Study On Channel Estimation Algorithm In OFDM-UWB Wireless Communication System

Ultra-wideband communication technology with high data rate, high capacity,low lost and resistance to multipath meet the requirements of ideal wirelesscommunication system, so UWB technology becomes a hotspot in the research and isconsidered as one of the key technology of the future communication. UWB wirelesscommunication have some characteristics which traditional communication can’tcompared. In the industry, Ultra-wideband communication technology are dividedinto multi-band orthogonal frequency division multiplexing (MB-OFDM UWB)scheme which uses carrier modulation and dieect sequence code division multipleaccess (DS-CDMA UWB) scheme which uses pulse radio.Because Channel estimation should be done before correlation detection,demodulation and equalization in wireless communication system, studying onchannel estimation algorithms has important significance. There are three kinds ofchannel estimation technology: The first one is non-blind channel estimation whichuses pilot through the whole course, the second is semi-blind channel estimationwhich uses pilot at the beginning and the third is blind channel estimation which doesnot use pilot. In this thesis we focused on channel estimation in OFDM-UWB systembased on training sequence (non-blind channel estimation). In this paper, theinformation for channel estimation can be got from two training sequence in one band.Channel estimation algorithm based on time redundancy (TR) has been analyzed.Improved algorithms based on TR with narrowband removed for CM1, CM2and withodd-even pilot for CM3, CM4are proposed. The DFT algorithm based on LSalgorithm can reduce noise by adopting rectangular window whose length is equal toprefix cycle to deal with the channel impulse response, but it doesn’t take into accountthe characteristics of channels. The channel estimation algorithm based on channel’sstatistical properties is put forward. It can effectively reduce the influence of noise,save signal energy and get better performance.At first, we introduced the LS estimation algorithm using time redundancy andits improved algorithms. In OFDM-UWB wireless communication system, twotraining sequences are sent in each sideband for channel estimation. The algorithmsends data only to the odd or even number sub-carriers and the corresponding even orodd sub-carriers are set to zero. In the receiver, interpolation algorithms are applied toget the whole channel impulse response. In the channel CM1and CM2, theperformance of this algorithm is superior to traditional LS estimation algorithm, but inchannel CM3and CM4, the performance of this algorithm is bad due to its strongly frequency selective fading. For CM3and CM4, channel estimation algorithm basedon odd-even training sequences is proposed. This algorithm can better resistfrequency selective fading. In addition, narrowband interference (NBI) existsinevitable. A channel algorithm that can remove NBI is proposed. It should constructtwo training sequences with opposite, both in amplitude and phase infrequency-domain. The simulation results show that the performance ofTR-ASC-based algorithm for CM1and CM2and the performance of TR-EOPalgorithm for CM3and CM4are better than that of the TR-based algorithm.Because there is no use of the correlation between frequency domain and timedomain in LS estimation, it is simple with poor performance. DFT-Based channelestimation algorithm can improve system performance in traditional OFDM wirelesscommunication system. DFT-based channel estimation based on LS algorithm adoptsrectangular window whose length is equal to prefix cycle to deal with the channelimpulse response in time domain. It can reduce the noise outside the window, so it canimprove performance. Due to the intensive multipath in OFDM-UWB channel anddifferent lengths of channel impulse for four channel surroundings, using rectangularwindow that length equal to prefix cycle can’t reduce noise and keep signalsufficiently, so it has no good performance. In the paper, I will do statistics for thefour different channels and construct the corresponding window function. Using thenew window to deal with the channel impulse response in time domain can reducenoise as much as possible. It can reserves signal energy as much as possible by gettingall multipath meetting certain standards when do the statistics and including all mult-ipath that energy attenuation of which is less than15dB when construct windowfunction. Beacause the channel estimation based on CSP can not only reduce noise butalso keep signal energy, performance of CSP-based algorithm is better than that ofDFT-based algorithm. The simulation results show that the performance of CSP-basedalgorithm has better performance.