Secret keys can be generated from reciprocal channels to be used for shared secret key encryption. However, challenges arise in practical scenarios from non-reciprocal measurements of reciprocal channels due to changing channel conditions, hardware inaccuracies and estimation errors resulting in low key generation rate (KGR) and high key disagreement rates (KDR). To combat these practical issues, we propose KNEW Key Generation using NEural Networks from Wireless Channels, which extracts the implicit features of channel in a compressed form to derive keys with high agreement rate. Two Neural Networks (NNs) are trained simultaneously to map each other's channel estimates to a different domain, the latent space, which remains inaccessible to adversaries. The model also minimizes the distance between the latent spaces generated by two trusted pair of nodes, thus improving the KDR. Our simulated results demonstrate that the latent vectors of the legitimate parties are highly correlated yielding high KGR (≈ 64 bits per measurement) and low KDR (<0.05 in most cases). Our experiments with over-the-air signals show that the model can adapt to realistic channels and hardware inaccuracies, yielding over 32 bits of key per channel estimation without any mismatch.
Secret keys can be generated from reciprocal channels to be used for shared secret key encryption. However, challenges arise in practical scenarios from non-reciprocal measurements of reciprocal channels due to changing channel conditions, hardware inaccuracies and estimation errors resulting in low key generation rate (KGR) and high key disagreement rates (KDR). To combat these practical issues, we propose KNEW Key Generation using NEural Networks from Wireless Channels, which extracts the implicit features of channel in a compressed form to derive keys with high agreement rate. Two Neural Networks (NNs) are trained simultaneously to map each other's channel estimates to a different domain, the latent space, which remains inaccessible to adversaries. The model also minimizes the distance between the latent spaces generated by two trusted pair of nodes, thus improving the KDR. Our simulated results demonstrate that the latent vectors of the legitimate parties are highly correlated yielding high KGR (≈ 64 bits per measurement) and low KDR (<0.05 in most cases). Our experiments with over-the-air signals show that the model can adapt to realistic channels and hardware inaccuracies, yielding over 32 bits of key per channel estimation without any mismatch.
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.
The basis of generating secret key from the common wireless channel at two communication parties is reciprocity. However, due to non-simultaneous measurements and unavoidable estimation error, bias always exists and key disagreement occurs. In this paper, performance analysis and improvements for key generation exploiting imperfect channel statement information (CSI) is studied. Firstly, the closed-form expression for key disagreement rate (KDR) is derived to evaluate the key mismatch caused by time delay and estimation error. Secondly, the energy efficiency involving power allocation between key generation and private data transmission is presented. After that, a neural network based prediction (NNBP) algorithm is proposed to improve the key consistency. Simulation results verify the theoretical analysis and show that NNBP achieves significant improvements. And it's interesting to find that NNBP algorithm achieves a higher performance gain in low signal-to-noise ration (SNR) and high delay condition, which is different from other methods, like guard band based quantizer (GBBQ).
The basis of generating secret key from the common wireless channel at two communication parties is reciprocity. However, due to non-simultaneous measurements and unavoidable estimation error, bias always exists and key disagreement occurs. In this paper, performance analysis and improvements for key generation exploiting imperfect channel statement information (CSI) is studied. Firstly, the closed-form expression for key disagreement rate (KDR) is derived to evaluate the key mismatch caused by time delay and estimation error. Secondly, the energy efficiency involving power allocation between key generation and private data transmission is presented. After that, a neural network based prediction (NNBP) algorithm is proposed to improve the key consistency. Simulation results verify the theoretical analysis and show that NNBP achieves significant improvements. And it's interesting to find that NNBP algorithm achieves a higher performance gain in low signal-to-noise ration (SNR) and high delay condition, which is different from other methods, like guard band based quantizer (GBBQ).
Physical layer secret key generation (PSKG) lever-ages the reciprocal properties of wireless channels to establish cryptographic keys between legitimate users. However, in time-division duplex (TDD) systems, this reciprocity can be compromised by hardware imperfections and channel noise, complicating the key generation process. To address these challenges, we propose a shared cross-prediction neural network (SCP-NN) model that operates in an unsupervised manner, eliminating the need for perfect channel state information (CSI). Specifically, the proposed SCP-NN model employs a cross-prediction strategy that minimizes the difference between the predicted channels of legitimate users, ensuring that the outputs converge toward each other. Furthermore, we present a comprehensive PSKG procedure that leverages the SCP-NN architecture to extract key information from the phase characteristics of the channel frequency responses. Experimental results demonstrate that the proposed SCP-NN significantly reduces the key disagreement ratio (KDR), providing a robust and effective solution for secure key generation in dynamic IoT environments.
Wireless information security generates shared secret keys from reciprocal channel dynamics. Current solutions are mostly based on temporal per-frame channel measurements of signal strength and suffer from low key generate rate (KGR), large budget in channel probing, and poor secrecy if a channel does not temporally vary significantly. This paper designs a cross-layer solution that measures noise-free per-symbol channel dynamics across both time and frequency domain and derives keys from the highly fine-grained per-symbol reciprocal channel measurements. This solution consists of merits that: (1) the persymbol granularity improves the volume of available uncorrelated channel measurements by orders of magnitude over per-frame granularity in conventional solutions and so does KGR; 2) the solution exploits subtle channel fluctuations in frequency domain that does not force users to move to incur enough temporal variations as conventional solutions require; and (3) it measures noise-free channel response that suppresses key bit disagreement between trusted users. As a result, in every aspect, the proposed solution improves the security performance by orders of magnitude over conventional solutions. The performance has been evaluated on both a GNU SDR testbed in practice and a local GNU Radio simulator. The cross-layer solution can generate a KGR of 24.07 bits per probing frame on testbed or 19 bits in simulation, although conventional optimal solutions only has a KGR of at most one or two bit per probing frame. It also has a low key bit disagreement ratio while maintaining a high entropy rate. The derived keys show strong independence with correlation coefficients mostly less than 0.05. Furthermore, it is empirically shown that any slight physical change, e.g. a small rotation of antenna, results in fundamentally different cross-layer frequency measurements, which implies the strong secrecy and high efficiency of the proposed solution.
Vector quantization schemes are proposed to extract secret keys from correlated wireless fading channels. By assuming that the channel between two terminals are reciprocal, its estimates can be used as the common randomness for generating secret keys at the two terminals. Most schemes in the literature assume that channels are independent over time and utilize scalar quantization on each element of the estimated channel vector to generate secret key bits. These schemes are simple to implement but yield high key disagreement probability (KDP) at low SNR and low key entropy when channels are highly correlated. In this work, two vector quantization schemes, namely, the minimum key disagreement probability (MKDP) and the minimum quadratic distortion (MQD) secret key generation schemes, are proposed to effectively extract secret keys from correlated channel estimates. The vector quantizers are derived using KDP and QD as the respective distortion measures. To further reduce KDP, each channel vector is first pre-multiplied by an appropriately chosen unitary matrix to rotate the vector away from quantization cell boundaries. The MKDP scheme achieves the lowest KDP but requires high complexity whereas the MQD scheme yields lower complexity but at the cost of slightly increased KDP. Computer simulations are provided to demonstrate the effectiveness of the proposed vector quantization schemes.