Roderick Cochran

As the proliferation of automation in smart transportation continues, there is a need to secure communication links of “on-the-go” future mobile platforms. In this effort, we implement decoy-state quantum key distribution (QKD), which provides provably secure communication, to mobile platforms such as drones and vehicles. Unlike demonstrations in fiber of fixed point-to-point, QKD between mobile platforms provides unique challenges such as designing systems with reduced size, weight, and power, establishing a stable line-of-sight as the platforms are in motion, and maintaining performance over a wide operating temperature range, etc. We design our QKD transmitter and receiver using a modular design that is platform-agnostic. This allows us to deploy the same QKD system on an octocopter drone and a car without any hardware or software modifications. We describe critical subsystems including our resonant-cavity QKD source, custom prepare and measure optics, pointing, acquisition, and tracking system, single-photon detector, field-programmable gate array-based time-tagger, and qubit-based time-synchronization algorithm. Our achievements include drone-to-drone QKD, drone-to-car quantum transmission, and high-speed (70 mph) vehicle-to-vehicle quantum transmission on a U.S. Interstate Highway.

Aerial Drones have been used in defense applications for decades, but recently the commercial use cases of drones have significantly increased to include package delivery, taxis, aerial photography, disaster relief, and even delivery of COVID-19 vaccines. Typically drones rely on a plurality of in-flight sensors for navigation and external command and control signals for tasking. As drones continue to proliferate our skies, the need to secure communication between drone constellations will become increasingly important, since the unmanned nature of drones offers new attack vectors which are not present for platforms with human operators. Quantum security protocols such as Quantum Key Distribution (QKD) offer unique advantages over classical approaches to secure the command-and-control signals of current and future drone constellations. In this presentation, we will report progress towards demonstrating QKD between two drones in flight. Critical subsystems and characterization data will be presented such as the QKD source, which is based on a resonant cavity Light Emitting Diodes (LED), as well as a secondary QKD source based on a fiber-coupled polarization modulator. The Pointing Acquisition, and Tracking (PAT) system provides both course alignment using Infrared (IR) beacons and cameras and fine alignment is achieved using Fast Steering Mirrors (FSM) and feedback position sensors. We will discuss QKD optical payloads, which were fabricated using a 3D printed bench to achieve a compact size and weight, single-photon detectors, an FPGA-based time-tagger and two time-synchronization approaches. Providing quantum security to emerging drone networks, including airborne and ground-based systems such as self-driving cars, is a critical enabling technology required to extend the future quantum internet to mobile platforms, with could play an essential role, e.g., for reconfigurable distributed quantum sensors.

While most current quantum network nodes are connected via fiber-based or free-space fixed point-to-point links, there have been many advancements in the last decade that expand these nodes to include mobile, re-configurable, and wireless platforms such as uncrewed aerial vehicles (UAVs) and satellites. The size, weight, and power (SWaP) restrictions of these platforms pose constraints that potentially impact the system performance of mobile nodes. Here, we will discuss our progress towards developing a low-SWaP mobile quantum key distribution (QKD) platform that can exchange quantum-secured random keys between both drones and cars. We implement a finite-key security proof that incorporates system imperfections in state preparation and analysis, including channel losses. These imperfections, present in any system, require consideration during key consolidation to minimize information leakage. We demonstrate average finite secure key rates between mobile platforms up to 19.6 kbit/s.

Quantum key distribution (QKD) provides a method for two users to exchange a provably secure key, which requires synchronizing the user’s clocks. Qubit-based synchronization protocols directly use the transmitted quantum states and thus avoid the need for additional classical synchronization hardware, but previous approaches sacrifice secure key either directly or indirectly. Here, we introduce a Bayesian probabilistic algorithm that incorporates all published information to efficiently find the clock offset without sacrificing any secure key [1]. Additionally, the output of the algorithm is a probability, which allows us to quantify our confidence in the synchronization. Our experimental system employs an efficient three-state BB84 prepare-and-measure protocol with decoy states. Our algorithm exploits the correlations between Alice’s published basis and mean photon number choices (which must already be published for the protocol) and Bob’s measurement outcomes to probabilistically determine the most likely clock offset. We perform cross-correlations using Fast Fourier Transforms to count the number of each type of event pairing for each potential offset (e.g., how many times Alice sent a decoy state in the horizontal/vertical polarization basis and Bob registered a click in the horizontal detector). Taking these along with a lookup table for the probabilities of the different event pairings, we determine the synchronization probability of the different potential offsets using Bayesian analysis. To demonstrate the robust nature of this algorithm, we tracked its performance using simulated data with varying parameters. We find that we can achieve a 95% synchronization confidence using a string length of only 4,140 communication bin widths, meaning we can tolerate clock drift approaching 1 part in 4,140 in this example when simulating this system with a dark count probability per communication bin width of 8⨉10-4 and a received mean photon number of 0.01. The relationship between the received mean photon number and the number of communication bin widths required to achieve a 95% synchronization confidence is shown in Fig. 1. We applied this algorithm to data collected from our drone-to-done QKD experiments, with a received mean photon number of 0.043, achieving quantum bit error rates of 0.0106, 0.0287, and 0.0361 for our 3 states.

Quantum key distribution (QKD) systems provide a method for two users to exchange a provably secure key that can be used to establish an unconditionally secure communication channel. Here we present an FPGA-controlled prepare-and-measure BB84 polarization-based decoy state protocol using light-emitting diodes (LEDs). Our setup uses three separate LEDs driven by a field-programmable gate array (FPGA) that go through different optical paths that set the state of polarization. Each LED is connected to two GPIO pins via a different resistive path. By setting one pin to high impedance and driving the other with a nanosecond-scale electrical signal, we can choose between signal and decoy states. We can thus send 3 signal states, 3 decoy states, and 3 vacuum states. To prevent side-channel attacks multi-source QKD systems require that each state is indistinguishable from the others in the spatial, spectral, and temporal degrees-of-freedom on the photon. We do this by passing the 3 photonic wavepackets through the same single-mode fiber and 1-nm-bandwith spectral filter and use dynamic shifting of the FPGA phase-locked-loops to control the phase and the width of the electrical pulses that drive the LEDs, which allows us to control the optical pulses produced by the LEDs. Both spectral and temporal profiles are shown in Figure 1. We control the timing of the photonic wavepackets to a resolution of 78 ps. Additionally, we use the FPGA to generate true random states as required by the BB84 protocol. To quantify the indistinguishability of Alice’s various states, we use the mutual information to calculate the fraction of the final sifted key that an eavesdropper would know after making temporal and/or spectral measurements on every state that is sent. We are able to achieve 2.39e-05 and 4.31e-05 mutual information fraction leaked in the spectral and temporal waveforms, respectively. Furthermore we put our scheme into practice with a simple tabletop QKD setup where we are able to achieve 1.7% quantum bit-error rate (QBER) in the L/R bases and 2.1% QBER in the H/V bases. Additionally, our system's SWaP restrictions make it very desirable for highly mobile platforms such as drones.

Quantum key distribution (QKD) provides a method for two users to exchange a provably secure key, which requires synchronizing the user’s clocks. Qubit-based synchronization protocols directly use the transmitted quantum states and thus avoid the need for additional classical synchronization hardware, but previous approaches sacrifice secure key either directly or indirectly. Here, we introduce a Bayesian probabilistic algorithm that incorporates all published information to efficiently find the clock offset without sacrificing any secure key [1]. Additionally, the output of the algorithm is a probability, which allows us to quantify our confidence in the synchronization. For demonstration purposes, we present a model system with accompanying simulations of an efficient three-state BB84 prepare-and-measure protocol with decoy states. Our algorithm exploits the correlations between Alice’s published basis and mean photon number choices (which must already be published for the protocol) and Bob’s measurement outcomes to probabilistically determine the most likely clock offset. We perform cross-correlations using Fast Fourier Transforms to count the number of each type of event pairing for each potential offset (e.g., how many times Alice sent a decoy state in the horizontal/vertical polarization basis and Bob registered a click in the horizontal detector). Taking these along with a lookup table for the probabilities of the different event pairings, we determine the synchronization probability of the different potential offsets using Bayesian analysis. In our simulations, we find that we can achieve a 95% synchronization confidence using a string length of only 4,140 communication bin widths, meaning we can tolerate clock drift approaching 1 part in 4,140 in this example when simulating this system with a dark count probability per communication bin width of 8⨉10-4 and a received mean photon number of 0.01. The relationship between the received mean photon number and the number of communication bin widths required to achieve a 95% synchronization confidence is shown in Fig. 1.

Quantum key distribution (QKD) systems provide a method for two users to exchange a provably secure key that can be used to securely exchange a cryptographic key. In prepare-and-measure QKD protocols, the indistinguishability of states is an important aspect for preventing side-channel attacks. Here we consider the indistinguishability of states in a prepare-and-measure three-state BB84 polarization-based decoy state protocol using light-emitting diodes (LEDs). In addition, our system is designed to operate under size, weight, and power (SWaP) restrictions such as that needed for drone-based QKD. Our setup uses three separate LEDs driven by a field-programmable gate array (FPGA) that go through different optical paths that set the state of polarization. Each LED is connected to two GPIO pins via a different resistive path. By setting one pin to high impedance and driving the other with a nanosecond-scale electrical signal, we can choose between signal and decoy states. We can thus send 3 signal states, 3 decoy states, and 3 vacuum states, using only 3 separate sources driven by a single low-cost and light-weight FPGA. We must guarantee that these sources are indistinguishable from each other in the spatial, spectral, and temporal degrees-of-freedom on the photon. We make them nearly indistinguishable by passing the 3 photonic wavepackets through the same single-mode fiber and 1-nm-bandwith spectral filter, and use dynamic shifting of the FPGA phase-locked-loops to control the phase and the width of the electrical pulses that drive the LEDs, which allows us to control the optical pulses produced by the LEDs. We control the timing of the photonic wavepackets to a resolution of 250 ps. To quantify spectral indistinguishability, we measure filtered spectra for all states, which are overlaid in Fig. 1a, and find that their overlap is 94.6%. To measure the temporal indistinguishability, we drive a single LED with a 10 ns wide electrical signal at a repetition rate of 12.5 MHz. The resulting photonic wavepacket is measured by a single-photon detector whose electrical output is measured by a time-to-digital converter and histogrammed. The temporal waveforms of all 6 states are overlaid and shown in Fig. 1b with a measured overlap of 97.1%.