Seamless Adaptation For Wireless Communication

The adaptation in current wireless communication system suffers from the follow-ing two key problems.· For binary data transmission, it is difficult to achieve high bandwidth efficiency in current physical layer due to the following three reasons. First, the performance of conventional rate adaptation heavily depends on the accuracy of channel esti-mation, while the accuracy and instancy of channel estimation cannot be achieved at the same time. Second, the rate cannot be continuously adjusted because of the limited number of available rates. Third, there exists a common assumption that all the data from upper layers are compressed already, which contradicts with the practical measurements.· As a special type of binary data, the wireless video streaming needs both video coding rate control and physical layer rate adaptation. The independent design between these two modules leads a lower transmission efficiency. Moreover, in multiple antenna video multicasting system, the channel condition and the num-ber of antennas at the receiver end are heterogeneous. The worst channel con-dition and the least number of antennas will become a bottleneck of the system performance.In order to tackle the above issues, this thesis proposed a novel concept seamless adaptation. In seamless adaptation, the information transmitted over wireless channel can be automatically, continuously and smoothly adapted with concurrent channel con-dition. For different type of data and applications, this research studied how to achieve seamless adaptation and the proposed techniques constitute the main contributions to the field of wireless communication.· For the seamless adaptation of binary data, this thesis proposed random projection (RP) coding and designed rate compatible modulation (RCM) and compressive coded modulation (CCM). First, the sender uses a fixed modulation to continu-ously and progressively generate coded symbols for modulation. The receiver accumulates the symbols until the bits can be recovered and delivers an ACK to the sender. Hence, the channel estimation and its feedback is unnecessary for rate selection. Second, RP code is a kind of rateless code. The transmission rate can be continuously tuned. Last but not least, sparse binary data can be compressed by an optimized RP code. Therefore, the system performance can be boosted by utilizing the potential redundancy in binary data. · For the seamless video adaptation, we proposed image processing based linear video transmission techniques, Cactus and AirScale. By linear video transmis-sion, we mean that after prediction, transform and power allocation, the video is directly transmitted without any non-linear processing, such as quantization, entropy coding and channel coding. The channel noise lies in the same arith-metic domain with the energy scaled data, the reconstructed video quality through MMSE detection can be continuously and smoothly adapted with the channel con-dition at the receiver end. Therefore, it avoids the problems in video coding rate control and physical layer rate adaptation.· In order to solve the heterogeneity in multi-antenna video multicast system, this thesis proposed multiple similar description coding (MSD) and multiplexed space time block coding (M-STBC). MSD generates multiple similar descriptions about the video. Based on conventional STBC, which transmits different versions of the same data in different time slots, M-STBC transmits the similar descriptions in different time slots. The common part in the similar data can be recovered even by the receiver with less antennas. While the difference part can be figured out by the receivers with more antennas. Therefore, the video quality at a receiver end can be seamlessly adapted not only with channel SNR but also with the number of effective antennas.· This thesis also studied the robustness of linear video transmission over fast fading channel and found that the resource allocation is an NP-hard problem. Through the analysis of the performance in fast fading channel, we proposed a fast algo-rithm to optimize the power and bandwidth allocation which improves the robust-ness to fast fading channel.