The argument of this thesis might be summed up as the exploitation of the noise to generate better noise. More specifically this work is about the possibility of exploiting classic noise to effectively transmit quantum information and measuring quantum noise to generate better quantum randomness. What do i mean by exploiting classical noise to transmit effectively quantum information? In this case I refer to the task of sending quantum bits through the atmosphere in order set up transmissions of quantum key distribution (QKD) and this will be the subject of Chapter 1 and Chapter 2. In the Quantum Communications framework, QKD represents a topic with challenging problems both theoretical and experimental. In principle QKD offers unconditional security, however practical realizations of it must face all the limitations of the real world. One of the main limitation are the losses introduced by real transmission channels. Losses cause errors and errors make the protocol less secure because an eavesdropper could try to hide his activity behind the losses. When this problem is addressed under a full theoretical point of view, one tries to model the effect of losses by means of unitary transforms which affect the qubits in average according a fixed level of link attenuation. However this approach is somehow limiting because if one has a high level of background noise and the losses are assumed in average constant, it could happen that the protocol might abort or not even start, being the predicted QBER to high. To address this problem and generate key when normally it would not be possible, we have proposed an adaptive real time selection (ARTS) scheme where transmissivity peaks are instantaneously detected. In fact, an additional resource may be introduced to estimate the link transmissivity in its intrinsic time scale with the use of an auxiliary classical laser beam co-propagating with the qubits but conveniently interleaved in time. In this way the link scintillation is monitored in real time and the selection of the time intervals of high channel transmissivity corresponding to a viable QBER for a positive key generation is made available. In Chapter 2 we present a demonstration of this protocol in conditions of losses equivalent to long distance and satellite links, and with a range of scintillation corresponding to moderate to severe weather. A useful criterion for the preselection of the low QBER interval is presented that employs a train of intense pulses propagating in the same path as the qubits, with parameters chosen such that its fluctuation in time reproduces that of the quantum communication. For what concern the content of Chapter 3 we describe a novel principle for true random number generator (TRNG) which is based on the observation that a coherent beam of light crossing a very long path with atmospheric turbulence may generate random and rapidly varying images. To implement our method in a proof of concept demonstrator, we have chosen a very long free space channel used in the last years for experiments in Quantum Communications at the Canary Islands. Here, after a propagation of 143 km at an altitude of the terminals of about 2400 m, the turbulence in the path is converted into a dynamical speckle at the receiver. The source of entropy is then the atmospheric turbulence. Indeed, for such a long path, a solution of the Navier-Stokes equations for the {atmospheric flow in which the beam propagates is out of reach. Several models are based on the Kolmogorov statistical theory, which parametrizes the repartition of kinetic energy as the interaction of decreasing size eddies. However, such models only provide a statistical description for the spot of the beam and its wandering and never an instantaneous prediction for the irradiance distribution. These are mainly ruled by temperature variations and by the wind and cause fluctuations in the air refractive index. For such reason, when a laser beam is sent across the atmosphere, this latter may be considered as a dynamic volumetric scatterer which distorts the beam wavefront. We will evaluate the experimental data to ensure that the images are uniform and independent. Moreover, we will assess that our method for the randomness extraction based on the combinatorial analysis is optimal in the context of Information Theory. In Chapter 5 we will present a new approach for what concerns the generation of random bits from quantum physical processes. Quantum Mechanics has been always regarded as a possible and valuable source of randomness, because of its intrinsic probabilistic Nature. However the typical paradigm is employed to extract random number from a quantum system it commonly assumes that the state of said system is pure. Such assumption, only in theory would lead to full and unpredictable randomness. The main issue however it is that in real implementations, such as in a laboratory or in some commercial device, it is hardly possible to forge a pure quantum state. One has then to deal with quantum state featuring some degree of mixedness. A mixed state however might be somehow correlated with some other system which is hold by an adversary, a quantum eavesdropper. In the extreme case of a full mixed state, practically one it is like if he is extracting random numbers from a classical state. In order to do that we will show how it is important to shift from a classical randomness estimator, such as the min-classical entropy H-min(Z) of a random variable Z to quantum ones such as the min-entropy conditioned on quantum side information E. We have devised an effective protocol based on the entropic uncertainty principle for the estimation of the min-conditional entropy. The entropic uncertainty principle lets one to take in account the information which is shared between multiple parties holding a multipartite quantum system and, more importantly, lets one to bound the information a party has on the system state after that it has been measured. We adapted such principle to the bipartite case where an user Alice, A, is supplied with a quantum system prepared by the provider Eve, E, who could be maliciously correlated to it. In principle then Eve might be able to predict all the outcomes of the measurements Alice performs on the basis Z in order to extract random numbers from the system. However we will show that if Alice randomly switches from the measurement basis to a basis X mutually unbiased to Z, she can lower bound the min entropy conditioned to the side information of Eve. In this way for Alice is possible to expand a small initial random seed in a much larger amount of trusted numbers. We present the results of an experimental demonstration of the protocol where random numbers passing the most rigorous classical tests of randomness were produced. In Chapter 6, we will provide a secure generation scheme for a continuos variable (CV) QRNG. Since random true random numbers are an invaluable resource for both the classical Information Technology and the uprising Quantum one, it is clear that to sustain the present and future even growing fluxes of data to encrypt it is necessary to devise quantum random number generators able to generate numbers in the rate of Gigabit or Terabit per second. In the Literature are given several examples of QRNG protocols which in theory could reach such limits. Typically, these are based on the exploitation of the quadratures of the electro-magnetic field, regarded as an infinite bosonic quantum system. The quadratures of the field can be measured with a well known measurement scheme, the so called homodyne detection scheme which, in principle, can yield an infinite band noise. Consequently the band of the random signal is limited only by the passband of the devices used to measure it. Photodiodes detectors work commonly in the GHz band, so if one sample the signal with an ADC enough fast, the Gigabit or Terabit rates can be easily reached. However, as in the case of discrete variable QRNG, the protocols that one can find in the Literature, do not properly consider the purity of the quantum state being measured. The idea has been to extend the discrete variable protocol of the previous Chapter, to the Continuous case. We will show how in the CV framework, not only the problem of the state purity is given but also the problem related to the precision of the measurements used to extract the randomness.
The argument of this thesis might be summed up as the exploitation of the noise to generate better noise. More specifically this work is about the possibility of exploiting classic noise to effectively transmit quantum information and measuring quantum noise to generate better quantum randomness. What do i mean by exploiting classical noise to transmit effectively quantum information? In this case I refer to the task of sending quantum bits through the atmosphere in order set up transmissions of quantum key distribution (QKD) and this will be the subject of Chapter 1 and Chapter 2. In the Quantum Communications framework, QKD represents a topic with challenging problems both theoretical and experimental. In principle QKD offers unconditional security, however practical realizations of it must face all the limitations of the real world. One of the main limitation are the losses introduced by real transmission channels. Losses cause errors and errors make the protocol less secure because an eavesdropper could try to hide his activity behind the losses. When this problem is addressed under a full theoretical point of view, one tries to model the effect of losses by means of unitary transforms which affect the qubits in average according a fixed level of link attenuation. However this approach is somehow limiting because if one has a high level of background noise and the losses are assumed in average constant, it could happen that the protocol might abort or not even start, being the predicted QBER to high. To address this problem and generate key when normally it would not be possible, we have proposed an adaptive real time selection (ARTS) scheme where transmissivity peaks are instantaneously detected. In fact, an additional resource may be introduced to estimate the link transmissivity in its intrinsic time scale with the use of an auxiliary classical laser beam co-propagating with the qubits but conveniently interleaved in time. In this way the link scintillation is monitored in real time and the selection of the time intervals of high channel transmissivity corresponding to a viable QBER for a positive key generation is made available. In Chapter 2 we present a demonstration of this protocol in conditions of losses equivalent to long distance and satellite links, and with a range of scintillation corresponding to moderate to severe weather. A useful criterion for the preselection of the low QBER interval is presented that employs a train of intense pulses propagating in the same path as the qubits, with parameters chosen such that its fluctuation in time reproduces that of the quantum communication. For what concern the content of Chapter 3 we describe a novel principle for true random number generator (TRNG) which is based on the observation that a coherent beam of light crossing a very long path with atmospheric turbulence may generate random and rapidly varying images. To implement our method in a proof of concept demonstrator, we have chosen a very long free space channel used in the last years for experiments in Quantum Communications at the Canary Islands. Here, after a propagation of 143 km at an altitude of the terminals of about 2400 m, the turbulence in the path is converted into a dynamical speckle at the receiver. The source of entropy is then the atmospheric turbulence. Indeed, for such a long path, a solution of the Navier-Stokes equations for the {atmospheric flow in which the beam propagates is out of reach. Several models are based on the Kolmogorov statistical theory, which parametrizes the repartition of kinetic energy as the interaction of decreasing size eddies. However, such models only provide a statistical description for the spot of the beam and its wandering and never an instantaneous prediction for the irradiance distribution. These are mainly ruled by temperature variations and by the wind and cause fluctuations in the air refractive index. For such reason, when a laser beam is sent across the atmosphere, this latter may be considered as a dynamic volumetric scatterer which distorts the beam wavefront. We will evaluate the experimental data to ensure that the images are uniform and independent. Moreover, we will assess that our method for the randomness extraction based on the combinatorial analysis is optimal in the context of Information Theory. In Chapter 5 we will present a new approach for what concerns the generation of random bits from quantum physical processes. Quantum Mechanics has been always regarded as a possible and valuable source of randomness, because of its intrinsic probabilistic Nature. However the typical paradigm is employed to extract random number from a quantum system it commonly assumes that the state of said system is pure. Such assumption, only in theory would lead to full and unpredictable randomness. The main issue however it is that in real implementations, such as in a laboratory or in some commercial device, it is hardly possible to forge a pure quantum state. One has then to deal with quantum state featuring some degree of mixedness. A mixed state however might be somehow correlated with some other system which is hold by an adversary, a quantum eavesdropper. In the extreme case of a full mixed state, practically one it is like if he is extracting random numbers from a classical state. In order to do that we will show how it is important to shift from a classical randomness estimator, such as the min-classical entropy H-min(Z) of a random variable Z to quantum ones such as the min-entropy conditioned on quantum side information E. We have devised an effective protocol based on the entropic uncertainty principle for the estimation of the min-conditional entropy. The entropic uncertainty principle lets one to take in account the information which is shared between multiple parties holding a multipartite quantum system and, more importantly, lets one to bound the information a party has on the system state after that it has been measured. We adapted such principle to the bipartite case where an user Alice, A, is supplied with a quantum system prepared by the provider Eve, E, who could be maliciously correlated to it. In principle then Eve might be able to predict all the outcomes of the measurements Alice performs on the basis Z in order to extract random numbers from the system. However we will show that if Alice randomly switches from the measurement basis to a basis X mutually unbiased to Z, she can lower bound the min entropy conditioned to the side information of Eve. In this way for Alice is possible to expand a small initial random seed in a much larger amount of trusted numbers. We present the results of an experimental demonstration of the protocol where random numbers passing the most rigorous classical tests of randomness were produced. In Chapter 6, we will provide a secure generation scheme for a continuos variable (CV) QRNG. Since random true random numbers are an invaluable resource for both the classical Information Technology and the uprising Quantum one, it is clear that to sustain the present and future even growing fluxes of data to encrypt it is necessary to devise quantum random number generators able to generate numbers in the rate of Gigabit or Terabit per second. In the Literature are given several examples of QRNG protocols which in theory could reach such limits. Typically, these are based on the exploitation of the quadratures of the electro-magnetic field, regarded as an infinite bosonic quantum system. The quadratures of the field can be measured with a well known measurement scheme, the so called homodyne detection scheme which, in principle, can yield an infinite band noise. Consequently the band of the random signal is limited only by the passband of the devices used to measure it. Photodiodes detectors work commonly in the GHz band, so if one sample the signal with an ADC enough fast, the Gigabit or Terabit rates can be easily reached. However, as in the case of discrete variable QRNG, the protocols that one can find in the Literature, do not properly consider the purity of the quantum state being measured. The idea has been to extend the discrete variable protocol of the previous Chapter, to the Continuous case. We will show how in the CV framework, not only the problem of the state purity is given but also the problem related to the precision of the measurements used to extract the randomness.
In the past decade, quantum communication protocols based on continuous variables (CV) has seen considerable development in both theoretical and experimental aspects. Nonetheless, challenges remain in both the practical security and the operating range for CV systems, before such systems may be used extensively. In this thesis, we present the optimisation of experimental parameters for secure randomness generation and propose a non-deterministic approach to enhance amplification of CV quantum state. The first part of this thesis examines the security of quantum devices: in particular, we investigate quantum random number generators (QRNG) and quantum key distribution (QKD) schemes. In a realistic scenario, the output of a quantum random number generator is inevitably tainted by classical technical noise, which potentially compromises the security of such a device. To safeguard against this, we propose and experimentally demonstrate an approach that produces side-information independent randomness. We present a method for maximising such randomness contained in a number sequence generated from a given quantum-to-classical-noise ratio. The detected photocurrent in our experiment is shown to have a real-time random-number generation rate of 14 (Mbit/s)/MHz. Next, we study the one-sided device-independent (1sDI) quantum key distribution scheme in the context of continuous variables. By exploiting recently proven entropic uncertainty relations, one may bound the information leaked to an eavesdropper. We use such a bound to further derive the secret key rate, that depends only upon the conditional Shannon entropies accessible to Alice and Bob, the two honest communicating parties. We identify and experimentally demonstrate such a protocol, using only coherent states as the resource. We measure the correlations necessary for 1sDI key distribution up to an applied loss equivalent to 3.5 km of fibre transmission. The second part of this thesis concerns the improvement in the transmission of a quantum state. We study two approximate implementations of a probabilistic noiseless linear amplifier (NLA): a physical implementation that truncates the working space of the NLA or a measurement-based implementation that realises the truncation by a bounded postselection filter. We do this by conducting a full analysis on the measurement-based NLA (MB-NLA), making explicit the relationship between its various operating parameters, such as amplification gain and the cut-off of operating domain. We compare it with its physical counterpart in terms of the Husimi Q-distribution and their probability of success. We took our investigations further by combining a probabilistic NLA with an ideal deterministic linear amplifier (DLA). In particular, we show that when NLA gain is strictly lesser than the DLA gain, this combination can be realised by integrating an MB-NLA in an optical DLA setup. This results in a hybrid device which we refer to as the heralded hybrid quantum amplifier. A quantum cloning machine based on this hybrid amplifier is constructed through an amplify-then-split method. We perform probabilistic cloning of arbitrary coherent states, and demonstrate the production of up to five clones, with the fidelity of each clone clearly exceeding the corresponding no-cloning limit.
In this thesis, we investigate how the combination of quantum physics and information theory could deliver solutions at the forefront of information security, and, in particular, we consider two focus applications: randomness extraction as applied to quantum random number generators and classical processing algorithms for quantum key distribution (QKD). We concentrate on practical applications for such tools. We detail the implementation of a randomness extractor for a commercial quantum random number generator, and we evaluate its performance based on information theory. Then, we focus on QKD as applied to a specific experimental scenario, that is, the one of free-space quantum links. Commercial solutions with quantum links operating over optical fibers, in fact, already exist, but suffer from severe infrastructure complexity and cost overheads. Free-space QKD allows for a higher flexibility, for both terrestrial and satellite links, whilst experiencing higher attenuation and noise at the receiver. In this work, its feasibility is investigated and proven in multiple experiments over links of different length, and in various channel conditions. In particular, after a thorough analysis of information reconciliation protocols, we consider finite-key effects as applied to key distillation, and we propose a novel adaptive real-time selection algorithm which, by leveraging the turbulence of the channel as a resource, extends the feasibility of QKD to new noise thresholds. By using a full-fledged software for classical processing tailored for the considered application scenario, the obtained results are analyzed and validated, showing that quantum information security can be ensured in realistic conditions with free-space quantum links.
Quantum technologies are rapidly transforming the landscape of secure communication and computation by exploiting the fundamental principles of quantum mechanics. Among these, Quantum Key Distribution (QKD) offers an unprecedented level of security in communications, provided by physical principles rather than computational assumptions. However, the realization of scalable and practical quantum networks remains challenging. Key technological hurdles include the need for high-sensitivity and high-speed single-photon detectors, stable and integrated optical platforms, and robust and fast true random number generators. Overcoming these challenges is essential to bridge the gap between laboratory demonstrations and the real-world deployment of quantum networks. This thesis addresses some of these critical issues through work in single-photon detection technologies, integrated QKD system implementation, and high-speed Quantum Random Number Generators (QRNGs). Single-Photon Avalanche Diodes are fundamental detectors for quantum technologies, enabling high-sensitivity measurements at the single-photon level. This thesis investigates the operating principles, limitations, and advancements of Single-Photon Avalanche Diodes, with a focus on high-speed dual-anode designs. Their photon detection efficiency, dark count rate, afterpulsing probability, and timing jitter are analyzed, while optimized control electronics and capacitance management strategies are developed. The characterization of Single-Photon Avalanche Diodes demonstrates their suitability for quantum communication, particularly in quantum key distribution and asynchronous heralded photon sources, with pathways for further improvements in speed and integration. Building on these detector technologies, a proof-of-principle integrated quantum key distribution system based on the BB84 protocol is presented. The system incorporates time-bin encoding, decoy-state methods, and photonic integrated circuits for both transmitter and receiver. Key challenges, such as chromatic dispersion and interferometer stability, are addressed through novel circuit and system-level designs. System-level testing and field deployment validate the feasibility of compact, high-performance integrated quantum key distribution implementations. Finally, the thesis explores quantum random number generation through semi-device-independent protocols. An experimental homodyne-based SDI-QRNG is realized, supported by custom transimpedance amplifier design and photonic integrated circuit integration. The setup achieves high bandwidth, strong quantum-to-classical noise separation, and stable operation, enabling practical random number generation rooted in fundamental quantum processes. Overall, this work advances the development of integrated quantum technologies by improving room-temperature single-photon detection, demonstrating scalable QKD prototypes, and implementing a high-speed QRNG. Together, these contributions pave the way toward secure, integrated quantum communication systems and practical quantum information applications ready to deploy in real world networks.
Cryptography plays a key role in today's digital society. The most widely used encryption method is the public-key algorithm, but its security relies on unproven computational assumptions. In contrast, quantum key distribution (QKD) is an unbreakable encryption algorithm. QKD has been proven to provide information-theoretic security rigorously based on the laws of physics. Despite the beauty of QKD theory, in practice, QKD still suffers from practical side-channel attacks, due to the imperfections of real-life implementations. Such discrepancy between the theory and the practice is fatal to the security of a QKD system. This thesis addresses a number of the discrepancies in a standard QKD implementation, and it offers efficient countermeasures to resolve these discrepancies. A novel protocol -- measurement-device-independent (MDI) QKD - was proposed to remove all detector side-channel attacks. To apply MDI-QKD in real life, I have proposed practical protocols, conducted the first rigorous finite-key security proof, designed methodologies to optimize parameters, and demonstrated MDI-QKD experimentally. These results allow MDI-QKD mature enough to be implemented in real-life. Based on the framework of MDI-QKD, the only security issue left is the source. To solve the loophole due to the imperfect encoder, I have experimentally implemented a recent protocol, which enables QKD loss-tolerant to source flaws. This experiment for the first time shows secure QKD with imperfect state preparations at long distances. Another vulnerable part in the source is the ability to generate true randomness. I have developed a compact and cost-effective quantum random number generator (QRNG) prototype, which has a real-time generation rate of 1 Gbit/s with excellent stability. The simplicity and speed of my prototype prove the feasibility of a robust, low-cost, and fast QRNG. In summary, MDI-QKD enables new scientific developments in the field of quantum cryptography. This thesis solves the major practical issues in the implementation of MDI-QKD. The combination of MDI-QKD, imperfect sources, and QRNG consists of the foundation of a future practical side-channel-free QKD protocol. The success of this protocol can substitute current classical cryptosystems by offering people an unconditionally secure infrastructure, which will be the ultimate dream for researchers in secure communication.
One of the most fascinating recent developments in research has been how different disciplines have become more and more interconnected. So much so that fields as disparate as information theory and fundamental physics have combined to produce ideas for the next generation of computing and secure information technologies, both of which have far reaching consequences. For more than fifty years Moore's law, which describes the trend of the transistor's size shrinking by half every two years, has proven to be uncannily accurate. However, the computing industry is now approaching a fundamental barrier as the size of a transistor approaches that of an individual atom and the laws of physics and quantum mechanics take over. Rather then look at this as the end, quantum information science has emerged to ask the question of what additional power and functionality might be realized by harnessing some of these quantum effects. This thesis presents work on the sub-field of quantum cryptography which seeks to use quantum means in order to assure the security of ones communications. The beauty of quantum cryptographic methods are that they can be proven secure, now and indefinitely into the future, relying solely on the validity of the laws of physics for their proofs of security. This is something which is impossible for nearly all current classical cryptographic methods to claim. \n \nThe thesis begins by examining the first implementation of an entangled quantum key distribution system over two free-space optical links. This system represents the first test-bed of its kind in the world and while its practical importance in terrestrial applications is limited to a smaller university or corporate campus, the system mimics the setup for an entangled satellite system aiding in the study of distributing entangled photons from an orbiting satellite to two earthbound receivers. Having completed the construction of a second free-space link and the automation of the alignment system, I securely distribute keys to Alice and Bob in two distant locations separated by 1,575 m with no direct line-of-sight between them. I examine all of the assumptions necessary for my claims of security, something which is particularly important for moving these systems out of the lab and into commercial industry. I then go on to describe the free-space channel over which the photons are sent and the implementation of each of the major system components. I close with a discussion of the experiment which saw raw detected entangled photon rates of 565 s^{-1} and a quantum bit error rate (QBER) of 4.92% resulting in a final secure key rate of 85 bits/s. Over the six hour night time experiment I was able to generate 1,612,239 bits of secure key. \n \nWith a successful QKD experiment completed, this thesis then turns to the problem of improving the technology to make it more practical by increasing the key rate of the system and thus the speed at which it can securely encrypt information. It does so in three different ways, involving each of the major disciplines comprising the system: measurement hardware, source technology, and software post-processing. First, I experimentally investigate a theoretical proposal for biasing the measurement bases in the QKD system showing a 79% improvement in the secret key generated from the same raw key rates. Next, I construct a second generation entangled photon source with rates two orders of magnitude higher than the previous source using the idea of a Sagnac interferometer. More importantly, the new source has a QBER as low as 0.93% which is not only important for the security of the QKD system but will be required for the implementation of a new cryptographic primitive later. Lastly, I study the free-space link transmission statistics and the use of a signal-to-noise ratio (SNR) filter to improve the key rate by 25.2% from the same amount of raw key. The link statistics have particular relevance for a current project with the Canadian Space Agency to exchange a quantum key with an orbiting satellite - a project which I have participated in two feasibility studies for. \n \nWanting to study the usefulness of more recent ideas in quantum cryptography this thesis then looks at the first experimental implementation of a new cryptographic primitive called oblivious transfer (OT) in the noisy storage model. This primitive has obvious important applications as it can be used to implement a secure identification scheme provably secure in a quantum scenario. Such a scheme could one day be used, for example, to authenticate a user over short distances, such as at ATM machines, which have proven to be particularly vulnerable to hacking and fraud. Over a four hour experiment, Alice and Bob measure 405,642,088 entangled photon pairs with an average QBER of 0.93% allowing them to create a secure OT key of 8,939,150 bits. As a first implementer, I examine many of the pressing issues currently preventing the scheme from being more widely adopted such as the need to relax the dependance of the OT rate on the loss of the system and the need to extend the security proof to cover a wider range of quantum communication channels and memories. It is important to note that OT is fundamentally different than QKD for security as the information is never physically exchanged over the communication line but rather the joint equality function f(x) = f(y) is evaluated. Thus, security in QKD does not imply security for OT. \n \nFinally, this thesis concludes with the construction and initial alignment of a second generation free-space quantum receiver, useful for increasing the QKD key rates, but designed for a fundamental test of quantum theory namely a Svetlichny inequality violation. Svetlichny's inequality is a generalization of Bell's inequality to three particles where any two of the three particles maybe be non-locally correlated. Even so, a violation of Svetlichny's inequality shows that certain quantum mechanical states are incompatible with this restricted class of non-local yet realistic theories. Svetlichny's inequality is particularly important because while there has been an overwhelming number of Bell experiments performed testing two-body correlations, experiments on many-body systems have been few and far between. Experiments of this type are particularly valuable to explore since we live in a many-body world. The new receiver incorporates an active polarization analyzer capable of switching between measurement bases on a microsecond time-scale through the use of a Pockels cell while maintaining measurements of a high fidelity. Some of the initial alignment and analysis results are detailed including the final measured contrasts of 1:25.2 and 1:22.6 in the rectilinear and diagonal bases respectively.
Although mathematical cryptography has been widely used, its security has only been proven under certain assumptions such as the computational power of opponents. As an alternative, quantum communication, in particular quantum key distribution (QKD) can get around unproven assumptions and achieve unconditional security. However, the key generation rate of practical QKD systems is limited by device imperfections, excess noise from the quantum channel, limited rate of true random-number generation, quantum entanglement preparation, and/or post-processing efficiency. This dissertation contributes to improving the performance of quantum communication systems. First, it proposes a new continuous-variable QKD (CVQKD) protocol that loosens the efficiency requirement on post-processing, a bottleneck for long-distance CVQKD systems. It also demonstrates an experimental implementation of the proposed protocol. To achieve high rates, the CVQKD experiment uses a continuous-wave local oscillator (CWLO). The excess noise caused by guided acoustic-wave Brillioun scattering (GAWBS) is avoided by a frequency-shift scheme, resulting in a 32 dB noise reduction. The statistical distribution of the GAWBS noise is characterized by quantum tomography. Measurements show Gaussian statistics upto 55 dB of dynamical range, which validates the security calculations in the proposed CVQKD protocol. True random numbers are required in quantum and classical cryptography. A second contribution of this thesis is that it experimentally demonstrates an ultrafast quantum random-number generator (QRNG) based on amplified spontaneous emission (ASE). Random numbers are produced by a multi-mode photon counting measurement on ASE light. The performance of the QRNG is analyzed with quantum information theory and verified with NIST standard random-number test. The QRNG experiment demonstrates a random-number generation rate at 20 Gbits/s. Theoretical studies show fundamental limits for such QRNGs. Quantum entanglement produced in nonlinear optical processes can help to increase quantum communication distance. A third contribution is the research on nonlinear optics of graphene, a novel 2D material with unconventional physical properties. Based on a quantum-dynamical model, optical responses of graphene are derived, showing for the first time a link between the complex linear optical conductivity and the quantum decoherence. Nonlinear optical responses, in particular four-wave mixing, is studied for the first time. The theory predicts saturation effects in graphene and relates the saturation threshold to the ultrafast quantum decoherence and carrier relaxation in graphene. For the experimental part, four-wave mixing in graphene is demonstrated. Twin-photon production in graphene is under investigation.