With the growing interest in underwater acoustic networks (UWANs), such as in the Internet of underwater things, we need mechanisms that guarantee communication security. In turn, security mechanisms often require legitimate devices to share secret keys. In this paper, we propose a physical-layer key generation (PKG) scheme for UWANs where a pair of legitimate users process the respective channel observation by two neural networks (NNs) to extract a raw key. Inspired by the lower bound on the secret key capacity, we train key extractors to maximize reciprocity and uniformity while minimizing the information leakage to the eavesdropper. Specifically, a custom architecture and loss function have been designed to accomplish these tasks. The performance of the proposed PKG mechanism based on machine learning (ML) has been evaluated considering both additive white Gaussian noise (AWGN) and UWAN scenarios, with the latter using measurements extracted from an experimental dataset. The results highlight the benefits of the proposed mechanism over the state-of-the-art solutions.
With the growing interest in underwater acoustic networks (UWANs), such as in the Internet of underwater things, we need mechanisms that guarantee communication security. In turn, security mechanisms often require legitimate devices to share secret keys. In this paper, we propose a physical-layer key generation (PKG) scheme for UWANs where a pair of legitimate users process the respective channel observation by two neural networks (NNs) to extract a raw key. Inspired by the lower bound on the secret key capacity, we train key extractors to maximize reciprocity and uniformity while minimizing the information leakage to the eavesdropper. Specifically, a custom architecture and loss function have been designed to accomplish these tasks. The performance of the proposed PKG mechanism based on machine learning (ML) has been evaluated considering both additive white Gaussian noise (AWGN) and UWAN scenarios, with the latter using measurements extracted from an experimental dataset. The results highlight the benefits of the proposed mechanism over the state-of-the-art solutions.
Underwater networks have become an important topic of research due to their potential impact on several military and civil applications. To secure these communications, many protocols require the users to have symmetric secret keys. To distribute the keys, we can resort to physical layer key generation schemes, where the keys are generated by each user from the channel itself, by exploiting the environment as a source of randomness. We focus on the advantage distillation step of the secret key agreement (SKA) procedure, where the users extract the bit sequence from the channel observations. We propose an adversarial autoencoder (AAE) model for advantage distillation, which has multiple tasks: it enhances the reciprocity between legitimated channel features, decreases the reciprocity to the attacker channel features, and provides uniformly distributed bit sequences. The proposed approach is adaptive and directly derived from the training dataset, thus it does not require the a priori knowledge of the channel statistics. The solution is tested using channel observations collected from a sea experiment and compared to existing advantage distillation solutions, showing a significant improvement over the state of the art.
With the growing use of underwater acoustic communications and the recent adoption of standards in this field, it is becoming increasingly important to secure messages against eavesdroppers. In this paper, we focus on a physical-layer security solution to generate sequences of random bits (keys) between two devices (Alice and Bob) belonging to an underwater acoustic network (UWAN); the key must remain secret to a passive eavesdropper (Eve) not belonging to the UWAN. Our method is based on measuring the propagation delay of the underwater acoustic channel over multiple hops of the UWAN: this harvests the randomness in the UWAN topology and turns the slow sound propagation in water into an advantage against eavesdropping. Our key generation protocol includes a route discovery handshake, whereby all UWAN devices at intermediate hops accumulate their message processing delays. This enables Alice and Bob to compute the actual propagation delays along each route and to map such information to a sequence of bits. Finally, from these bit sequences, Alice and Bob obtain a secret key. We analyze the performance of the protocol theoretically and assess it via extensive simulations and field experiments.
Underwater acoustic communications (UWACs) are being adopted in several applications, and their security mechanisms require periodically renewed symmetric keys among legitimate devices. To this end, a secret key agreement (SKA) exploiting the UWAC context can be implemented. In this paper, we focus on one step of the source-based SKA protocol, i.e., the advantage distillation, when the legitimate users pre-process the channel measurement to increase the agreement while decreasing the attacker's knowledge about the key. We propose to learn the advantage distillation process by using a dataset of observed channels from the legitimate parties and the attacker, respectively. In turn, also the attacker uses such data to extract the key from its own underwater channel observations. The key extraction functions are derived through a computationally efficient data-driven adversarial strategy based on a metric related to the secret key rate that can be computed with a lower computational cost. The proposed strategy is tested considering experimental UWAC data, derived from a sea experiment. Results prove the quick convergence of the algorithm and the advantage of using the proposed techniques.
Physical-layer key generation exploits the randomness and unpredictability of wireless channels to enhance the security of wireless communications. However, factors such as asynchronous measurements, channel noise, and hardware impairments undermine the ideal reciprocity of channel state information (CSI) observed by legitimate users, thereby limiting key generation performance. To address this issue, this paper proposes a jointly optimized contrastive learning network (JOCLNet), which consists of a joint learning model and a contrastive learning model. The joint learning model, built upon a convolutional neural network (CNN) and a feedforward neural network (FNN), separates non-reciprocal components such as noise and measurement errors from the raw CSI of legitimate users, while extracting reciprocal components for key generation, thus improving robustness in noisy environments. To overcome the inability of existing deep-learning-based SKG schemes to resist passive eavesdropping attacks, the contrastive learning model introduces a contrastive loss on top of the joint learning model. This enables a dual objective: enhancing reciprocity between the processed CSI of legitimate users while ensuring that the processed data of an eavesdropper remains highly uncorrelated with that of legitimate parties. Furthermore, a complete key generation scheme is designed based on the proposed JOCLNet. Experimental results demonstrate that the proposed scheme achieves strong resistance to eavesdropping and robust adaptability to noise.
Physical layer encryption is a promising solution to provide transmission security for Internet-of-Things (IoT). To realize physical layer encryption, channel state information (CSI) as a random source to generate secret keys should be exploited. The reciprocity of wireless channel between legitimate communication nodes which is typically assumed for time division duplex (TDD) systems, is of critical importance to ensure the consistency in generated keys at two nodes. However, in practical TDD systems, due to the asynchronous measurement of CSI, the receiver noise and the the difference in hardware, etc., the reciprocity of channel measurements between legitimate parties is hard to be guaranteed, which makes physical layer key generation approach infeasible. To address this issue, we propose a wireless key generation scheme based on deep learning techniques. Firstly, a feature extraction network, which consists of two auto-encoders trained jointly, is developed to extract the common feature from the different yet correlated channel observations at two legitimate nodes, which will be used as the source to generate keys. Then a quantizer with an adaptive guard band width is devised to output a binary secret-key sequence from the feature vector. Numerical results show that the proposed scheme achieves remarkable improvement in terms of key disagreement rate (KDR) and key generation rate (KGR) compared existing counterparts in the literature.
Physical layer encryption is an emerging security paradigm which supplements the higher-layer encryption solutions. To realize physical layer encryption, channel measurement is typically utilized as a randomness source to generate secret keys. Due to non-perfect reciprocity of the channels and the inevitable channel estimation errors resulting from noises and interferences, the channel measurements at the legitimate transceivers are not the same, yielding a mismatch in the generated keys. To address this issue, a distributed physical layer key generation scheme based on deep learning is proposed in this paper. A channel state information (CSI) learning neural network (CLNet) based on the autoencoder is deployed at both the transmitter and the receiver side to refine the initial channel estimation results. The CLNet takes the least square (LS) channel estimations as input and outputs the channel estimations more similar to the real channel. To train this CLNet, a channel equalization module together with a pre-trained maximum-likelihood (ML) detection network is deployed, which is used to make a decision on the transmitted pilot signal. Since the pilot is known as a priori, the cross entropy loss between the known pilot and the recovered one can be calculated, which implicitly indicates the accuracy of the refined CSI. Further, the weighted sum of the aforementioned cross entropy loss and the mean square loss between the original and the refined CSI estimates is utilized for training, which not only improves the CSI estimation quality but also avoids the over-fitting effect. Extensive simulation results show that compared with benchmark schemes, the proposed scheme improves the key agreement rate and the key generation rate at the legitimate users. Moreover, our method ensures a higher key disagreement rate between the legitimate user and the eavesdropper as well, especially in the low signal-to-noise ratio (SNR) regime.