Semi-blind Channel Estimation Of MB-OFDM UWB System

The main description methods of the ultra-wideband (UWB) system are S-V indoorchannel model and IEEE802.15.3a one. However, IEEE802.15.3a indoor channel model isvery complex to be simulated and achieved due to many parameters. So, an UWB frequencydomain parameter model is proposed aimed to the problems above. In this paper, firstgeneralized stationary noncorrelation scattering hypothesis of UWB channel is demonstrated.It shows that UWB channel can be used to describe the autoregression-based channel model.Then, we can reach a conclusion that a precise method of channel description can be achievedwhen the model order is equal to2according to calculating the AIC criterion of the newmodel.UWB system based on multiband orthogonal frequency division multiplexingmodulation (OFDM) is defined in ECMA-368standard where coherent demodulation is usedat the receiver. So, the channel estimation must be introduced in order to provide the channelinformation for the receiver to restore the signal in the UWB system. However, the traditionalmethods of channel estimation are very complex, expensive and sensitive with the noise. Animproved method based on subspace decomposition semi-blind channel estimation isproposed aimed to the problems above in this paper. The key idea of this method is that theminimal eigenvalues of the matrix which is orthogonal with the channel vector arerepresented the influence caused by the noise in this channel. Then, the influence can beremoved by subspace decomposition and the channel estimation accuracy can be improved. Inaddition, a new channel estimation method based on multi-level correction matrix is proposedon the basis of the mentioned method in this paper where the correction matrix is obtained bysubspace decomposition according to the way of using the pilot in subspace decomposition.So, the pilot information can be well optimized to improve the precision of the pilotinformation.