【作者】 邱吉芳；

【导师】 林金桐；

【作者基本信息】 北京邮电大学 ， 物理电子学， 2010， 博士

【摘要】 全光逻辑信号处理技术在光交换节点的大多数功能单元中都起着至关重要的作用,比如分组头处理、净荷定位、时钟提取、信号再生、光分组自路由以及光信号编码等节点功能都要依靠逻辑器件来实现。因此,一旦此技术获得突破,不仅能为目前的光网络开辟新的发展方向,而且具有极其广泛的应用,它将会给人类信息社会带来一场深刻的、革命性的变革,如同晶体管及集成电路技术在20世纪信息社会中的地位和作用一样。鉴于此,本论文着眼于光网络中基于高非线性器件的全光可重构多逻辑门的实现技术,包括全光AND、OR/NOR、XOR和NOT门等。本论文的主要工作如下：介绍了SOA中非线性偏振旋转(NPR)效应产生的物理机制,以及SOA非线性偏振相关的理论模型,从理论上熟悉NPR效应的物理层含义；然后建立了与实验中SOA参数吻合的理论模型,分析研究了SOA偏振相关的增益特性、相位特性和啁啾特性。实验方面,首次研究了SOA的NPR与偏振相关增益(PDG)之间的关系。通过研究SOA的PDG,巧妙地利用泵浦-探测结构在单个SOA中实现了对NPR与PDG关系的研究,得到结论：1)当泵浦光在SOA的增益峰值波长处,SOA中NPR效应最强；2)SOA的PDG越大,NPR效应越强。同时还研究了输入光功率、偏振态对SOA中NPR效应的影响。实验表明,选择合适的SOA和输入光功率、波长及偏振态对于基于SOA中NPR的影响很大。为之后基于SOA中NPR效应的多逻辑门研究做了扎实的准备工作。基于上述SOA的理论模型,仿真研究了基于SOA中NPR和窄带滤波特性的开关器件的性能,为后面基于NPR和窄带滤波特性的逻辑门研究做了铺垫工作。在完成对NPR和窄带滤波的实验及仿真研究基础上,首先提出了一种基于SOA中NPR效应和窄带滤波特性的全光可重构多逻辑门实现方案。仿真上实现了速率为10Gbit/s、Q因子分别为15.95、8.35、20.84、12.89的AND、XOR、OR、NOT和NOR门。实验方面,在2.5Gbit/s的实验平台上同时实现了A.B和A.B门,验证了XOR门的可行性。通过研究调制码长和调制速率对输出结果的影响,得到结论：如果采用增益恢复时间更快的SOA,该方案可工作在更高速率上。该方案具有结构简单、可集成、可重构的优点。鉴于HNLF的优势和发展潜力,论文提出一种基于HNLF中XPM效应的可重构多逻辑门实现方案。该方案有三个显著的优点：1)只采用了一种非线性效应(即XPM),2)输入只有两路信号光(没有附加的探测光),3)在不改变输入光功率、波长、偏振等参数的条件下,只需通过调节输出端滤波器中心波长的设置即可实现高质量的无误码输出。实验实现了速率为10Gbit/s的XOR、OR和AND门,各逻辑门相应的消光比分别为：14.0dB、16.5dB和22.2dB,功率代价分别为：5.9 dB,2.9 dB和-1.0 dB。由于该方案基于快速响应的Kerr效应,该方案可工作在速率高于100Gbit/s的系统。仿真深入研究了该方案在波长范围内的适用性,就接收灵敏度与波长之间的关系做了仿真验证和分析；最后研究了系统性能与高非线性光纤长度的关系。研究表明,该方案在波长间隔大于5nm的整个C波段范围内都可以实现高性能无误码的输出,并且由于走离效应和自相位展宽的存在,选择合适长度的高非线性光纤可以进一步优化系统性能。在高速光信号处理技术中,高重复频率超短光脉冲的产生是实现系统运行的必需条件。因此,本论文探索了超窄脉冲的产生技术。实验研究基于自级联EAM的短脉冲产生技术,产生了10.2ps的10GHz窄脉冲信号,相对于基于单个EAM产生窄脉冲的技术,基于自级联EAM的结构可以让脉冲宽度压缩50%以上。通过对自级联EAM的开关窗口的研究,首次将该自级联EAM结构用于高速长距离传输的OTDM系统中的解复用，实验成功地搭建了一个稳定可靠、速率为80Gbit/s、传输距离为300km的高速OTDM系统,采用自级联EAM结构的时分解复用器成功地实现了80Gbit/s到10Gbit/s的解复用；在接收端,接收灵敏度为-19.0dBm,功率代价为1.4dB。更多还原

【Abstract】 High-speed all-optical logic gates are key elements in all-optical switching networks, which can realize header recognition, clock recovery, signal regeneration, all-optical packet self-routing, optical signal coding, buffering, etc. Once these technologies become commercially mature, current optical network would benefit from these technologies significantly, like what transistor and integrated circuit (IC) did to the information society in 1900s.All-optical logic gates cover XOR, AND/NAND, OR/NOR, NOT, and other complex logic gates, like half-adder, full-adder, etc. Many approaches have been proposed to achieve various all-optical logic functions including simple logic gates and complex logic gates, based on the nonlinear effects in semiconductor material, in optical fibers or in waveguides. However, it is more attractive to realize multi-functional logic gates by using the same architecture or few alterations because of the flexible logic operations and potential applications in massive photonic integration.Therefore, this dissertation mainly focuses on the technologies of highly-nonlinear-materials-based all-optical logic gates realization, including all-optical AND, OR/NOR, XOR and NOT, etc. The main research efforts are summarized as follow. A theoretical model considering length effect and polarization effect in a SOA has been built to study the polarization-dependent gain, polarization-dependent phase and polarization-dependent chirp. Experimental characteristics based on a simple pump-probe setup have been carried out to investigate the influence of polarization dependent gain (PDG) on the nonlinear polarization rotation (NPR) response in SOAs. The results show that PDG enhances the NPR effect in SOAs. Additionally, the influences of input probe power, input probe polarization and input pump polarization in the NPR. response in SOAs are also measured experimentally. The results indicate that the PDG, input optical power, wavelength and polarization affect NPR effect significantly.The performance of an all-optical switch based on NPR and sideband filtering in a SOA has been theoretically studied to help the following all-optical multi-logic gate scheme, which is based on NPR and sideband filtering in a SOA. Specifically, the influences of pump pulsewidth, the bandwidth and center wavelength setting of the bandpass filter have been investigated.Based on the above experimental and theoretical results, a novel multi-logic gate based on NPR and sideband-filtering in a single SOA has been proposed. Digital logic operations including AND, XOR, OR, and NOT are realized at 10 Gb/s numerically, with Q factor of 15.95,8.35, 20.84,12.89 respectively. Then, A·B and A·B gates, which can be combined together to realize XOR, are realized simultaneously in a same experimental setup at 2.5 Gb/s, indicating the feasibility in experimental realization of XOR gate. Further experiment on the influences of pattern length and operation speed on the logic gates have been investigated, indicating that by exploiting a SOA with faster gain recovery time, this scheme can work at higher speed.Cross phase modulation (XPM) in a HNLF is proposed to realize all-optical multi-logic gate. Digital logic operations including XOR, AND, and OR of two input data signals are demonstrated using a single device. We obtain error free operation at 10 Gb/s and the receiver sensitivity for a BER of 10-9 is below-15.0 dBm. Power penalties for XOR, AND, and OR logic operations are 5.9 dB,-1.0 dB and 2.9 dB, respectively.Further analyses of the feasible application of the scheme in the whole C-band have been carried out numerically, concluding the influences of wavelength choosing of signals and wavelength separation between two signals on the performance of the scheme. Moreover, the influence of the length of HNLF on the performance has been studied either.1. we use only one nonlinearity (i.e., XPM), which makes the experiment easy to control,2. we use only two input data signals, thereby avoiding the need for additional sources such as a continuous wave (CW) or a clock signal, and3. the experiment is configurable, which means all three logic gates (XOR, AND, and OR) are realized by only adjusting two bandpass filters, i.e., there is no need to change input variables/parameters such as power, wavelength, or polarization.A novel scheme for pulse generation with a self-cascaded EAM is presented and experimentally demonstrated at 10 GHz. In the case of optimal tuning of time delay in the fiber loop, the improvement of 50% on pulsewidth with improved extinction ratio is obtained and the narrowest pulse generated with this method is about 11 ps.The self-cascaded EAM has also been applied in a 300km 8×10Gb/s OTDM transmission system as a demultiplexer. With the more than 50% decreased switching window, an error free demultiplexing is obtained with a sensitivity of-19dBm, a power penalty of 1.4dB is achieved compared with the back-to-back case.更多还原