System Performance Improvement by Applying Artificial Intelligence Algorithm of MIMO Log-normal Channel for OFDM/OCDMA System

• Main Authors: TSAI, HSIN-HAN, 蔡昕翰
• Other Authors: YEN, CHIH-TA
• Format: Others
• Language: zh-TW
• Published: 2018
• Online Access: http://ndltd.ncl.edu.tw/handle/2z8d49

碩士 === 國立虎尾科技大學 === 電機工程系碩士班 === 106 === In this paper, we use an orthogonal frequency division polarization multiplexing multiplexing optical code division multiplexing system, and adopt MIMO (Multiple Input Multiple Output) architecture on the free optical channel. In this system, we use Walsh Hadamard codes (WHC) coding. As the user's codeword, the technology uses the characteristics of fixed cross-correlation between each group of WHCs, and uses the balanced detection method at the receiving end to effectively eliminate the interference between the base stations, so it has a stable transmission rate and data. The benefits of security and low transmission costs. Finally, the optimization algorithm is proposed to resist the attenuation in harsh environments. In this paper, the gene algorithm and neural network are used as the algorithm of the optimization system. In terms of simulation software operation, this paper mainly uses the commercial optical communication software Opitsysytem 15.0 to verify the feasibility of the architecture, and collect data and simulation through the built-in Matlab link original. In the fifth chapter, the formula is deduced, and the log-normal distribution channel and the noise value are added, and the floating climate state is added to make the simulated environment more in line with the real climate. Use Matlab to simulate and compare the results of the optisystem simulation to verify that the architecture has been deduced correctly. After the verification is correct, use the genetic algorithm, add the compensation parameters, optimize the BER formula, and analyze the original architecture and optimization framework. Comparison of received power with BER, and user comparison of BER. As is known from the simulation results, the use of gene algorithms and neural network-like compensation as channel attenuation can indeed improve the system performance when the channel is not ideal. After the two algorithms are simulated and compared, the neural network optimization part is more suitable than the gene algorithm to fit the proposed structure.