Experimental Quantum-enhanced Cryptographic Remote Control

TL;DR

This paper demonstrates a quantum-enhanced cryptographic remote control system for UAVs, utilizing quantum randomness and one-time pad encryption to ensure secure, tamper-proof command transmission in IoT environments.

Contribution

It introduces a practical quantum cryptography scheme for secure remote control, preloading quantum keys for UAV command encryption, enhancing security beyond standard quantum communication methods.

Findings

Commands are fully encrypted and chaotic without matching keys.

UAV only responds to correctly encrypted commands, showing immunity to interference.

Quantum randomness verified as secure and effective for control encryption.

Abstract

The Internet of Things (IoT), as a cutting-edge integrated cross-technology, promises to informationize people's daily lives, while being threatened by continuous challenges of eavesdropping and tampering. The emerging quantum cryptography, harnessing the random nature of quantum mechanics, may also enable unconditionally secure control network, beyond the applications in secure communications. Here, we present a quantum-enhanced cryptographic remote control scheme that combines quantum randomness and one-time pad algorithm for delivering commands remotely. We experimentally demonstrate this on an unmanned aircraft vehicle (UAV) control system. We precharge quantum random number (QRN) into controller and controlee before launching UAV, instead of distributing QRN like standard quantum communication during flight. We statistically verify the randomness of both quantum keys and the converted ciphertexts to check the security capability. All commands in the air are found to be completely chaotic after encryption, and only matched keys on UAV can decipher those commands precisely. In addition, the controlee does not response to the commands that are not or incorrectly encrypted, showing the immunity against interference and decoy. Our work adds true randomness and quantum enhancement into the realm of secure control algorithm in a straightforward and practical fashion, providing a promoted solution for the security of artificial intelligence and IoT.

Figures

Tables

Table 1. Table 1: Balance and runs properties. P ​ - ​ v ​ a ​ l ​ u ​ e ​ s 𝑃 - 𝑣 𝑎 𝑙 𝑢 𝑒 𝑠 P\text{-}values for uniformity check, and proportions for examination of the sequences that pass a certain statistical test (Success Rate). 20 pieces of commands are tested.

Test Index	$P\text{-}value$	Proportion	Result
Frequency	0.4861	1	SUCCESS
Runs	0.4719	1	SUCCESS

Table 2. Supplementary Table 1: Intercepted values of plaintexts sent by the UAV controller. Red values show the difference.

Connection	36	77	60	16	105	221	5	221	5	219	5	220	5	220	5	220
	5	221	5	0	0	166	0	0	0	0	0	0	221	255	223	255
Backward	36	77	60	16	105	221	5	221	5	219	5	232	3	221	5	220
	5	221	5	0	0	149	0	0	0	0	0	0	221	191	215	255
Turn Left	36	77	60	16	105	221	5	220	5	234	3	219	5	220	5	220
	5	221	5	0	0	168	0	0	0	0	0	0	221	255	215	255
Turn Right	36	77	60	16	105	221	5	221	5	208	7	219	5	220	5	221
	5	220	5	0	0	168	0	0	0	0	0	0	221	255	215	255
Forward	36	77	60	16	105	220	5	220	5	220	5	208	7	220	5	219
	5	221	5	0	0	168	0	0	0	0	0	0	221	255	215	255

Full text

With the rapid development of artificial intelligence and IoT, greater demands are being placed on the security by growing hacking incidents. To implement cryptographic remote control, two general types of key-based algorithms, public-key and symmetric, are being widely investigated. Public-key algorithms use two different keys for encryption and decryption, and are often based on computational complexity; while they are imperfect in the real world for being slow, and vulnerable to chosen-plaintext attacks Schneier_1996 . Conventional symmetric algorithms require that communication parties share matched and secret keys in advance; while the security of such algorithms rely on the shared keys. One-time pad Gingerich_Isis_1996 , as a powerful symmetric algorithm, has been proved by Claude Shannon to be impossible to crack Shannon_Bell_1949 , as long as crucial problems of generating and sharing real random sequences are settled.

Randomness is a fundamental resource with significant applications in cryptography and numerical simulation. Real random sequences, however, are hard to generate mathematically Knuth_1981 , but have to rely on unpredictable physical processes Stefanov_JMO_2000 ; Jennewein_RSI_2000 ; Uchida_nphoton_2008 ; Dynes_APL_2008 ; Fiorentino_PRA_2006 . Although different mechanics, such as chaotic effects Stojanovski_IEEE_2001_partI ; Stojanovski_IEEE_2001_partII , thermal noise Petrie_Computers_2004 , biometric parameters Szczepanski_IEEE_2000 and free-running oscillators Kohlbrenner_2004 are employed in the generation of physical random number, they are faced with some problems like hard to detect failure Herrero_RMP_2017 . The inherent uncertainty of quantum mechanics makes quantum systems an excellent stochastic source, with the fact that a single photon incident on a 50:50 beam splitter be transmitted or reflected is intrinsically random. More importantly, the randomness is precisely balanced and immune to environmental perturbations.

Sharing randomness is another crucial problem to be settled for realizing symmetric algorithms. One best-known scheme is quantum key distribution (QKD), which is quite mature so far for applications, with enormous progresses Bennett_IEEE_1984 ; Ekert_PRL_1991 ; Lo_nphoton_2014 ; JiLing_OE_2016 and is even ready for constructing secure networks Qiu_Nature_2014 ; Elliott_nCommun_2006 ; Fujiwara_OE_2011 ; Frohlich_PRA_2013 ; Tang_PRX_2016 . Nevertheless, many situations of IoT control are not compatible with QKD schemes. For example, sensor networks require low cost, low power and miniature devices, which is hard to be met by QKD systems Gubbi_FGCY_2012 , especially for large-scale and distributed sensor networks.

Interestingly, in many situations, real-time sharing of randomness is not really necessary in practice. For instance, UAVs or satellites are essentially well identified before being launched, and are well isolated with other parties during their missions. For all these situations, we could precharge quantum keys into controllers and controlees, and implement cryptographic remote control with quantum enhancement in a straightforward and practical way. In this work, we present this quantum-enhanced cryptographic remote control scheme that combines quantum randomness and one-time pad algorithm for delivering commands remotely, and experimentally demonstrate this on a UAV control system.

The cryptographic system is composed of three subsystems: key management unit, controller unit and controlee unit, as is shown in Figure 1(a). The randomness derived from quantum nature of single photons is charged into IoT devices through secure key storage (SKS) chips to perform one-time pad encryption and decryption. The connection of flexible small-scale SKS chips to QRN generator is realized by a home-built key management unit. SKS chips are planted into controllers and controlees before they are detached. Commands are encrypted by one-time pad algorithm with a certain section of keys, which can only be decrypted correctly by the controlee with corresponding keys. On the controller side, with QRN keys, a bitwise exclusive OR is performed on commands before sent; and conversely, with identical keys, commands can be deciphered and executed on the controlee side, as is shown in Figure 1(b).

The data transmission diagram of our UAV control system is illustrated in Figure 2(a). Encryption keys are generated by Quantis, which is a reliable QRN generator, employing a quantum process as the source of randomness, and producing random sequences at a bit rate of 4 Mb/s. To be specific, a photon incident on a semi-transparent mirror will be reflected with half the probability, leading to a “0”; or transmitted with half the probability, leading to an “1”. A microcontroller is dedicated for charging or updating quantum keys into SKS chips.

The successful execution of one-time pad algorithm depends on the synchronization of keys. Unfortunately, it happens that commands get lost or error, leading to key mismatch between controllers and controlees. Any minor key mismatch may cause control system failure. To solve this problem, we assign unique address information to each command, so that each command with quantum keys is independent and well labeled. Once error happens, the corresponding command will be discarded directly together with its encryption and decryption keys to maintain the synchronization. Such address information doesn’t have to be encrypted, because they include no information about commands.

The randomness of quantum keys is a crucial parameter that determines system security. Quantum random number based on the uncertainty principle of quantum mechanics provides the honest-to-goodness randomness in the world, with the properties of unpredictable and unreproducible Schneier_1996 . We use NIST suites to perform statistical tests Rukhin_APL_2015 , and the final results are shown in Figure 2(b). The results are $P\text{-}valuess$ of all 15 tests: indicating how a sequence is identical to purely random number, ideally $P\text{-}values$ equals to one. The results of NIST tests prove an excellent statistical randomness of our quantum keys.

Furthermore, according to one-time pad algorithm, where ciphertexts are the XOR values of quantum keys and plaintexts, the randomness of quantum keys determines that of ciphertexts. We intercept a section of commands in the air sent by the controller, and test them with three characteristics of random binary sequences proposed by Gobomb: balance property, runs property and auto-correlation property Golomb_SF_1967 . The good properties of balance, runs (see Table I), and auto-correlation (see Figure3) indicate that our ciphertexts are statistically random. Meanwhile, since the quantum keys are unpredictable and unrepeatable, the ciphertexts intercepted here are expected and experimentally observed to be truly random, which is impossible to be deciphered without matched quantum keys.

Experimental commands between the remote control and the aircraft are shown in Figure 4. We intercept five pieces of different functional commands, with and without one-time pad encryption for comparison. For plaintexts, we can see that five repeating commands share exactly identical values; while for ciphertexts, five repeating commands are bought into chaos, and there is no correlation between any two commands or even any two bytes, which guarantees the security as have been proved statistically in Table I and Figure 3, according to three postulates proposed by Golomb Golomb_SF_1967 .

Since the security of the commands depends on the one-time pad, the communication capacity in this cryptographic control scheme is mainly limited by the number of pre-charged QRNs. To extend the key updating period, on one hand, the capacity of secure key storage device should be large enough, while it might take more bytes in the commands for storing keys’ address information. On the other hand, the encryption commands can be optimized according to different structures, and some trivial information in a certain command could be ignored to save keys, as well as to increase decryption speed.

In summary, we have proposed and experimentally demonstrated a quantum-enhanced cryptographic remote control scheme that combines quantum randomness and one-time pad algorithm for delivering commands remotely. The quantum-enhanced cryptographic scheme are expected to be generalized to bidirectional systems: controlees can be securely controlled and also be able to send encrypted recorded flight data back to controllers, such as position, direction and speed. Besides, the point-to-point solution can be extended to point-to-multipoint or distributed networks. More importantly, such scheme can be combined with fixed QKD channels Ursin_nphy_2007 ; Frohlich_PRA_2013 ; Marsili_nphoton_2013 ; Wang_nphoton_2013 for long-distance quantum keys charging, providing a flexible solution for control security of artificial intelligence and IoT in large scale.

Acknowledgments

The authors thank J.-W. Pan for helpful discussions. This work was supported by National Key R&D Program of China (2017YFA0303700), the National Natural Science Foundation of China (NSFC) (61734005, 11761141014, 11690033), the Science and Technology Commission of Shanghai Municipality (STCSM) (15QA1402200, 16JC1400405, 17JC1400403), the Shanghai Municipal Education Commission (SMEC)(16SG09, 2017-01-07-00-02-E00049), and the Zhiyuan Scholar Program (ZIRC2016-01). X.-M.J. acknowledges support from the National Young 1000 Talents Plan.

Supplemental Information: Experimental Quantum-enhanced Cryptographic Remote Control

Supplementary Note 1: Commands reliability testing

To test the reliability of cryptographic remote control scheme, commands emitted by the UAV controller are intercepted and processed by another microcontroller (STM32). As for general cases, without encryption, the five-time repeated commands intercepted are demonstrated in FIG. S1(a). Inserted is the command structure which consists of 32 bytes, and each byte consists of eight “0” or “1” bits. Thus the value of a byte ranges from 0 to 255. The key address information added in the command is not encrypted, which contains no information without local quantum key storage devices.

As for situations with quantum-enhanced encryption, commands sent by the UAV controller are encrypted by one-time pad algorithm. The ciphertexts intercepted are shown in FIG. S1(b). We can see that the five-time repeated commands has been brought into chaos, except for the bytes which carry key address information (marked in shadows). According to the key address information, the controllee can read out quantum keys and perform one-time pad decryption process. The randomness of ciphertexts has been tested and presented in Figure 3 and Table I in the main text.

A brief structure of a command is shown in the insert of FIG. S1. Even different functional commands share parts of the same bytes to maintain stable communication process. Therefore, it is not necessary to encrypt all 32 bytes in a command, and we can just perform one-time pad encryption on the bytes that carry fatal information, to save quantum keys. Values of plaintexts of five commands (blue lines) presented in Figure 4 in the main text are shown in Table I. According to the command structure, they are similar but not the same.