Performance Research Of Massive-MIMO Detection Algorithm In Spatial Division Multiplexing Systems

Spatial-division Multiplexing(SDM)is considered as a potential way to increase the capacity of optical communication systems,which uses the spatial degrees of freedom of Few-mode Fiber(FMF)/Multi-mode Fiber or coupled Multi-core Fiber to transmit datas in propagation mode.The demand for data traffic of global users is increasing at an astonishing rate with the vigorous development of digital communication services,and Single-mode Fiber transmission is far from being able to satisfy the demand for the amount of future information.In this context,the SDM systems have become the focus of research.However,the SDM systems will inevitably generate coupling between different modes and Mode Dependent Loss(MDL)caused by optical devices under longdistance transmission,thereby causing in crosstalk and interference,which seriously affects the increase of system capacity.Therefore,alleviating effectively the influence of MDL is the key to achieve reliable signal transmission.The Multiple Input Multiple Output(MIMO)detection algorithms can separate parallel data streams and reduce effectively the performance impact of MDL in FMF,which improves the reliability of received signals.However,the existing MIMO detection algorithms for FMF will present excessive computational complexity when the number of modes is too high,which leads to high hardware cost and cannot be implemented.Therefore,this thesis proposes to introduce Massive-MIMO detection algorithm technology into the SDM systems to alleviate the performance loss caused by MDL,and main contributions of the thesis are summarized as follows:(1)The SDM fiber link model with D propagation modes is established based on the principle of random matrix transmission,and the system properties such as dispersion,mode coupling and MDL in the SDM fiber channel are introduced.On this basis,the statistical characteristics of MDL and the influence of accumulated MDL on channel capacity of SDM optical communication systems are analyzed by simulation.And the variation trend of MDL with the transmission distance in the SDM systems is shown.(2)The implementation process of five Massive-MIMO detection algorithms are introduced,including Minimum Mean Square Error equalization,Conjugate Gradient detection,Gauss-seidel detection,Optimized Coordinate Descent based Box-constraint(OCDBOX)detection,Alternating Direction Method of Multipliers based Infinity-norm(ADMIN),which provides theoretical basis for the application of Massive-MIMO detection algorithms to SDM optical fiber transmission systems and the improvement of the algorithm.(3)In 32-mode and 128-mode optical communication systems under the presence of MDL,the convergence,complexity and Bit Error Rate(BER)of five Massive-MIMO detection algorithms are studied.The research results show that OCDBOX has the best BER performance,followed by ADMIN detection algorithm for SDM systems with different accumulated MDLs.However,the complexity of OCDBOX detection algorithm is much higher than ADMIN.(4)The ADMIN detection algorithm is improved,and the simulation is carried out in the 128-mode optical communication systems.The results show that the improved ADMIN detection algorithm greatly improves the system performance without changing the computational complexity,and it outperforms the OCDBOX detection algorithm in both complexity and BER performance.In addition,a new Massive-MIMO detection algorithm: Adaptive momentum based box-constraint detection algorithm is proposed and simulated based on the idea of the box constraint and its performance advantages,its performance has been greatly improved compared with the traditional Adaptive momentum detection algorithm.