Signal design for Multiple-Antenna Systems and Wireless Networks
This dissertation is concerned with the signal design problems for Multiple Input and Multiple Output (MIMO) antenna systems and wireless networks. Three related but distinct problems are considered.
The first problem considered is the design of space time codes for MIMO systems in the case when neither the transmitter nor the receiver knows the channel. We present the theoretical concept of communicating over block fading channel using Layered Unitary Space Time Codes (LUSTC), where the input signal is formed as a product of a series of unitary matrices with corresponding dimensionality. We show the channel capacity using isotropically distributed (i.d.) input signaling and optimal decoding can be achieved by layered i.d. signaling scheme along with a low complexity successive decoding. The closed form layered channel capacity is obtained, which serves as a design guideline for practical LUSTC. In the design of LUSTC, a successive design method is applied to leverage the problem of optimizing over lots of parameters.
The feedback of channel state information (CSI) to the transmitter in MIMO systems is known to increase the forward channel capacity. A suboptimal power allocation scheme for MIMO systems is then proposed for limited rate feedback of CSI. We find that the capacity loss of this simple scheme is rather small compared to the optimal water-filling solution. This knowledge is applied for the design of the feedback codebook. In the codebook design, a generalized Lloyd algorithm is employed, in which the computation of the centroid is formulated as an optimization problem and solved optimally. Numerical results show that the proposed codebook design outperforms the existing algorithms in the literature.
While it is not feasible to deploy multiple antennas in a wireless node due to the space limitation, user cooperation is an alternative to increase performance of the wireless networks. To this end, a coded user cooperation scheme is considered in the dissertation, which is shown to be equivalent to a coding scheme with the encoding done in a distributive manner. Utilizing the coding theoretic bound and simulation results, we show that the coded user cooperation scheme has great advantage over the non-cooperative scheme.
Advisor:Heung-No Lee; xinfu chen; Zhi-Hong Mao; Robert Boston; Mickle Marlin
School:University of Pittsburgh
School Location:USA - Pennsylvania
Source Type:Master's Thesis
Date of Publication:09/25/2007