SECURE COMMUNICATION METHOD

Abstract

Disclosed is a method for a secure communication method having a secret key generation technique. The novelty of the proposed method stems from enhancing physical layer security (PHY) by using channel-adaptive keys, after manipulating a channel by introducing an artificial component into the channel. An adaptively designed artificial component is cascaded with the legitimate user’s channel. In an orthogonal frequency division multiplexing (OFDM) system, subcarriers corresponding to a channel gain higher than a threshold value are selected to extract the keys. Since the number of the selected subcarriers is adaptive, the length of the generated key sequences is changing adaptively as well. Thus, the channel reciprocity property in a time division duplexing (TDD) system is utilized.

Description

TECHNICAL FIELD

The present disclosure relates to a communication method for enhancing physical layer security. In this invention; a novel method for generating secret keys depending on a wireless channel is proposed.

BACKGROUND

The broadcast nature of wireless transmission causes vulnerability to the communication between legitimate transmitter and receiver in the presence of a passive eavesdropper. Since there is a great development in 5G and beyond wireless networks, classical encryption techniques might not be sufficient enough to ensure security. Techniques that generate keys using the characteristics of random channels should be developed in order to overcome the problem.

According to many studies including Ren K. et al (2011), Zeng K. (2015), Zhang J. et al. (2016) and Hamamreh J. M. et al. (2018); one of the innovative concept of channel-based security methods is to generate secret keys by exploiting the characteristics of the channel between the legitimate transmitter and receiver, while preventing eavesdroppers to access the secret data content. Badawy A. et al. (2016) states that channel-based key generation techniques are predominantly based on the assumption of channel reciprocity.

In time division duplexing (TDD) communication systems, the reciprocity property enables to generate highly correlated keys for legitimate parties because of observing the same link at the both ends of the channel. Moreover, Zhang J. et al. (2016) proposes that a key generation system should be evaluated by considering the performance metrics of randomness, key disagreement rate (KOR), and key generation rate (KGR). Further; Zhang J., Marshall A. et al. (2016) accepts randomness as a crucial characteristic for a generated key sequence.

A random key will provide a secure cryptographic system in the presence of passive eavesdropper attacks. KDR indicates the ratio of the different bits between the keys generated by each legitimate party to the length of the total key. In addition, KGR defines the quantity of key bits produced in each measurement.

For increasing KGR in secret key generation, Li G. et al. (2016) suggests using OFDM technique. Exploiting the randomness of the legitimate user’s channel is a rising technology to make key cracking troublesome, as proposed in Li G. et al. (2017). According to Li G. et al. (2017) and Patwari N. et al. (2010), the reason behind the randomness of the channel can be presented as the movement of the user and object movements in the communication environment or introducing artificial components into the channel.

Artificial components such as filters are performed to increase the security in wireless communication systems. For example, in Zhang J., Marshall A. et al. (2016), a low pass filter is used to reduce the KDR and increase the SNR range of the system since cross correlation of channel measurements are impacted by noise. Further, in Khiabani Y. S. et al. (2013) a first-order digital filter is performed to map the automatic repeat request (ARQ) transmission into destination set. On the other hand, in Wu Y. et al. (2018), m-windows filtering method is used to have low key generation consistency probability of a multi-bit adaptive quantization scheme. In m-windows filtering system, a plurality of the keys are exploited to generate a key bit at the same interval and preprocessing channel eigenvalues through the m-windows helps to increase the key agreement rate. In addition, in Patwari N. et al. (2010), a finite impulse response implementation, Farrow filter, is performed as a fractional interpolation filter. Furthermore, a low pass filter is used to improve the reciprocity of the channel by reducing the noise in Zhang J. et al. (2015). In this study, low pass filter is designed to suppress the high frequency components of the estimated channel so that the correlation between legitimate transmitter’s and receiver’s channel estimation is improved. The low pass filter not only increases channel reciprocity, but also decreases KDR and improves the achievement of key generation. In addition, in Fang S. et al. (2017) and Fang S. (2018), a delay-weight-sum module is used for channel manipulation and this is implemented by a finite impulse response (FIR) filter. The delayed copies of the original transmitted signal are summed as if sending the signal through a real multipath channel. However, some of the mentioned methods have some drawbacks such as complexity and non-adaptivity which will make it easy to brute force attack.

For future networks such as large-scale heterogeneous wireless network key management and distribution are big issues. There will be demand in future networks to support new wireless technologies such as ULLRC, remote surgery and Internet of Things (IoT). But problems with these devices which are used in these applications are as follows: power-limitation, delay-sensitive and processing-restricted which make cryptography-based techniques fail for such type of technologies. Future networks need to support various kind of services and scenarios with different security requirements. The encryption-based method cannot provide scenario specific security. For these reasons, providing secure communication without depending on the conventional cryptography-based security solutions which their key needs to be shared in legitimate nodes is crucial.

In the state of the art, the patent document numbered CN104901795 discloses a physical layer secret key extraction method based on channel characteristics. However, in this document, it is not disclosed that receiver shares a reference signal to transmitter before communication for channel estimation. Further, it is also not disclosed that peak points, which corresponds to subcarriers, from the frequency selective channel between transmitter and receiver should be selected in order to create a cascaded channel.

SUMMARY

The invention aims to provide a method for a secure communication method comprising a secret key generation technique. The novelty of our proposed method stems from enhancing physical layer security (PHY) by using channel-adaptive keys, after manipulating a channel by introducing an artificial component into the channel. An adaptively designed artificial component is cascaded with the legitimate user’s channel. In an orthogonal frequency division multiplexing (OFDM) system, for example, subcarriers corresponding to a channel gain higher than a threshold value might be selected to extract the keys. Since the number of the selected subcarriers is adaptive, the length of the generated key sequences is changing adaptively as well. Thus, we can utilize the channel reciprocity property in a time division duplexing (TDD) system.

Different from previous works, this invention discloses a novel method by introducing an artificial component to ensure that receiver’s newly created channel is highly frequency selective and its number of subcarriers corresponding to a channel gain higher than a threshold value is increased. Our method inherently extracts keys after cascading the channel between a transmitter and legitimate receiver with the newly designed artificial component. The artificial component consists of the copies of the legitimate user’s channel values corresponding to the selected subcarriers. In addition, it is important to mention that since the keys are dependent on the legitimate user’s cascaded channel, their length and values are changing adaptively depending on the number of the selected subcarriers. Therefore, eavesdropper cannot decode the data even if she is stronger than the legitimate user.

The proposed method can enhance physical layer security of wireless systems and it can provide secure communication without depending on the conventional cryptography-based security solutions which their key needs to be shared in legitimate nodes. This algorithm also can solve the problems in physical layer security such as key sharing in conventional methods. Plus, the proposed method is a promising way to solve some problems, such as co-located attacks and high temporal correlation, which are critical issues in secret key generation technique as Jiao L. et al. (2019) mentioned, when increasing the randomness of user’s channel.

To sum up; the proposed algorithm can provide more reliable key by extracting bits from strong subcarriers (subcarriers corresponding to a channel gain higher than a threshold value). In addition, the proposed algorithm can provide high level secure key by providing adaptive length of key. By having adaptive length of key, it is too hard for eavesdropper to find the key and somehow impossible to decode the data. Besides, the proposed method can ensure secure communication without depending on the conventional cryptography based secure solutions which in these methods their secret key needs to be shared in legitimate nodes which this sharing is critical in terms of how to share this key securely. Lastly, the proposed method is applicable to ensure flexible security.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 shows proposed system model for secure communication in presence of a passive eavesdropper.

FIG. 2 shows OFDM transceiver structure of the proposed method

FIG. 3 shows the structure of the proposed channel-based key generation technique

FIG. 4a shows frequency response graph of receiver’s channel

FIG. 4b shows frequency response graph of receiver’s newly created channel

FIG. 5 shows an illustration of reshaping the key for symbol encryption

BRIEF DESCRIPTION OF REFERENCE NUMBERS

• A. Transmitter
• B. Receiver
• E. Eavesdropper
• S_ref. Reference signal
• h_b. Cascaded channel
• h_e. Eavesdropper channel
• b_t. Transmitted bits
• b_r. Received bits
• M. Modulation
• DM. Demodulation
• DC. Decoding
• K. Key
• S/P. Serial to parallel
• P/S. Parallel to serial
• IFFT. Inverse fast Fourier transform
• FFT. Fast Fourier transform
• CP. Cyclic prefix
• CPR. Cyclic prefix removal
• f. Frequency
• G. Gain
• KS. Key stream
• B_b. Binary key
• C_b. Complex key
• EQ. Channel equalization

DETAILED DESCRIPTION

FIG. 1 shows proposed system model for secure communication in presence of a passive eavesdropper. Particularly, the purpose of a transmitter (A) is to send a secret data and communicate confidentially with a legitimate user, or receiver (B) in the presence of a passive eavesdropper (E). Eavesdropper’s (E) aim is to access the secret data content from the communication link between transmitter (A) and receiver (B) by its own observations of the signals it received. Besides, it is stronger than the receiver (B) by having multiple antennas, better signal processing skills, more power, hardware capabilities, processing capability, distributed reception capability, and offline processing. Its antennas are distributed where their locations are not known by the transmitter (A). It is assumed that the channels of both the receiver (B) and eavesdropper (E) are independent and uncorrelated of each other, i.e. eavesdropper (E) is located at least half wavelength away from the receiver (B). Furthermore, all received signals experience Rayleigh fading frequency selective channel. It is assumed that the transmitter (A) knows channel state information (CSI) of the receiver (B) by using the reciprocity property in a time division duplexing (TDD) system, but it doesn’t have any knowledge about eavesdropper’s (E) channel, since it is passive. Therefore, both channels, transmitter-to-receiver and receiver-to-transmitter, are assumed to be estimated as correlated and same with each other in TDD mode as proposed in Zhou X. et al. (2016).

In the proposed communication method, the receiver (B) sends a reference signal (S_ref) to transmitter (A) for channel estimation. Assuming N corresponds to total number of complex data symbols; the proposed method fundamentally comprises the steps of:

• a. Selecting the first m point out of M number of peak-points from the frequency selective channel between transmitter (A) and receiver (B), where in points correspond to subcarriers and where in M < N,
• b. Creating an artificial channel, F_b ∈ C^Nx1, by using the selected m points,
• c. Creating a new channel, H_b ∈ C^Nx1, by cascading receiver’s (B) channel, _Ab ∈ C^Nx1, with the artificial channel, F_b; as H_b = _Ab ⊙ F_b,
• d. Selecting peak points from the cascaded channel, H_b,
• e. Quantization of the selected subcarriers which their gains are corresponds to the peak points from cascaded channel by transmitter (A) and receiver (B) to generate a binary key (B_b),
• f. Converting the binary key (B_b) into a complex key (C_b),
• g. Reshaping the complex key (C_b) to the closest multiplication of N.

In a preferred embodiment, M number of peak-points are selected where channel gain (G) is above average channel gaina (G) of all the frequency (f) indices considered by the transmitter (A) to extract keys.

In another embodiment, if the length of last key block is less than N, key samples from the head are added as a suffix to reshape the complex key (C_b).

Yet in another embodiment, after reshaping the complex key (C_b), transmitted signal, x, is sent to the receiver (B) by applying cyclic prefix (CP) to the time-domain encrypted symbols as y_b = h_b ∗ x + n_b; where y_b is received signal at receiver (B) , h_b is the cascaded channel in time-domain, and n_b is the zero-mean complex additive white-Gaussian noise (AWGN) at the receiver’s (B) side.

AN EXAMPLE IMPLEMENTATION

In this example implementation, firstly the receiver (B) transmits a reference signal (S_ref) to the transmitter (A) for channel estimation. Thus, as an advantage of the channel reciprocity property in TDD mode, the downlink channel is obtained from its uplink as suggested in Goldsmith A. (2005).

The proposed OFDM transceiver structure of the proposed method is depicted in FIG. 2. Frequency-domain complex data symbols having the length of N is represented by:

$S=\left[S_1,S_2,\dots ,S_N\right]$

where S ∈ (ℂ^1xN. These symbols, obtained by using BPSK modulation, are going to be send by the transmitter (A) to the receiver (B) in the presence the eavesdropper (E).

To encrypt data, a secret key is used. Generation of this key at the transmitter is illustrated in FIG. 3. The frequency response of the channel experienced by the receiver (B) is denoted by _Ab ∈ ℂ^Nx1 and by the eavesdropper (E) is by A_e ∈ ℂ^Nx1 where N is the channel length of each. M selected subcarriers (M < N) out of N points are selected from the frequency response of the receiver’s (B) channel,

$A_b={\left[A_{b_1},A_{b_2},\dots ,A_{b_N}\right]}^T$

of length N. These M subcarriers correspond to the points where the channel gain is above the average gain of all the frequency indices are considered by the transmitter (A) to extract the secret keys. Both to increase the number of the subcarriers corresponding to a channel gain higher than a threshold value and ensure that the channel is more selective, an artificial channel is designed by using the selected M points. The values of _Ab at selected M frequency values are copied till the length of the artificial channel, F_b ∈ C^Nx1, equals to the length of the receiver’s (B) channel, _Ab ∈ ℂ^Nx1. A new channel for the receiver (B), H_b ∈ C^Nx1, is created by cascading the receiver’s (B) channel, _Ab ∈ ℂ^Nx1, with the artificial channel, F_b ∈ C^Nx1 as expressed:

$H_b=A_b\odot F_b$

The number of selected points corresponding to the frequency indices where the channel gain values are above the average gain of all values is shown in FIGS. 4a and 4b. As presented in FIG. 4a, the number of the selected subcarriers of the receiver’s (B) channel is very less to extract keys securely. Therefore, the transmitter (A) adds an artificial component to manipulate the channel. Thus, the number of the selected parts of the cascaded channel given in FIG. 4b is greater than the number of selected points in FIG. 4a. After selecting subcarriers of H_b, the receiver’s (B) enhanced channel is quantized by the transmitter (A) and the receiver (B) to construct a key. The cascaded channel gain measurements are equally divided into regions and each region is quantized into multi-bit quantization levels by the transmitter (A) and the receiver (B). Each of the highest peak points corresponds to a bit stream. Therefore, length of the generated key depends on the number of the selected peak points of the receiver’s (B) cascaded channel. Since our proposed encryption is based on symbol level, the binary key for the receiver (B), B_b ∈ {0,1}, is converted into a complex key, C_b, as i(2xB_b - 1). Since the length of the key is longer than the number of symbols, the key is reshaped to the closest multiplication of the total number of symbols, N.

As it is seen in FIG. 5, the key stream (KS), K_b, which has an adaptive length, is divided into multiple blocks and length of each block should be equal to the number of symbols, N. If the length of the last key block is less than N, the key samples from the head are added as a suffix. At transmitter’s (A) side, after reshaping the key, the encrypted symbols, E_b, can be written as the multiplication of each key block of with the symbols, S as:

$E_b=K_{b_i}\odot S^T$

where i = 1, ..., n and n is the number of key blocks. E_b is reshaped to obtain a vector of encrypted symbols of length (N×n)×1. The transmitted signal, x, of having an adaptive length is sent to the receiver (B), after applying cyclic prefix (CP) to the time-domain encrypted symbols to avoid inter symbol interference (ISI). The received signal at the receiver’s (B) side is defined as:

$y_b=h_b\ast x+n_b$

where h_b is the receiver’s (B) cascaded channel in time-domain, x is the transmitted signal, and n_b is the zero-mean complex additive white Gaussian noise (AWGN) at the receiver (B). Since the length of x is adaptive, the length of the received signal, y_b, is adaptive as well.

After removing cyclic prefix (CP) and then applying S/P conversion on the time-domain received signal, y_b, the receiver uses FFT on the resulted signal. A zero-forcing channel equalization process is performed to reduce the effects of noise from the channel for a better decoding. Thus, the received signal at the receiver’s (B) side after channel equalization process is found by element-wise division of the received signal and his channel is expressed as:

${\widehat{Y}}_b=Y_b\varnothing H_b$

Where H_b is the receiver’s (B) cascaded channel in frequency-domain and y_b is the frequency-domain received signal after S/P conversion shown in FIG. 2. After P/S conversion, the receiver (B) generates the decoded data by using the key stream (KS), K_b, that it extracted from its own channel and the decoded data is observed as:

${\widehat{X}}_b={\widehat{Y}}_b\varnothing K_b$

The eavesdropper has access to the transmitted signal, x, as well. As it has stronger skills and a more versatile receiver than the receiver (B), it follows the same steps with the receiver (B) as shown in FIG. 2 to generate its own keys from its channel. The eavesdropper (E) also designs an artificial component as the same way of the receiver (B) and cascades it with its channel, A_e, to obtain its cascaded channel, H_e. The signal it captures from the channel is defined as:

$y_e=h_e\ast x+n_e$

where h_e is the eavesdropper’s (E) cascaded channel in time-domain and n_e is the zero-mean complex AWGN at the eavesdropper (E). The received signal at the eavesdropper’s (E) side after channel equalization process is expressed as:

${\widehat{Y}}_e=Y_e\varnothing H_e$

where Y_e is the frequency-domain received signal after S/P conversion and H_e is the eavesdropper’s (E) cascaded channel. The eavesdropper (E) generates the decoded data by using the key, K_e, it extracted from its channel and the decoded data is expressed as:

${\widehat{X}}_e={\widehat{Y}}_e\varnothing K_e$

It is important to note that both the receiver (B) and the eavesdropper (E) follow the same steps as shown in FIG. 2 in order to decode the secret data they received from the transmitter (A). Since the transmitter (A) transmits the data using the keys she extracted from the receiver’s (B) channel, the receiver (B) can securely receive this signal in the presence of the eavesdropper (E). Although the eavesdropper (E) has multiple antennas and more skills than the receiver (B), it cannot decode the data correctly even if it extracts its own keys after manipulating its channel by adding an artificial component into it.

REFERENCES

K. Ren, H. Su, and Q. Wang, “Secret key generation exploiting channel characteristics in wireless communications,” IEEE Wireless Commun., vol. 18, no. 4, pp. 6-12, 2011.

K. Zeng, “Physical layer key generation in wireless networks: challenges and opportunities,” IEEE Commun. Mag., vol. 53, no. 6, pp. 33-39, 2015.

J. Zhang, T. Q. Duong, A. Marshall, and R. Woods, “Key generation from wireless channels: A review,” IEEE Access, vol. 4, pp. 614-626, 2016.

J. M. Hamamreh, H. M. Furqan, and H. Arslan, “Classifications and applications of physical layer security techniques for confidentiality: A comprehensive survey,” IEEE Commun. Surveys & Tut., 2018.

A. Badawy, T. Elfouly, T. Khattab, C.-F. Chiasserini, A. Mohamed, and D. Trinchero, “Robust secret key extraction from channel secondary random process,” Wireless Commun. Mob. Comp., vol. 16, no. 11, pp. 1389-1400, 2016.

J. Zhang, A. Marshall, R. Woods, and T. Q. Duong, “Efficient key generation by exploiting randomness from channel responses of individual OFDM subcarriers,” IEEE Trans. Commun., vol. 64, no. 6, pp. 2578-2588, 2016.

G. Li, A. Hu, L. Peng, and C. Sun, “The optimal preprocessing approach for secret key generation from OFDM channel measurements,” in IEEE Global Commun. Conf. (Globecom), 2016, pp. 1-6.

G. Li, A. Hu, J. Zhang, and B. Xiao, “Security analysis of a novel artificial randomness approach for fast key generation,” in IEEE Global Commun. Conf. (Globecom), 2017, pp. 1-6.

N. Patwari, J. Croft, S. Jana, and S. K. Kasera, “High-rate uncorrelated bit extraction for shared secret key generation from channel measurements,” IEEE Trans. Mob. Comp., vol. 9, no. 1, pp. 17-30, 2010.

Y. Wu, H. Xia, and C. Cheng, “Improved mult-bit adaptive quantization algorithm for physical layer security based on channel charscteristics,” in Int. Conf. Sys. Inf. (ICSAI), 2018, pp. 807-811.

Y. S. Khiabani and S. Wei, “ARQ-based symmetric-key generation over correlated erasure channels,” IEEE Trans. Inf. Forensics Sec., vol. 8, no. 7, pp. 1152-1161, 2013.

J. Zhang, R. Woods, A. Marshall, and T. Q. Duong, “An effective key generation system using improved channel reciprocity,” in IEEE Int. Conf. Acoust., Speech Sig. Proc. (ICASSP), 2015, pp. 1727-1731.

S. Fang, I. Markwood, and Y. Liu, “Manipulatable wireless key establishment,” in IEEE Conf. Commun. Net. Sec. (CNS), 2017, pp. 1-9.

S. Fang, “Channel camouflage and manipulation techniques in wireless networks,” PHD theses and dissertations, 2018, https://scholarcommons.usf.edu/etd/7284.

L. Jiao, N. Wang, P. Wang, A. Alipour-Fanid, J. Tang, and K. Zeng, “Physical layer key generation in 5G wireless networks,” IEEE Wireless Commun. Mag., 2019.

X. Zhou, L. Song, and Y. Zhang, Physical Layer Security in Wireless Communications. CRC Press, 2016.

A. Goldsmith, Wireless Communications. Cambridge University Press, 2005.

Claims

1. A secure communication method, wherein a receiver (B) sends a reference signal (Sref) to a transmitter (A) for channel estimation and wherein N corresponds to a total number of complex data symbols; the method comprising the steps of: a. Selecting the a first m point out of M number of peak-points from the a frequency selective channel between the transmitter (A) and the receiver (B), where in points correspond to subcarriers and where in M < N,b. Creating an artificial channel, Fb ∈ ℂNx1, by using the selected m points,c. Creating a new channel, Hb ∈ ℂNx1, by cascading receiver’s (B) channel, Ab ∈ ℂNx1, with the artificial channel, Fb; as Hb = Ab ⊙ Fb,d. Selecting peak points from the cascaded channel, Hb,e. Quantization of the selected subcarriers which their gains are corresponds to the peak points from cascaded channel by the transmitter (A) and the receiver (B) to construct a binary key (Bb),f. Converting the binary key (Bb) into a complex key (Cb),andg. Reshaping the complex key (Cb) to the closest multiplication of N.

2. The secure communication method according to claim 1, wherein M number of peak-points are selected where channel gain (G) is above average gain (G) of all the frequency (f) indices from cascaded channel, Hb, considered by the transmitter (A) to extract keys.

3. The secure communication method according to claim 1, wherein if the length of a last key block is less than N, key samples from the head are added as a suffix to reshape the complex key (Cb).

4. The secure communication method according to claim 1; wherein after reshaping the complex key (Cb), transmitted signal, x, is sent to the receiver (B) by applying cyclic prefix (CP) to the time-domain encrypted symbols as yb = hb ∗ x + nb; where yb is received signal at receiver (B), hb is the cascaded channel in time-domain, and nb is the zero-mean complex additive white-Gaussian noise (AWGN) at the receiver’s (B) side.