Study Of Detection Algorithms In MIMO-MDM Optical Transmission Systems

Driven by high-definition video streaming, multimedia file sharing, cloud computing, mobile networking, online game and other information technologies, worldwide data traffic is growing at an unbelievable rate. In order to keep up with the increasing traffic demand, today’s researches need to concentrate on technologies that can improve the channel throughput and reduce the cost of the entire process at the same time. In the recent years, techniques in optical networks have already made some big breakthroughs in physical dimensions, including some new types of low-loss fibers and other multiplexing technologies in various physical dimensions. However, studies of current techniques are quite mature, and the capacity of optical fiber communication almost reaches the limit of Shannon. Studies of mode division multiplexing based on few mode fiber develop quickly. With the use of different modes as its carriers and few mode fiber as its transmission link, a mode division multiplexing transmission system using MIMO configuration can expand its capacity by times and improve the spectral efficiency at the same time.Current studies are mostly about mode or optical components, thus this paper focuses on the algorithms of optical MIMO mode division multiplexing systems. Use MATLAB as the tool to search the performance of different algorithms. The main contents are as follow:(1) Studies the structure, key components, principle and implementation of a MIMO mode division multiplexing system in few mode fiber. Analyze the key technologies in mode division multiplexing systems and the model of MIMO. And then, Study into the detection algorithms in depth, including maximum likelihood detection algorithm, minimum mean square error detection algorithm, zero forcing detection algorithm and sphere decoding algorithm and other nonlinear detection algorithms.(2) Establish the model of MIMO-MDM, and apply the MIMO detection algorithms to the system. After simulation on MATLAB, It’s clear about the performance of the algorithms according to the curves of signal to noise ratio-bit error rate. Use different modulation schemes in the process of simulation, such as BPSK, QPSK and 16QAM. Compare the curves of different algorithms with the same modulation scheme, and study the curves of different modulation schemes with the same detection algorithm. At last, change the gain matrix of and the noise at the receiver, repeat the simulation of the two types comparison as mentioned.(3) Depending on the curves of different algorithms at relatively the same modulation manner, sphere decoding algorithm is better than other linear algorithms with its performance closet to that of maximum likelihood algorithm. By studying the curves of different modulation schemes with the same algorithm, QPSK seems to be a good choice considering both the SNR needed and the complexity. At last, the results seems to be the same while the channel model is changed. Sphere decoding algorithm is still quite good considering both the computational complexity and the performance.