【作者】 程木；

【导师】 吴重庆；

【作者基本信息】 北京交通大学 ， 光学， 2009， 博士

【摘要】 由于全光分组交换(OPS)具有对数据速率、调制格式、传输协议等透明的优点,同时OPS具有高吞吐量、高传输速率,能有效的解决目前交换节点中存在的“电子瓶颈”问题,全光分组交换被认为是下一代网络交换技术的首选方案。光缓存是实现全光分组交换的关键技术之一。光缓存器可以在光域内完成数据包的存储而不需要经过光—电—光的变换,是实现光分组同步、竞争解决和流量整形的关键。本文提出了一种基于半导体光放大器(SOA)非线性偏振旋转的光缓存器结构,并且实验实现了该缓存器的多级级联。该缓存器具有结构简单、可级联、易集成、功耗小、重构速度快、易于控制和成本低等优点。另外,本文对基于半导体光放大器非线性偏振旋转的光缓存器的关键问题进行了研究和讨论。本文的创新点和主要工作包括以下几点:1对半导体光放大器的非线性偏振旋转进行了理论研究。在建立非线性偏振旋转的理论模型的基础上,提出了一种寻找SOA偏振主态,以及通过调节注入电流获得SOA两个正交线偏振态的方法,从而实现了两个正交线偏振态之间的转化。利用上述方法,实现了基于SOA非线性偏振旋转的光开关。对光开关的分光比和开关速度进行了深入研究,结果表明,光开关在“通态”和“切换态”时,开关的两个输出端口的分光比并不相同,且光开关的分光比随着输入光功率的变大而变大,增加输入信号光的功率可以改善光开关的开关速度,但考虑到SOA的增益饱和效应,输入信号光的功率不易大于OdBm。2提出了一种基于SOA非线性偏振旋转的光缓存器的新结构。该缓存器利用SOA中同样的非线性偏振旋转,既可以实现“读”操作,也可以实现“写”操作,控制简单,读写速度快。与NOLM结构的光缓存器相比,没有被缓存的数据可以直接通过而不需要在光纤环中绕行一周;由于SOA不在光纤环中放置,缓存器的缓存深度可以达到80%以上。所以,该缓存器具有结构简单、读写速度快、功耗小、易级联等优点。利用该缓存器,实现了2.5Gbps速率的1024比特数据的12圈的缓存,每一圈的缓存时间为0.91μs。实验中,还提出了一种利用PBS和耦合器在线监测系统实现两个正交线偏振态之间相互转化的方法。实验测量了信号的椭圆度对PBS分光比的影响,指出缓存结果中出现的“漏光”现象是由于信号光的椭圆度降低导致PBS的分光比降低造成。3提出了基于SOA非线性偏振旋转的光缓存器的级联结构,该级联结构具有控制元件少的优点,每一级的缓存单元只需一个控制元件(SOA),且级联结构中没有2×2光开关,易于集成。首次实现了两级、三级该缓存器的级联,两级光缓存器可提供60个不同的缓存时间,最大缓存时间为～40μs;三级缓存器可提供140多个不同的缓存时间,最大缓存时间为～50μs。更多还原

【Abstract】 Optical packet switching networks are transparent to the data rate,modulation format,transmission protocol,and so on.Therefore,the optical packet switching(OPS) emerge as first scheme for the next generation switching systems,which accommodate huge data-centric traffic at the packet level granularity and match the high transmission capacity provided by the dense wavelength division multiplexing(DWDM) technologies.OPS are the key approach to build future compact and power-efficient all-optical data routers with greater scalability,flexibility,and switching capacity.One of the most critical components in such an all-optical router is the optical buffer,which stores the data packets for a desired amount of time during destined output switch fabric port contention and releases the data packets while the destined output port is cleared. Optical buffer can also be use in optical packet synchronization and traffic shaping.In the dissertation,we proposed a novel optical buffer based on the nonlinear polarization rotation in SOA,which is cost-effective,compact and low power consumption,and can provide variable buffer length,fast dynamic reconfiguration in picoseconds.These conclusions can also be applicable for all the other SOA based optical buffers.What I have done during the PHD period are as follows:For the first time we have presented a simple principal states of polarizations (PSP) model that is capable of describing the behavior of polarization switches based on nonlinear polarization rotation in SOA with tail fibre.Using this theory we experimentally study the nonlinear polarization switching based on the polarization rotation in SOA.During the experiment,the output SOP evolution in SOA can easily be tracked and controlled on line.We have also demonstrated a nonlinear polarization switching based on the polarization rotation induced by bias current in SOA. Experimental results show that the static contrast ratio is more than 20dB and the dynamic contrast ratio is more than 10dB.We present a novel optical buffer based on nonlinear polarization rotation induced by inject current of SOA.This buffer is simple,compact,cost effective,low power consumption and easy to cascade.With the buffer configuration,the signal can pass directly if the signal does not need buffer.In addition,the buffer depth can exceed 50% and the configuration is easy to cascade.With the novel design,we had successfully demonstrated the loading and reading operation at bit rate of 2.5 Gb/s.Random data of 1024-bit was buffered for 12 circulations(11μs).We present the cascaded buffer scheme without space switching to achieve variable-length delay buffer.The most interesting features of this new buffer are that it has more than 1000 optional delay lengths,fast dynamic reconfiguration in picoseconds. With two stages cascaded buffer,we had successfully demonstrated 60 buffering times at 2.5 Gb/s,the maximum buffer time is～40μs.With three stages cascaded buffer,we had successfully demonstrated 144 buffering times at 2.5 Gb/s,the maximum buffer time is～50μs.更多还原