New techniques for relay network beamforming on frequency-selective channels

• Main Author: Wang, Tao
• Other Authors: Er Meng Hwa
• Format: Theses and Dissertations
• Language: English
• Published: 2014
• Subjects: DRNTU::Engineering::Electrical and electronic engineering::Electronic systems::Signal processing
• Online Access: https://hdl.handle.net/10356/61026
• Institution: Nanyang Technological University

User-cooperative techniques have recently gained tremendous interests in wireless communication society, for they are able to exploit the spatial diversity among distributed network nodes. In cooperative relay networks, data transmissions between sources and destinations can be assisted by relays which form a virtual antenna array. In this thesis, we investigate relay network beamforming using the filter-and-forward (FF) relaying scheme on frequency-selective channels. The thesis makes original contributions by introducing two classes of new techniques for relay network beamforming design. One class involves designing the beamformers to achieve a distortionless response. The other class uses an alternative optimization criterion by minimizing the output power at the destination. For the first class of technique for relay beamforming design, we consider a relay network on frequency-selective channels, in which the received signals are affected by the inter-symbol-interference (ISI), and the FF scheme is adopted to combat ISI. To adaptively select the optimal delay decision which crucially affects the performance, we formulate and solve the single-user relay beamforming problem using the frequency-domain approach. A new technique of distortionless design is introduced which aims at forcing the frequency response of equivalent channel from the source to the destination to be flat and linear phase. Hence the proposed approach eliminates the ISI completely, and it is shown to have lower computational complexity in selecting optimal decision delay according to channel realizations. As the second class of technique for relay beamforming design, we develop a new criterion which amounts to minimize the destination output power also subject to the distortionless constraint and is described in frequency domain. An adjustable scalar amplifier at the destination side is introduced and its range of value and effects on the system performance are discussed. With the scalar amplifier simultaneously optimized with the relay beamformer coefficients, we demonstrate that the proposed new technique is equivalent to the distortionless SINR maximization criterion in terms of output SINR performance. In addition, the output power minimization scheme is extended to a multi-antenna destination scenario, where the destination also deploys a beamformer which is jointly designed with the relay beamformer. By adopting the new technique of output power minimization, we propose a relay beamforming design without the distortionless constraint. We demonstrate that the proposed approach attains the optimal output SINR that is given by the SINR maximization design scheme. Moreover, although the same output SINR performance is achieved, when the individual relay transmit power constraint is taken into account, the proposed approach requires lower computational complexity than the SINR maximization scheme. In addition, a time-domain version of the distortionless constraint is proposed, and with this constraint, the efficient selection of decision delay is also considered with a time-domain approach. Subsequently, we extend the output power minimization scheme to cater for multiple peer-to-peer FF relay beamforming. Two criteria are considered, in which one is to minimize the maximal output power among all destinations, and another is minimize the sum of the output power. The former one is proved to be equivalent to the worst SINR maximization scheme in terms of output SINR performance, and the latter one is shown to give the optimal average-output-SINR. Iterative algorithms based on second-order cone programming (SOCP) are proposed to solve these optimization problems.