A Study Of Space-Frequency Coding And Signal Detection In MIMO-OFDM Systems

The combination of multiple-input multiple-output (MIMO) technology and orthogonal frequency division multiplexing (OFDM) systems can greatly expend the application area of MIMO, and it can work well in frequency selective fading channels. It owns the both advantages of MIMO and OFDM. MIMO-OFDM system can exploit the space and frequency diversity simultaneously to improve the performance of wireless communication system, and achieves the theoretic margin of system. One of the most important ways to achieve this margin is space-frequency coding (SFC).According to the decoding method using in the receiver, we assort the SFC into three types: coherent SFC, noncoherent SFC, and differential space-frequency modulation.In this dissertation, an equivalent signal model of the transmission of space-frequency codes is proposed in the first place. It can be used to transform the transmission model of space-frequency codes under frequency selective fading channels into the model of space-time codes in flat-fading channels, and this simplified the analysis of the performance of space-frequency codes significantly. With this equivalent model, some important conclusions of space-time codes can be extended smoothly into the description of space-frequency coding. This facilitates the prediction of the property of space-frequency coding and help to guide our research, and these conclusions are also verified by the following research and simulation.Based on the equivalent model, a novel analysis method of coherent space-frequency coding is proposed and an expression with excellent form for the performance is obtained. From this expression, the two important indexes of space-frequency codes, diversity order and coding gain, are inferred strictly, and they follow the concept firstly proposed by Tarokh in the research of space-time coding. With these two indexes, the coding criterion and a realization structure are proposed. Simulation results are provided to verify our conclusion.Up to now, the performance analysis of noncoherent space-frequency coding whose receiver does not know the channel state information and the methods to reduce the encoding-decoding complexity have not been solved, and this has constituted a bottleneck of the research of noncoherent space-frequency coding. Here a sub-block noncoherent space-frequency coding is proposed to reduce the size of codeword and the complexity. Using the equivalent model proposed previously, the performance expression of noncoherent space-frequency coding is deduced, and the coding criterion to achieve full diversity and optimize the coding gain is obtained. It also provides an implementary construction of noncoherent SFCs and uses computer simulation to verify our conclusion.The former researches of space-frequency coding are all assume that channel coefficients on the different antenna pairs are mutually independent, but it is not true in reality. Under our analysis of pairwise error probability, we provide some propositions to describe the performance of SFCs (both coherent and noncoherent) designed under the assumption of independent fading but transmitted in a correlated channel. We also point out that how to encode the system under a correlated channel.Differential modulation is also a widely used method to the case of no knowledge of the channels in the receiver. Here we propose noncoherent sequence detection for differential space-frequency modulation, which can be used to alleviate the error floor that occurs in a multipath channel with long time delay expansion. We also propose a performance evaluation method to our detection algorithm. Simulation results show that our algorithm enforces the adaptability of differential space-frequency modulation to the time delay expansion.Finally, we give our work to reduce the complexity of multiple symbol differential sphere decoding in differential space-time modulation. Through the modification of the enumeration strategy of the search tree in sphere decoding, we significantly reduce the complexity of detection at the expense of little loss on performance.