We review current silicon photonic devices and their performance in connection with energy consumption. Four critical issues are identified to lower energy consumption in devices and systems: reducing the influence of the thermo-optic effect, increasing the wall-plug efficiency of lasers on silicon, optimizing energy performance of modulators, and enhancing the sensitivity of photodetectors. Major conclusions are (1) Mach–Zehnder interferometer-based devices can achieve athermal performance without any extra energy consumption while microrings do not have an efficient passive athermal solution; (2) while direct bonded III–V-based Si lasers can meet system power requirement for now, hetero-epitaxial grown III–V quantum dot lasers are competitive and may be a better option for the future; (3) resonant modulators, especially coupling modulators, are promising for low-energy consumption operation even when the power to stabilize their operation is included; (4) benefiting from high sensitivity and low cost, Ge/Si avalanche photodiode is the most promising photodetector and can be used to effectively reduce the optical link power budget. These analyses and solutions will contribute to further lowering energy consumption to meet aggressive energy demands in future systems.

Silicon photonics has become one of the major technologies in this very information age. It has been intensively pursued by researchers and entrepreneurs all over the world in recent years. Achieving the large scale silicon photonic integration, particularly monolithic integration, is the final goal so that high density data communication will become much cheaper, more reliable, and less energy consuming. Comparing with the developed countries, China may need to invest more to develop top down nanoscale integration capability (more on processing technology) to sustain the development in silicon photonics and to elevate its own industry structure.

研究了小波变换和Contourlet变换的不足,利用非抽样Contourlet变换的多尺度性、多方向性以及平移不变性特征进行彩色图像数字水印。先对图像进行非抽样Contourlet变换得到不同尺度不同方向上的系数,然后为了满足水印的不可见性和稳健性,在低频系数和高频系数中采用不同的嵌入强度,并依据人类视觉的彩色感知特性,在蓝色分量中嵌入较多水印,在其他颜色分量选取变换系数能量最大的区域嵌入水印。与同样嵌入规则的Contourlet变换算法和平稳小波变换算法相比,该算法具有更好的稳健性,利用该算法得到的水印相似度更高,最高可提高约0.5。

提出了一种用神经网络鉴别退化图像的模糊类型的方法。由于采用不同降质方法得到退化图像的频谱差异较大，以此作为判别依据，用概率神经网络实现了对四种模糊类型：离焦，矩形，运动和高斯模糊的鉴别。根据神经网络的鉴别结果决定点扩散函数的初始估计值，可大大地提高盲解恢复算法的复原质量和系统点扩散函数的估计精度，扩大了算法的实用范围。

以掺铒磷酸盐玻璃中铒离子寿命随OH含量变化为依据，系统研究了碱土金属离子类型和含量对磷酸盐铒玻璃除水性质的影响。结果表明，采用通干燥O2和CCl4除水工艺可有效降低玻璃中OH含量，除水效果在通气的最初阶段最为明显，玻璃中OH含量降低速率随通气时间延长而减慢，最终趋于动态平衡。值得指出的是，在相同通气时间，磷酸盐铒玻璃的荧光寿命随碱土金属离子半径增大而减短。含有较高浓度碱土金属离子的磷酸盐铒玻璃具有较短的荧光寿命，并且在停止通气后熔体重新吸水的能力变大。