future generations of wireless communication systems.
EB lecture hall, Hörsalsvägen 11, Chalmers
Opponent: Prof. Pere Lluis Gilabert Pinal, Department of Signal Theory and Communications, Universitat Politècnica de Catalunya, Barcelona, Spain