Research On MIMO Signal Processing Technology Of Optical MDM System

With the continuous expansion of Internet users and the rapid increase of broadband data services,the data carrying capacity of global backbone communication system needs to be improved urgently.Therefore,research on ultra-large capacity information transmission technology becomes the inevitable trend of future development of optical fiber communication.After growing for over two decades,the capacity of single mode fiber is rapidly approaching the non-linear Shannon limit and is unable to meet the growing capacity demand.At present,the technology of increasing spectral efficiency by adding multiplexing dimension(time,frequency,phase,polarization)has been proposed gradually,but it has been proved that the expanded capacity space is limited.In this context,the mode division multiplexing(MDM)system based on few-mode fiber emerges as the times require.It can multiply the capacity of the system by multiplexing modes and is regarded as a promising technology which is expected to solve the capacity bottleneck of communication system.Therefore,it has attracted wide attention of domestic and overseas scholars.Different spatial modes in MDM system always experience different mode dependent loss,which can destroy the orthogonality of each mode in the transmission process.In addition,random coupling of mode signals occurs along the optical fiber links.These will seriously affect the performance of system.This paper focuses on MDM system and starts from the basic principle of mode division multiplexing to construct a model of MDM system.We emphatically studied how to use MIMO digital signal processing technology to solve mode crosstalk and suppress mode-related losses in MDM system,thus improve the reliability of transmission.The main work is summarized as follows:First,according to the mode transmission theory of few-mode fiber,a long-haul 3?3 MDM system with accumulated mode dependent loss is established and the threaded algebraic space-time coding scheme is employed in the MDM system.Considering the exponential computational complexity of maximum-likelihood detection,a soft-input and soft-output detection scheme based on probabilistic data association method of MDM system is proposed.By using the “iterative Gaussian approximate”,besides,taking the probabilities of the potential candidate symbols as the soft input/output information and using the approximate form of Bayesian formula,we can obtain the decision probability of each transmitted symbol.The simulation and analysis results show that the proposed algorithm can achieve low complexity and approximately optimal detection performance at the same time.Second,aiming at the difficulty in calculating the posterior probability of a transmission symbol by using Bayesian theory,the classic Markov chain Monte Carlo(MCMC)method is applied to the MDM system.The performance of Gibbs sampling based MCMC algorithm in signal detection of MDM system is emphatically studied.The simulation results show that the performance of MCMC algorithm is superior to other sub-optimal detection algorithms.Moreover,the complexity of MCMC is only related to the number of modes and iterations.On this basis,a new MCMC detection scheme based on QRD-M algorithm(QRD-MCMC)is proposed.This scheme combines QRD-M algorithm with MCMC algorithm.With the help of QRD-M algorithm,the high signal-to-noise ratio problem of classic MCMC detection is alleviated.The simulation results show that the proposed QRD-MCMC detection scheme can achieve near maximum-likelihood detection performance and owns lower bit error rate than the classic MCMC algorithm when the increase of complexity can be ignored.