Research On Land Mobile Satellite Channel Modeling And Long-range Prediction At S-band

With the rapid improvement of the transmission rate for communication systems and the high-quality requirements of multimedia service from growing users,satellite mobile communication,an important complement to 4G and future 5G land communication networks as well as an indispensable part of the air-space-ground information networks,needs to improve the transmission rate and spectral efficiency.Long-range prediction is a driving factor of fading-compensation techniques and has capacity to achieve a perfect trade-off between power efficiency and spectral efficiency for land mobile satellite system(LMSS)adaptive transmission schemes.LMS channel has openness,randomness and complicated variability samilar to satellite channel and land channel,it would tend to seriously affect the reliability of transmitted schemes due to ionospheric scintillation,multipath fading,shadowing fading together with Doppler effect.In addition,LMS channel may become instable given that ever increasing transmission rate.Therefore,it is an important research topic to model the propagation characteristics of LMS channel and study long-range prediction of fading signals.Against this background,LMS channel models,capable of describing shadowed conditions for different environments,different satellite elevation angles and different azimuth angles,are firstly established.On the basis of models,the long-range prediction schemes are researched in the dissertation.Firstly,this dissertation overviews the propagation characteristics of radio in LMS channel,analyzes the first-and second-order statistics of fading signals,and establishes simulation methods and modeling ideas.On the basis,single-state C.Loo model along with Corazza model,and two-state Lutz model is furture researched.LMS channel state model and adaptive long-range prediction schemes based on three-or two-state LMS channel model are investigated intensively.The main contributions of this paper are organized as follows:Secondly,state modeling of narrowband LMS channel at S-band is presented.Given that single-and dual-satellite reception systems,an exhaustive analysis on four narrowband LMS channel measurement scenarios is provided.Considering multi-state transitions process for the single-satellite case,a state modeling of narrowband LMS channel suitable for dual-satellite reception is proposed based on first-order,discrete-time Markov chain.Though the joint state probabilities and correlation coefficients can be represented accurately under various environment types,elevation and azimuth angles,the proposed modeling still exists limitations in state duration and its statistics.To improve the accuracy of state modeling further,the semi-Markov dual-satellite state modeling is proposed by the approximation of the state duration probability density function(SDPDF)for each state using lognormal distribution.The results from numerical simulation have shown that both of the proposed modeling are closer to that of measured data in terms of state probabilities,correlation coefficients and state duration statistics.The semi-Markov modeling ensures a correct SDPDF,on the condition that state transitions depends directly on its duration.On the contrary,the Markov modeling only depends on given state length that conducives to realize the long-range prediction of future channel state.Thirdly,the adaptive long-range prediction(ALRP)schemes of fading signals are researched.The LMS channel in single-satellite reception case has the variable shadowing conditions and large delay.It's a tough work to realize fading signals long-range prediction when the algorithms of land mobile channel are applied to the LMS channel.In order to deal with this problem,ALRP schemes are proposed based on LMS channel models.The scheme models the propagation conditions at S-band as three-state LMS channel model with the given model parameters per state.The channel states and fading signals within future long-range are predicted by combing normalized weighted idea with autoregression(AR)model after generating the channel theoretical values.Model parameters are updated by employing least mean square adaptive tracking method.Subsequently,the ALRP algorithm is applied in the versatile two-state LMS channel model to resolve the complex and limitation of three-state model.This proposed prediction scheme uses "good" and "bad" states to represent a range of shadowing respectively,and updates model parameters per state by joint distribution.Simulation results prove that the proposed schemes can achieve a reliable prediction for channel state and fading signals compared with long-range prediction(LRP)scheme,while also having prominent advantages of real-time and low-complexity.Finally,alternating variable step-size adaptive long-range prediction(AVSS-ALRP)scheme of fading signals is investigated.To solve some issues,such as sensitivity to channel model,a wide dynamic range of received signal as well as stable AR model step-size in ALRP scheme,AVSS-ALRP scheme is proposed based on two-state LMS channel model suitable for more scenarios.The proposed AVSS-ALRP scheme enhances versatile two-state LMS channel model.The main differences include two aspects;one is to perform the interpolation and rate conversion before a conversion into lognormal distributed samples and the other is to introduce azimuth angles relative to the driving directions only in slow fading components.ALRP algorithm combined with normalized weighted prediction has good performance gain,but this is accompanied by a loss in prediction performance due to low correlation of theoretical values,abrupt change of shadowing states and stationary step-size parameter.Therefore,the proposed scheme smoothes the channel fading signals to get prediction samples,and introduces an eigendecomposition idea to predict channel states.Based on variable step-size adaptive long-range prediction(VSS-ALRP)algorithm,the future fading signals are predicted well if the current shadowing condition corresponds "good" state.Otherwise the predicted results will be modified by combining with adaptive prediction of predicted errors within "bad" state.The simulations demonstrate the superiority of the proposed scheme in complexity and prediction performance.