Research On Universal Receiver Architecture And Algorithm For Multi-Mode Wireless Communication Terminal

The evolution of wireless communication has gone through more than thirty years from the early-stage analog 1G wireless communications to the modern digital 2G/3G/4G and the on-going 5G.Wireless communication brings convenience to human'life and has great impact on human's habits with technology upgrading.Due to historical and technical reasons,the 2G,3G and 4G standard coexist in current wireless network,which causes diversity of specifications with different frame structures and results in differences in base station,handset terminal and operators.These differences bring extreme mankind inconvenience to purchase and use handset and change operator.Under this circumstance,demand for multi-mode handset has come into being.Previously,most multi-mode chips adopt independent single-mode method,which only supports one communication standard with obvious drawbacks in large size,high power consumption,high cost,etc.Chip companies integrated these single-mode chips into one chip by using technologies(e.g.,System in Packet and System on a Chip)which can only reduce a little die size.In order to solve these problems thoroughly,it is necessary to solve the convergence problem of different algorithms among different modes at the beginni60ng of the chip design.Some research institutions focus on configurable receiver and wide-band receiver used in the multi-mode RF part,while the receiver back-end demodulation module has less research and is limited to the fusion between two standards.This dissertation proposes one general architecture for equalizer and channel estimation to diminish the chip die size and power by reducing the implement complexity of the receiver algorithm.The architecture,which meets the needs of different traffic channels in multi-mode systems for different times of urgency,is acceptable to all time-domain system and shares part hardware with frequency-domain system.Meanwhile,key algorithm challenges in the implementation of the general architectures for the inner part,including channel estimation,equalizer,interference cancellation are presented..Under the multi-cell and/or user interference scenario,the interference cancellation for channel estimation and equalizer model are also analyzed and discussed in this dissertation.The major contributions are summarized as three-fold:1)Different kinds of frame structure and multi-access induce different implementation for channel equalizer algorithm.This dissertation proposes a general equalizer algorithm for the receiver of time-domain multi-mode system(including GSM/EDGE/CDMA/TD-SCDMA/CDMA2000/WCDMA),namely,the auto-switch method between JMMSE-DFE-OSR and Notch-IC-RAKE.The auto-switch method is considered in order to keep the system performance robust.The receiver mode would switch among RAKE,Notch-IC-RAKE and JMMSE-DFE-OSR varies with the different circumstances(i.e.the channel multipath,the signal quality and the modulation),and requirements(i.e.time response and performance of the control channels).The implements fulfill a major of performance improvement except that,under a few certain situations,the system performance would be slightly weakened2)Considering that the channel estimate algorithm varies from different frame structures,this dissertation proposes a general channel estimation method for multi-mode including GSM/EDGE/TD-SCDMA/CDMA2000/WCDMA,which applies different parameters,i.e.relevant length and cell number,to fulfill the ability to support a variety of modes.Furthermore,the algorithm is designed automatically adjustable when configuring to support multi-cell channel estimation.The algorithm applies the iteration method instead of the matrix inverse method to reduce the algorithm complexity.This method introduces one new training sequence combined the training sequence and take the shape filter coefficient into consideration,achieving the improvement of the precision and the acceleration of the converge speed.Moreover,in order to enable this method suitable in the multi-cell scenario,the multi-path would have the cell information into consideration.Compared with the single model system,the proposed method has the equivalent performance as well as a less complexity.3)A fusion scheme of sharing FFT/IFFT and other modules to reduce hardware resources is proposed when considering the fusion of 4G frequency domain modes with other time-domain modes.The 4G mode is hard to be fused with the time-domain system on the perspective of the physical layer.In order to reduce the hardware source,this dissertation discusses the amalgamation solution as follows:the JMMSE-DFE-OSR is implemented in frequency domain to enable the 4G mode can be shared the FFT/IFFT model with 2G/3G model.In addition,other hardware accelerator modules such as sub-modules and CORDIC can also be shared.The pre-coding in TD-LTE system is studied and a non-codebook pre-coding scheme with the upper bound of the character error rate as the optimization criterion is proposed.This scheme can guarantee fast convergence of non-codebook pre-coding algorithm and maintain less computational complexity.The common algorithm proposed in the dissertation uses MATLAB,C ++ and FPGA to realize its floating point and fix point platform.Compared with the former single-mode system,the proposed scheme is evaluated in terms of performance and complexity.In terms of single-mode,other modes are more complicated except the GSM/EDGE mode;however,from multi-mode perspective,the complexity of all modes is significantly reduced.From the point of performance,these simulation results show that GSM system has the biggest performance loss(about 0.3dB)under sensitivity scenario,but can still pass the specification requirement,and TD-SCDMA system has little performance loss(less 0.1 dB)under sensitivity scenario.Except for these mentioned modes,some of these remaining modes have no performance loss,and another have performance improvement.From the analysis,the conclusion can be obtained that this multi-mode architecture and algorithm can met the specification requirement,and reduce the complexity at the same time,which can further obtain lower cost and lower power multi-mode baseband chip.As a novel reference,the solution putting forward a new direction of expansion provide a possible scheme with lower cost and power for the chip manufacturers.