注鎖與回授控制穩頻寬線寬雷射量子密鑰分發

Abstract

隨著網際網路的蓬勃發展，行動支付的需求日漸增加，對於行動支付的資安是不可忽視的。由於傳統加密方式最常用的是RSA、AES等等，其安全性依賴於某些數學問題的困難度，例如大質數分解，對於現有電腦其破解速度需要好幾年。但近幾年量子電腦的快速發展，針對特殊問題的運算速度大幅提升，在短時間內破解已非不可能，因此亟需一個具有極高安全性的加密方式，量子密鑰分發正好符合此需求。 在第二章中，本篇論文首先針對商用DFBLD的基本特性進行分析，如P-I-V 曲線、閃爍雜訊、相對強度雜訊以及線寬。閃爍雜訊以及相對強度雜訊影響脈衝調製頻率的選擇範圍，當雷射的偏壓電流操作在7 Ith以上時，脈衝頻率可以操作在200 MHz以上。雷射線寬以及延遲干涉儀的自由光譜範圍大小對於QKD傳輸的影響十分顯著，使用寬線寬雷射進行傳輸，會因為干涉儀兩路輸出互相干擾，使得接收端無法解碼，本章使用線寬為293 kHz的商用DFBLD，其造成量子誤碼率為0.38%。除此之外，溫度改變造成的波長飄移遠大於電流改變造成的波長飄移，因此雷射需要精準的溫度控制，以防止干涉儀輸出擾動。單光子偵測器的各項操作參數也需要最佳化。脈衝產生器的脈衝頻率需要與延遲干涉儀的延遲時間對準，使得兩個脈衝能完全干涉。因為單光子偵測器的休息時間為1 µs，每個密碼之間的間隔時間也需要為1 µs。還有因為光路跟電路的延遲，單光子偵測器的閘門偵測時間需要與脈衝抵達時間對準。 第三章首先分析主雷射和從雷射的基本參數。主雷射在偏壓電流為60 mA時P-I 曲線有一不連續點，會造成線寬拓寬，需要避開此操作點。由於從雷射的線寬以及相對強度雜訊不適合用於傳輸，因此使用注入鎖定技術縮減雷射線寬，從10.62 GHz降低至6.37 MHz。再來分析注入鎖定範圍，使用Q值較低的從雷射可以更容易注入鎖定，但缺點是有較寬的線寬。在注入鎖定前，由於從雷射線寬太寬，3種不同自由光譜範圍之延遲干涉儀皆會因為兩路輸出互相干擾，而無法進行量子密鑰分發解調。在注入鎖定後，線寬大幅下降，因此在這3個延遲干涉儀可以觀察到明顯的干涉輸出，而192 MHz的延遲干涉儀之洩漏率仍有5%以上，其破壞性干涉無法降至sub-mW以下，這對於可見度的影響是劇烈的，除此之外，1.45 GHz的延遲干涉儀洩漏率最低，完全建設性干涉的功率比另外兩個還要大。因為系統的時序顫動，加密長度過長會使得量子誤碼率上升。對於192 MHz的延遲干涉儀，4種加密長度皆無法通過錯誤糾正的標準，而1 GHz和1.45 GHz的延遲干涉儀在128、256、512碼皆能通過標準，但1024碼則會略為超過。

With the flourishing development of the Internet, the demand for mobile payments is steadily increasing, making information security in mobile payments an essential consideration. Traditional encryption methods, such as RSA and AES, rely on the difficulty of certain mathematical problems, like prime factorization, to ensure security. For current computers, breaking these encryption methods would take several years. However, the rapid advancement of quantum computers in recent years has significantly increased the computational speed for specific problems, making it feasible to break traditional encryption methods in a relatively short time. Hence, there is an urgent need for an encryption method that offers exceptionally high security, and quantum key distribution (QKD) is perfectly suited to meet this demand. In Chapter 2, this paper commences by analyzing the fundamental characteristics of commercial distributed feedback laser diode (DFBLD), including P-I-V curves, flicker noise, relative intensity noise (RIN), and linewidth. Flicker noise and RIN influence the selection range for pulse modulation frequencies. When the laser operates at bias currents greater than 7 Ith, the pulse frequency can exceed 200 MHz. The linewidth of the laser and the free spectral range of the delay interferometer significantly impact the transmission of QKD. Using lasers with wide linewidths for transmission can lead to interference between the two output channels of the interferometer, rendering the decoding at the receiver end impossible. In this chapter, a DFBLD with a linewidth of 293 kHz was used, resulting in a quantum bit error rate (QBER) of 0.38%. Furthermore, temperature-induced wavelength drift is much greater than the drift caused by changes in current, necessitating precise temperature control of the laser to prevent disturbances in the interferometer's output. Parameters of single-photon detectors also require optimization. The pulse frequency of the pulse generator must align with the delay time of the delay interferometer (DI), allowing the two pulses to interfere completely. Given the resting time of the single-photon detectors is 1 µs, the interval time between each code must also be 1 µs. Furthermore, due to delays in the optical path and circuits, the gate detection time of the single-photon detectors needs to align with the pulse arrival time. Chapter 3 initiates by analyzing the fundamental parameters of the master and slave lasers. The master laser exhibits a discontinuity point in its P-I curve when biased at 60 mA, resulting in an increase in linewidth. It is advisable to avoid operating at this point. Due to the wide linewidth and RIN of the slave laser, they are unsuitable for transmission. Therefore, the linewidth of the laser was reduced from 10.62 GHz to 6.37 MHz using optical injection locking (OIL). Subsequently, the injection locking range is analyzed, where a slave laser with a lower Q-value is easier to inject lock but comes with a wider linewidth. Before injection locking, the wide linewidth of the slave laser leads to mutual interference between the two output channels of the DI for three different free spectral ranges, making quantum key distribution decryption impossible. However, after injection locking, the linewidth significantly reduces. As a result, clear interference output can be observed in all three delay interferometers. The delay interferometer with an FSR of 192 MHz still has a leakage rate of over 5%, and its destructive interference cannot be reduced to below sub-mW. This significantly impacts visibility. In contrast, the delay interferometer with an FSR of 1.45 GHz has the lowest leakage rate, and the power of fully constructive interference is greater than the other two. Because of system timing jitter, excessively long encryption lengths lead to an increase in quantum error rates. For the 192 MHz DI, none of the four encryption lengths meet the error correction standard. In contrast, for 1 GHz and 1.45 GHz DIs, encryption lengths of 128, 256, and 512 codes meet the standard, but the 1024 code length slightly exceeds it.