Feedback design and application of beamforming in cellular and multicell environment
Abstract
Transmit beamforming is a popular low complexity transmission technique that sends data along the channel’s strongest singular mode. Through a feedback channel with limited rate, the transmitter receives knowledge of the channel and adapts its transmit beamforming vector accordingly. As feedback channel typically has finite capacity, the design of the vector quantizer is critical. This thesis includes studies on several issues related to the design and application of feedback in a beamforming system. First, we analyze and derive closed-form expressions for key performance metrics of a system that uses random vector quantization (RVQ) codebook, which is known to be asymptotically optimal. Second, we cast the beaforming vector selection problem as a selection combining (SC) problem, thereby leveraging existing literature on SC technique. Third, the multiuser downlink channel with limited feedback suffers interference from inaccurate quantization. At high SNR, the system enters interference limited region and the sum-rate cannot surpass a ceiling. We propose a low-cost technique to overcome this ceiling by dynamically switching between multiuser and single-user mode. Fourth, we study a point-to-point system where both users have data to transmit to the other user. Through power allocation, each user can tradeoff between its forward transmission rate versus the transmission rate of the other user. The overall achievable rate tradeoff curve is characterized as a dual-objective optimization problem. Fifth, we study the multicast problem where a transmitter sends common data to a number of receivers, who individually feeds back channel information back to the transmitter. By coding the data over multiple coherence times, the reliable multicast rate is increased from coding for a single coherence time. Lastly, we study the quantization problem in a cooperative beamforming setting. We identified trellis quantization as the most suitable technique for this application. We present trellis design techniques, trellis simplification technique, low-complexity tail-biting technique, and a nonuniform quantization technique for the beamforming vector quantization problem.
Degree
Ph.D.
Advisors
Love, Purdue University.
Subject Area
Electrical engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.