Efficient Channel Estimation Algorithms For Downlink Narrowband Internet Of Things Systems

The Internet of Things(IoT)belongs to an innovative networking platform that revolutionizes our daily lives and workplaces by enabling the widespread Internet connection of everyday smart devices including actuators,sensors,consumer electronics,and radio frequency identification(RFID)tags.By allowing a significant number of geographically dispersed and heterogeneous things/IoT objects to link to the Internet using a unified addressing approach in the near future,new intelligent technological systems that are accessible everywhere will develop.Nevertheless,several present and emerging technologies for wireless communication have numerous challenges in realizing the goal of the IoT framework such as low data rates,expanded coverage,delay sensitivity,and excellent power efficiency.Many standardization groups have worked tirelessly in recent years to conform cellular networks to the needs of IoT network technologies.Particularly,the Third Generation Partnership Project(3GPP)standardized a revolutionary and unique IoT technology named narrowband IoT(NBIoT)as part of release-13 in the middle of 2016.For integrating the IoT into pre-existing cellular communication infrastructure,this new low-power wide-area(LPWA)network technology seems promising.The philosophy of developing NB-IoT is to reutilize the baseband numerology and radio frequency(RF)in a restricted and simplified manner from the legacy Long Term Evolution(LTE)communication technology.Importantly,the accuracy of channel estimators used during signal transmission is crucial for both improving coverage and correctly decoding of data from the user equipment of NB-IoT(NB-IoT UE).Consequently,the choice of channel estimation technique will have a substantial impact on both the system performance and hardware complexity.This research focused on this essential feature of "channel estimation" in downlink NB-IoT systems.From this viewpoint,several novel methods of channel estimation are proposed and validated that make use of simplified NB-IoT numerology in order to attain a proper balance between complexity and performance.Firstly,the narrowband reference signal(NRS)assisted efficient frequency dimension low complexity channel estimation method named narrowband least square(NB-LS)is presented based on the modification of the traditional least square(LS)algorithm.This proposed method aims to minimize the difficulty of severe noise acceptability of the traditional LS channel estimator in the newly standardized downlink NB-IoT systems and keeps the complexity as lowest as possible.In order to accomplish this target,the additional channel taps that consist of only noise are eliminated instead of considering all the channel taps by adding a novel threshold value which is developed based on the power of the channel noise.Secondly,two hybrid and reduced complexity channel estimation methods are proposed based on the maximum likelihood estimator(MLE)and 2D Wiener filtering method.The MLE is simple in complexity but subjected to low estimation precision,whereas well-known 2D Wiener filtering offers efficient performance to estimate the channel but entails severe complexity.The application of two 1D Wiener filters(for frequency and time domain)can reduce its complexity but degrades the performance.In this research,the time domain 1D Wiener filter and MLE in the frequency domain are exploited in two different orders.As the first approach,the time domain Wiener filter is exploited to compute all the estimated channel data for several subcarriers in entire OFDM symbols.Subsequently,the MLE algorithm is employed in the frequency domain to compute the entire channel matrix of all the OFDM symbols.In the second approach,the frequency domain MLE algorithm is applied for computing the estimated channel response of several OFDM symbols having NRS sequences,followed by the time domain Wiener filter application to obtain the whole channel estimation.Both proposed methods have a good trade-off between complexity and system performance.Thirdly,a modified reduced complexity linear minimum mean square(LMMSE)channel estimator is presented in order to accomplish a proper balance between performance and complexity in the channel estimation for the downlink NB-IoT systems.Basically,this estimator addresses the severe computational complexity problem of the optimum LMMSE estimator.To reach the goal,the optimal rank reduction formulae named singular value decomposition(SVD)technique is applied for simplifying the complex channel matrix in the first phase,which minimizes the complex matrix inversion problem in the traditional LMMSE algorithm.Henceforth,the autocorrelation matrix of the channel is split into a few submatrices for an efficient understanding of the channel estate.Lastly,the submatrices are overlapped to optimize the performance of the channel estimation method and to reduce the complexity furthermore.The proposed overlap banded SVD LMMSE consists of mentionable lower complexity in computation compared to the traditional LMMSE by sacrificing a negligible performance degradation.Finally,the performance loss issue of the optimum LMMSE channel estimation algorithm while reducing the complexity for the downlink NB-IoT systems is addressed.In this light,a minimal complexity optimum performance channel estimation technique named narrowband LMMSE(NB-LMMSE)is proposed by applying the Toeplitz matrix property on the complex LMMSE channel matrix.Furthermore,a preconditioning algorithm is applied to make the estimation process faster,which preserves the performance of the optimum LMMSE even after complexity reduction.This channel estimation method exhibits excellent spectral efficiency with the highest estimation accuracy as well as very low complexity.It can preserve the optimum performance in the situation of few received CIR in the very worst channel environment.The results of link level computer simulations are presented to confirm the efficacy of the proposed technique with comparison to their comparable candidates.In addition,a comprehensive analysis of the computational complexity of the suggested channel estimation procedure compared to the state-of-art methods is presented.