The use of multiple antennas in wireless communication systems can significantly increase bandwidth efficiency. Many different techniques have been developed to take advantage of the added dimension provided by the multiple transmit and multiple receive antennas. In particular, space-time codes are 2-dimensional forward error correcting codes designed specifically for multi-input multi-output (MIMO) channels created by multiple antennas. In this thesis, we propose several methods to enhance space-time code design. In particular, we study the design criteria for good space-time codes and propose various approaches to design 2-dimensional space-time codes based on traditional 1-dimensional codes....[ Read more ]

The use of multiple antennas in wireless communication systems can significantly increase bandwidth efficiency. Many different techniques have been developed to take advantage of the added dimension provided by the multiple transmit and multiple receive antennas. In particular, space-time codes are 2-dimensional forward error correcting codes designed specifically for multi-input multi-output (MIMO) channels created by multiple antennas. In this thesis, we propose several methods to enhance space-time code design. In particular, we study the design criteria for good space-time codes and propose various approaches to design 2-dimensional space-time codes based on traditional 1-dimensional codes.

We consider a generalized layered space-time architecture with parallel space- time codes. It had been shown that using a pre-fixed layer decoding order with appropriate power levels at different layers could improve system performance. We derive an optimal power allocation scheme over the space-time code layers to 1minimize the overall frame error rate. We show that the optimal power allocation scheme should equate the derivatives of the frame error rate functions of all the space-time codes. Moreover, the gain is achieved by simply using the optimized parameters with no increase in complexity.

We also study the design of space-time codes for systems with large number of antennas in quasi-static Rayleigh fading environment. We prove that the interference between signals transmitted on different transmit antennas vanishes as the number of receive antennas approaches infinity. This result suggests that the code design can be largely simplified by dividing the antennas into groups with each group using a small space-time code. This design can be further improved by adding an outer code to utilize the full range of antenna diversity.

Finally, we design code structures that can achieve the error probability lower bound attainable by all space-time codes with a given minimum Euclidean distance. We propose two transformation structures that generate 2-dimensional space-time codes using 1-dimensional codes with given minimum Euclidean distance. Also, we show that these transformation structures are asymptotically optimal as the frame size goes to infinity. To achieve this bound, we propose a space-time code based on a conventional code such as convolutional code together with temporal interleaver and unitary matrix spatial transformation, and show that this code has performance close to the optimal error probability bound.