Training signal design for MIMO one-way and two-way relay systems in the presence of interference

Abstract

In this dissertation, we propose an optimal channel training method in the sense of linear minimum mean square error (LMMSE) for multiple-input multiple-output (MIMO) amplify-and-forward (AF) one-way and two-way relay systems in the presence of interference. In the environment where the interference signals are received from the adjacent cells, there is a need for channel estimation method which avoids such an interference signals and suppresses the influence of interference. Because of the characteristics of the multi-antenna system, the training signal for channel estimation is transmitted to several signal subspaces. However, in the conventional method, it is very difficult to select an optimal signal subspace in which influence of interference signals are minimized. In this paper, we propose a new channel estimation method that determines the optimal signal subspace and power allocation to minimize the influence of interference . In this dissertation, we consider three relaying protocols: one-way relaying protocol, multiple access broadcast (MABC) two-way relaying protocol, and time division broadcast (TDBC) two-way relaying protocol. We propose a channel training method that minimizes the LMMSE between the channel of the link and the channel estimated by each protocol.

Firstly, an one-way relaying is a protocol that a relay transmits a signal received from a source to a destination and is used in the one way relay system. In the one-way relay system with a direct path between the source and destination, the destination must consider both the signal received via the relay and the signal received directly from the source. The interference signals received in each phase and each path should be considered to design the training signal. Particularly, the interference signal received at the relay in the first transmission phase is amplified like the channel training signal and is received at the receiver in the second transmission phase. Therefore an appropriate interference suppression process is required. In Chapter 2 of this dissertation, we propose an optimal channel estimation method in this environment. Specifically, the relay precoding matrix was used to properly control the interference signal, and the channel estimation performance could be improved by allocating more power to the subspace having a good channel gain and a low interference signal power. We also propose a channel estimation method to reduce complexity and show that it is an optimal method for high signal-to-interference and noise ratio (SINR).

Secondly, MABC is a protocol in which a relay re-transmits a signal received from two users at the first transmission phase to two users at the next transmission time, and the two users simultaneously transmit signals through a relay without a direct path. In the MABC two-way relay system, two training signals transmitted from each user at the first transmission phase are received by each user at the second transmission phase, because two users exchange signals at the same time. Therefore, unlike the one-way system, there is interference between the transmission signals of two users. Therefore, it is necessary to appropriately control both the inter-user interference signal and the externally received interference signal. In Chapter 3 of this dissertation, we propose an optimal channel estimation method in this environment. Specifically, the relay precoding matrix was used to properly control the external interference signal, and the training signals of the two users were designed to have orthogonality so that the channel estimation performance could be improved, and we show that the orthogonal training signals is optimal. We also propose a channel estimation method to reduce complexity and show that it is an optimal method for high SINR.

Thirdly, TDBC is a protocol in which a relay re-transmits a signal received from one user at the first transmission phase and another user at the second transmission phase to two users at a third transmission phase. TDBC protocol is widely used for the two-way relay systems with direct path between two users. Similar to the MABC two-way relaying protocol, there are inter-user interference, because two users exchange each signal at the same time. The difference is that two users send and receive signals over the relay during three transmission phase, and there is a direct path between the users. Therefore, the number of received interference signals to be controlled increases, and a direct path channel as well as a channel passing through the relay must be estimated simultaneously. In Chapter 4 of this paper, we propose an optimal channel estimation method in this environment. Particularly, the TDBC two-way relaying protocol can use the two relay precoding matrices for the signals received at the first and second transmission times at the relay. Therefore, it is possible to appropriately control the external interference signals received during the three transmission phases. Also, the training signals of two users are designed to have orthogonality so that the channel estimation performance can be improved. Also, we show that the orthogonal training signals are optimal when the direct path channel is simultaneously estimated. We also propose a channel estimation method to reduce complexity and show that it is an optimal method for high SINR.

Simulation results demonstrate that the two proposed methods outperform the conventional training signal design methods in terms of mean square error (MSE), bit error rate (BER), and spectral efficiency.