Peak Power Optimization Schemes For Multicarrier Communication Systems

To meet the needs of future wireless communications and realize high-speed data transmission, multicarrier communication technologies have attracted much attention due to the immunity to multipath fading and the high spectral efficiency. As the representative of multicarrier communication technologies, orthogonal frequency division multiplexing (OFDM) has been adopted in many fields of wireless communications because of the robustness to frequency-selective fading and the high-rate data transmission, and it has become one of core techniques for4G. Moreover, system performance can be further improved by combining OFDM and other technologies. For example, multiple input multiple output OFDM (MIMO-OFDM) can increase both system capacity and spectral efficiency effectively.Although OFDM has many advantages, peak power problem is always one of its main drawbacks. Especially, when the number of subcarriers in OFDM systems is large, this problem becomes considerably serious, which can be described by peak-to-average power ratio (PAPR). As a result, the high PAPR brings on OFDM signal distortion, and the signal distortion induces system performance degradation. This situation requires that the linear region of high power amplifier (HPA) in the transmitter must be very large. However, the implementation of hardware system is very hard, and its cost is also very high. On the other hand, because MIMO-OFDM belongs to one of multicarrier communication systems, peak power problem is also its shortcoming.In this dissertation, the basic principles of multicarrier communication systems and the peak power problem are firstly introduced. Then, based on the research status of this problem, we concentrate our efforts on the research of peak power optimization in OFDM and MIMO-OFDM. The main contributions are summarized as follows:(1) Partial transmit sequence (PTS) is an effective scheme for PAPR reduction in OFDM systems. However, two important shortcomings should be solved in this scheme: large computational complexity and the need of side information. To reduce computational complexity of conventional PTS, two phase weighting methods with low computational complexity for PTS, named grouping phase weighting (GPW) and recursive phase weighting (RPW), are proposed. In PTS with GPW, all the subblocks are divided into several groups, and in each group, its own phase weighting process can be implemented and its own subcandidate signals can be obtained. Finally, subcandidate signals from different groups can be combined by employing complex additions to achieve all the OFDM candidate signals, and the one with the lowest PAPR is selected for transmitting. In PTS with RPW, by making use of the relationships among phase weighting sequences, the common terms can be found to simplify the computation for most of candidate signals, which can result in computational complexity reduction. At last, the candidate signal with minimum PAPR is selected for transmitting. Compared with conventional PTS, these two proposed schemes can not only dramatically reduce computational complexity but also have an advantage of no loss in PAPR reduction performance. Moreover, the combination of these two proposed schemes can obtain further computational complexity reduction. That is to say, all the subblocks are firstly divided into several groups, and subcandidate signals in each group can be achieved by employing RPW. Lastly, subcandidate signals from different groups can be combined by using complex additions to obtain all the candidate signals.(2) For OFDM systems employing phase shift key (PSK), two scrambling schemes for PAPR reduction are proposed, i.e., scrambling scheme based on selected mapping (SLM) and scrambling scheme based on information hiding. In the scrambling scheme based on SLM, instead of phase weighting process employing complex multiplications in conventional SLM, simple scrambling employing the logical NOT operation is adopted. Meanwhile, by making use of signal constellations in different shapes, the side information can be reliably embedded in the transmitted signal, which avoids transmitting explicit side information. Compared with optimal SLM, the proposed scrambling scheme based on SLM can obtain the same PAPR reduction performance. In the scrambling scheme based on information hiding, the hided binary sequences are viewed as the scrambling sequences, and the scrambling process can be implemented by changed mapping, which improves PAPR reduction performance clearly. Most importantly, the original demodulating process of OFDM is still adopted, and no side information needs be transmitted to the receiver.(3) Because conventional block interleaving is implemented on one OFDM signal, the correlation between candidate signals is strong, which has a negative impact on the improvement in PAPR reduction. Concerned on the limitation of conventional block interleaving, time-frequency block interleaving schemes for improving peak power in OFDM are proposed. In the proposed schemes, block interleaving is adopted among different OFDM signals in a timeslot, which improves the freedom of generating candidate signals and reduces the correlation between different candidate signals. Compared with conventional block interleaving, the proposed schemes can obtain better PAPR reduction performance with the same computational complexity.(4) For MIMO-OFDM systems, PTS can be easily applied to each transmitting antenna in a straightforward manner, called ordinary PTS (oPTS). However, large computational complexity is still its main drawback. To reduce computational complexity, a cooperative PTS (co-PTS) scheme is proposed for MIMO-OFDM. In co-PTS, alternate optimization is firstly employed for reducing computational complexity, and then the use of spatial subblock circular permutation is able to increase the number of candidate signals indirectly, which equivalently improves the performance for PAPR reduction. Compared with oPTS, co-PTS can reduce computational complexity dramatically and obtain better PAPR reduction performance.In the last part, the problems to be solved in this dissertation and the future research work are summarized.