Study On Key Techniques Of MB-OFDM UWB Systems

Ultra Wideband (UWB), which has been speed up to develop in recent years, is a novel technology and could be applied to future wireless communication systems. The UWB wireless communications technique is outstanding of high data rates, low power consumption, low equipment cost, and high positioning accuracy, so it is very suitable for the construction of wireless personal area network (WPAN) in the future. Compared with the wireless communication systems in existence, the UWB systems have higher data rates and bandwidth, in addition, their transmit power are restricted strictly, all of these lead to some technical difficulties to be solved. In UWB wireless communication field, there are mainly two schemes at present: DS-CDMA based UWB and MB-OFDM based UWB. In this dissertation, some key techniques of MB-OFDM systems are discussed:1. Study on channel estimation and diversity combining for MB-OFDM systems.Some channel estimation methods and diversity combining techniques that could be used in MB-OFDM systems are investigated. By using the characteristics of subcarriers diversity provided in MB-OFDM system, an adaptive joint equalization scheme is proposed. The proposed scheme can implement frequency-domain adaptive equalization and maximum-radio combining (MRC) jointly without weighting the diversity branches that is essential in tranditional schemes. Firstly, it's proved that the minimum mean-squared error (MMSE) combining is equivalent to MRC when there is no inter-symbol interference, so it can be completed with the adaptive algorithm in accordance with MMSE criterion. Then, it's found that the tranditonal LMS algorithm is not complied with the homogeneity and the overlap-and-add principle, as a result, it converges slowly when the input signals are highly correlated. Some constraints are also derived, under which the LMS algorithm converges to the optimum weight values directly, and the converged weight values have both the function of equalization and MRC. The derived analytical expressions indicate that the convergence behavior of the proposed algorithm is only governed by the normalized step size, regardless of the input signal correlation matrix. The theoretic conclusions are verified by the simulation results.2. Study on the bit and power allocation algorithms for OFDM systems. The bit and power allocation algorithms aiming at maximization the data rate in OFDM systems are investigated. Under the constraint of integer bit and the requirment of low computational complexity in MB-OFDM systems, an optimum algorithm named as bit-water-level algorithm is proposed. Firstly, the bit-water-level is defined and its property is proved, then principle and formulations of the algorithm are obtained. It's theoretically proved that the algorithm is optimum and can achieve any possible results. Also, it's found that the algorithm has a similar computational complexity with the suboptimum algorithms of iteration water-filling algorithm or improved iteration water-filling algorithms, and need less number of computations in practice. The algorithm can also be adapted to minimize the transmit power. It's confirmed in simulations that the algorithm yields the same bits and power allocation results with the Greedy algorithm.3. Study on the problem of narrowband interference in MB-OFDM systems.Methods of narrowband interference (NBI) suppression are researched. Firstly, several suppression schemes accomplished recently are discussed, and then some foundational methods and their characteristics are analyzed and compared. Because the subcarriers elimination method is suggested to use for NBI suppression by the MB-OFDM scheme, the simple and accurate NBI detection method is needed. A detection method based on sequences construction is proposed. In this method, different sequences are constructed and different signal to interference and noise ratio (SINR) detection formulas are derived for pilot subcarriers and data subcarriers respectively. The method has a low computational complexity and its estimation results are not influenced by the channel parameters. Simulation results show that the SINR can be estimated with an error not exceeded±0.8dB.