Research On Efficient Channel Estimation And Equalization Algorithms For Uplink NB-IoT Systems

The Internet of Things(IoT)is a novel paradigm that can facilitate the massive connectivity of smart devices,such as sensors,actuators,radio frequency identification(RFID)tags,and consumer electronics to the Internet for revolutionizing our living and working systems.In the near future,IoT enables a massive number of heterogeneous and geographically distributed things/objects will be connected to the Internet through a common addressing scheme,resulting in the evolution of new intelligent service systems available worldwide.However,various existing and emerging wireless communication technologies are facing a lot of challenges to realize the vision of IoT framework,including extended coverage,high power efficiency,low data rates,and delay sensitivity.In recent years,many standardization bodies provided tremendous efforts to accommodate such requirements of IoT applications in current cellular networks.Specifically,a new cellular-based low power wide-area(LPWA)IoT enabling technology named narrowband IoT(NB-IoT)has been standardized by the 3rd Generation Partnership Project(3GPP)as part of release-13 in middle of 2016.The NB-IoT aims to reuse radio frequency(RF)baseband processing and numerology of existing Long Term Evolution(LTE)systems with restricted scope.Coverage enhancement and accurate decoding of NB-IoT user equipment(NB-IoT UE)data mostly depend on the quality of employed channel estimator and equalizer during transmission.Therefore,the selection of channel estimation and equalization methods will significantly affect both the hardware complexity and system performance.This research addresses two crucial aspects of uplink NB-IoT systems:channel estimation and equalization.For the channel estimation perspective,various novel algorithms are presented and verified that exploit the simplified numerology of NB-IoT and combine reduced complexity with reasonable performance.For the channel equalization aspect,the focus is on the frequency-domain interference-free one-tap equalizer that considers the mismatched sampling rate and RF baseband processing between the NB-IoT UE transmitter and LTE base station(BS)receiver.Firstly,the narrowband demodulation reference signal(NDMRS)-aided frequency-domain low complexity effective channel estimation algorithms are proposed,modifying the conventional least squares(LS)and minimum mean square error(MMSE)algorithms.These proposed methods aim to minimize the difficulty of realizing the traditional LS and MMSE estimators in the newly standardized uplink NB-IoT systems as minimum as possible.In order to achieve this goal,the ill-conditioned matrix problem is addressed by adding a normalization matrix in the conventional LS algorithm.Then,the real-time matrix inversion problem of channel auto-covariance matrix is eliminated with the identity matrix in the traditional MMSE algorithm.Secondly,improved discrete Fourier transform(DFT)-based low complexity channel estimation algorithms are proposed named random sorting LS(RS-LS)and de-noising LS(D-LS),stemming from the simple LS method.Another sub-optimal estimator based on the filtered channel estimate called MMSE-Approximation(MMSE-A)is also studied.The initial channel frequency response(CFR)is estimated at pilot frequencies using the original LS method,and then,apply several additional operations in the transform-domain to suppress LS estimation error without exploiting extra frequency-band resources and increasing computational complexity.Then,an efficient time dimensional linear interpolation method is defined to estimate the CFR for the remaining orthogonal frequency-division multiplexing(OFDM)symbols within an NB-IoT uplink subframe.Thirdly,the signal aliasing and border effect problems experienced in DFT-based transform domain methods are addressed by employing modified discrete cosine transform type-I(DCT-1)based channel estimation approaches.In order to overcome the limitations of the classical DFT and DCT-I based methods,three options of DCT-I are proposed by modifying the original definition of DCT-I.These three modified DCT-I(MDCT-I)approaches can reduce the high-frequency distortion and aliasing error in the time-domain when non-sample-spaced path delays exist in multipath fading channels.Then,the proposed methods are applied to the initial LS estimates to suppress estimation noise as well as improve the radio coverage without increasing significant computational complexity,compared to the conventional DFT and DCT-I based methods.Furthermore,the proposed MDCT-I scheme is realized using a fast DCT algorithm,which has fewer computational steps than a fast DFT algorithm.Finally,the issue of channel equalization is addressed for the LTE-based uplink NB-IoT systems,which has mismatched RF baseband processing and sampling rate between the transmitter and receiver.A relation between the channel impulse response and channel frequency-domain equalization coefficients is derived under imbalanced baseband processing and sampling rate environment.Consequently,a novel one-tap frequency-domain channel equalization algorithm is explicitly proposed for uplink NB-IoT systems based on the resultant channel coefficients.Furthermore,the performance and effectiveness of the proposed channel estimation and equalization algorithms are investigated within the system context of the 3 GPP NB-IoT standards,compared with state-of-the-art algorithms through extensive numerical simulations and computational complexity analysis.