Method for increasing physical layer security in OFDM based ISAC system

Abstract

The invention relates to a method realized by an Orthogonal Frequency Division Multiplexing (OFDM) based Integrated Sensing and Communication (ISAC) system comprising at least one base station (100) and at least one user equipment (201). Accordingly, its novelty is characterized in that comprising the steps of; performing channel probing between the base station (100) and the user equipment (201); if a channel mismatch detected, performing reconciliation between the base station (100) and the user equipment (201); determining a channel between the base station (100) and the user equipment (201); by the base station (100), accessing a pre-trained generative adversarial network (GAN) model which the GAN model is configured to generate adversarial channels from an inputted channel, wherein the GAN model is stored on both the base station (100) and the user equipment (201); inputting, by the base station (100), determined channel to the GAN model and adversarial channels from GAN model; generating sequences based on adversarial channels; arranging the order of OFDM symbols of a message to be transmitted according to generated sequences; transmitting, by the base station (100), arranged OFDM symbols to user equipment (201); receiving, by the user equipment (201), OFDM symbols; inputting, by the user equipment (201), determined channel to the GAN model and acquiring adversarial channels; generating sequences based on adversarial channels; rearranging OFDM symbols based on generated sequences for acquiring the message.

Description

[0001] DESCRIPTION

[0002] METHOD FOR INCREASING PHYSICAL LAYER SECURITY IN OFDM BASED ISAC SYSTEMS

[0003] TECHNICAL FIELD

[0004] The invention relates to a method realized by an Orthogonal Frequency Division Multiplexing (OFDM) based Integrated Sensing and Communication (ISAC) system comprising at least one base station and at least one user equipment.

[0005] PRIOR ART

[0006] Orthogonal Frequency Division Multiplexing (OFDM)-based Integrated Sensing and Communication (ISAC) is a system that uses OFDM waveforms to combine radar sensing and communication within the same framework. OFDM, a modulation technique common in wireless communication systems like LTE and Wi-Fi, is known for its ability to handle multipath fading and its efficiency in using available spectrum. By leveraging these characteristics, OFDM-based ISAC allows a single signal to simultaneously perform sensing and communication tasks.

[0007] The integration works by sharing the OFDM waveform between these two functions, eliminating the need for separate signals for sensing and communication. This approach is efficient because it reduces the complexity and cost of the system while maximizing the use of spectrum resources. OFDM’s ability to divide the available bandwidth into multiple subcarriers provides flexibility, making it possible to allocate resources to meet the specific needs of both sensing (such as precise range or velocity resolution) and communication (like high data rates or low latency). This makes OFDM-based ISAC suitable for applications like autonomous vehicles, smart cities, and next-generation wireless networks, where sensing and communication often need to coexist.

[0008] The integration of multiple functionalities within a single system has gained significant attention, this aims to share resources among services or functionalities in a more efficient way, or to enhance the capabilities of one or more services through synergy. In this regard, extensive research is being carried out on OFDM schemes for integrated sensing and communication (ISAC). This is because OFDM has already been widely adopted in various wireless communication systems (i.e. 4G, 5G and Wi-Fi standards from Wi-Fi 4) due to its high spectrum efficiency and resilience to multipath spreading, while there is a need to integrate both sensing and communication functionalities over a single waveform with small modification of the hardware structure and signaling strategies [1].

[0009] In conventional OFDM-based communication, OFDM pilots are commonly used for channel estimation. Currently, pilot signals are being leveraged for cost-effective ranging, sensing, or self-localization [2], By utilizing the pilots for both sensing and channel estimation for conventional communication, the need to generate new carrier signals for sensing is thus being avoided. Despite the apparent benefits of resource sharing, integrating sensing and communication can introduce significant information leakage and privacy risks, especially when the sensing and communication users are two different entities. For example, with knowledge of the pilot signals, a legitimate sensing user has the ability to equalize the communication signal, making him a potential eavesdropper against the conventional communication. This poses a critical security risk, necessitating new or enhanced security mechanisms for ISAC systems, particularly OFDM-based ISAC systems.

[0010] Traditionally, encryption schemes have been employed to secure and maintain the confidentiality of communications by using pre-shared keys to encrypt and decrypt data through specific methods. These include symmetric-key cryptography, which uses a single key for both encryption and decryption (e.g., Advanced Encryption Standard (AES) and Data Encryption Standard (DES), BlowFish, Ronald Rivest (RC) ); public-key cryptography, which relies on a pair of keys — a public key for encryption and a private key for decryption (e.g., Rivest-Shamir-Adleman (RSA) and Diffie-Hellman) [3]-[5]. There are also Hash functions such as Secure Hashing Algorithm (SHA), RACE Integrity Primitives Evaluation Message Digest (RIPEMD).and Message Digest Algorithm (MD), that transform input messages of any length into fixed-length hash values, often referred to as message digests [3]. Despite the extensive use of crypto-based algorithms over the past decades, these techniques lack high key dynamicity, making them unsuitable for long-term communication, as the same key is often reused for extended periods. This vulnerability becomes particularly critical in the face of computationally strong attackers as of the current era. In large-scale applications with a vast number of devices, generating multiple keys for each device poses additional challenges and strains existing key generation techniques. Furthermore, the broadcast nature of wireless systems allows any entity to capture and analyze wireless signals anonymously, increasing the risk of key compromise and potential information leakage. Other constraints on traditional encryption techniques include the complexity and latency introduced by mathematical encryption functions, which depend on file size, type, and platform [3]. This worsens performance for low-latency communication and heavy data applications such as augmented reality, real-time gaming, and more.

[0011] Physical Layer Security: Traditional cryptography techniques have primarily focused on achieving information secrecy at higher layers, generally overlooking the physical or first layer [6]. In wireless communication, however, it has been discovered that the physical characteristics of devices, waveforms, and the uniqueness and random nature of the channel between transmitter (TX) and receiver (RX), can be leveraged to implement reliable, strong, low-cost, and less complex security measures [7], This has paved the way for a new concept in information security, known as Physical Layer Security (PLS), which adds a new layer of safeguards against eavesdropping, spoofing, impersonation, and other attacks.

[0012] Particularly, the wireless channel and its properties have been widely utilized in PLS for link adaptation to nullify the signal or significantly degrade the signal-to-noise ratio (SNR) at the eavesdropper [7], In this scenario, the eavesdropper is unable to achieve sufficient performance to reconstruct the signal. To address the issue of key dynamicity, which is often lacking in traditional cryptographic techniques, the unique properties of the channel, including randomness and reciprocity between the transmitter (TX) and receiver (RX), have been exploited to generate reciprocal keys (identical keys) that are dynamic and unique based on the observed channels. This approach, known as channel-based key generation [7], eliminates the need for a new key exchange system by leveraging the channel reciprocity between legitimate TX and RX. Furthermore, other studies have developed methods for interference or noisy signal injection to hinder the eavesdropper's performance [7] [8].

[0013] In OFDM-based communication systems, such as 4G, 5G, and Wi-Fi standards from Wi-Fi 4 onwards, security against eavesdropping has been extensively discussed, and promising techniques tailored to the OFDM waveform have been proposed. Techniques such as cyclic prefix (CP) shortening, randomized cyclic prefix, randomized pilots, intentional introduction of interference, and joint time-frequency domain artificial noise (AN) have been proposed to complicate the eavesdropping equalization capability.

[0014] When OFDM pilots are used for both sensing and channel estimation, the need to generate new carrier signals for sensing is avoided. However, this approach also makes the communication signal vulnerable to potential eavesdropping by legitimate sensing users, as legitimate sensing users with knowledge of the pilot signals could equalize the communication signal. Thus, the communication signal should also be protected from sensing users. In other words, the existing PLS techniques tailored to OFDM-based communication systems may fail to ensure secure coexistence between communication and sensing

[0015] Randomized CP or channel shortening PLS technique: the CP is adjusted based on an agreement between the transmitter (TX) and the conventional receiver (RX) or the maximum excess delay of their channel. Since this CP is not designed based on the maximum excess delay of a potential eavesdropper (Eve), it may be too short for an illegitimate user (Eve), leading to higher inter-carrier interference (ICI) and inter-symbol interference (ISI) [9]

[0010]

[0011] . This implies that a legitimate sensing user, who is now assumed to be illegitimate for communication, will experience high ICI and ISI in the OFDM pilot signals, resulting in failed or reduced sensing performance.

[0016] Pilot Randomization: This technique involves changing the location, sequence, or pattern of pilot tones across different OFDM symbols, making it more difficult for an attacker (EVE) to identify and extract pilots for channel estimation. As a result, the legitimate sensing user cannot also identify the pilots, preventing him from performing sensing

[0012]

[0013] .

[0017] Intentional introduction of interference: Intrinsic interference is created, following a pattern known to the legitimate receiver, who is assumed to have the capability to cancel this interference. The intrinsic interferences are broadly distributed and overlap across the timefrequency grid, making it highly challenging for the eavesdropper to fully eliminate without knowing the loading pattern. The intrinsic interference can also affect pilot carriers significantly, degrading the performance for sensing users in in ISAC systems

[0014] ,

[0018] Artificial Noise (AN): AN is transmitted along with OFDM signals and can be effectively removed at the legitimate receiver as it is generated based on the receiver's channel state information (CSI) [6]

[0015]

[0016] . However, for a sensing user, the AN, which may be distributed across OFDM pilot subcarriers, can significantly bias the sensing process, potentially leading to inaccurate sensing decisions. When frequency-domain artificial noise (AN) is applied to specific data carriers, it may not significantly improve the secrecy rate in single-antenna wiretap channels or devices with a limited number of antennas, as it is more suitable for systems with a large number of antennas at the TX and RX. Considering the above discussion, it can be concluded that securing OFDM symbols based on the communication user's channel characteristics can hinder sensing capabilities in ISAC systems. Interestingly, studies in

[0017] and

[0018] prosed FFT-based speech scrambling techniques aiming at minimizing residual intelligibility in speech transmission systems, addressing the limitations of previous methods that leave identifiable patterns exploitable by eavesdroppers. In

[0017] , a scrambling algorithm that it is based on the Hadamard transform in the frequency domain is proposed, offering a larger key space, low residual intelligibility, and fairly good recovered speech quality. The work in

[0018] combines Quadrature Amplitude Modulation (QAM) mapping with Orthogonal Frequency Division Multiplexing (OFDM) to scramble speech by permuting various frequency components, resulting in scrambled speech with no residual intelligibility. Additionally, it allows multiple unique scrambling sequences to be generated from the speech. However, a limitation of these method is the lack of a reliable process for regenerating scrambling keys at both ends, as the receiver does not have prior knowledge of the speech content, making descrambling challenging unless the keys are pre-shared.

[0019] All the problems mentioned above have made it necessary to make an innovation in the relevant technical field as a result.

[0020] 1. S. Mura, D. Tagliaferri, M. Mizmizi, U. Spagnolini and A. Petropulu, "Waveform Design for OFDM-Based ISAC Systems Under Resource Occupancy Constraint," 2024 IEEE Radar Conference (RadarConf24), Denver, CO, USA, 2024, pp. 1-6, doi: 10.1109 / RadarConf2458775.2024.10548861 .

[0021] 2. Y. Wan, Z. Hu, A. Liu, R. Du, T. X. Han and T. Q. S. Quek, "OFDM-Based Multiband Sensing For ISAC: Resolution Limit, Algorithm Design, and Open Issues," in IEEE Vehicular Technology Magazine, vol. 19, no. 2, pp. 51-59, June 2024, doi: 10.1109 / MVT.2024.3368205. All the problems mentioned above have made it necessary to make an innovation in the relevant technical field as a result.

[0022] 3. Alenezi, M. N., Alabdulrazzaq, H., & Mohammad, N. Q. (2020). Symmetric encryption algorithms: Review and evaluation study. International Journal of Communication Networks and Information Security, 12(2), 256-272.

[0023] 4. C. S. KLINE and G. J. POPEK, "Public key vs. conventional key encryption," 1979 International Workshop on Managing Requirements Knowledge (MARK), New York, NY, USA, 1979, pp. 831-838, doi: 10.1109 / MARK.1979.8817073.

[0024] 5. Hellman, M. E. (2012). An overview of public key cryptography. IEEE Communications Magazine, 40(5), 42-49.

[0025] 6. Q. Xu, P. Ren, Q. Du and L. Sun, "Security-Aware Waveform and Artificial Noise Design for Time-Reversal-Based Transmission," in IEEE Transactions on Vehicular Technology, vol. 67, no. 6, pp. 5486-5490, June 2018, doi: 10.1109 / TVT.2018.2813318. 7. M. S. J. Solaija, H. Salman and H. Arslan, "Towards a Unified Framework for Physical Layer Security in 5G and Beyond Networks," in IEEE Open Journal of Vehicular Technology, vol. 3, pp. 321-343, 2022, doi: 10.1109 / OJVT.2022.3183218.

[0026] 8. Q. Xu, P. Ren, Q. Du and L. Sun, "Security-Aware Waveform and Artificial Noise Design for Time-Reversal-Based Transmission," in IEEE Transactions on Vehicular Technology, vol.

[0027] 67, no. 6, pp. 5486-5490, June 2018, doi: 10.1109 / TVT.2018.2813318.

[0028] 9. Solaija, M. S. J., Salman, H., & Arslan, H. (2021 ). Enhancing Channel Shortening Based Physical Layer Security Using Coordinated Multipoint. arXiv preprint arXiv:2109.14346.

[0029] 10. H. M. Furqan, J. M. Hamamreh and H. Arslan, "Enhancing physical layer security of OFDM systems using channel shortening," 2017 IEEE 28th Annual International Symposium on Personal, Indoor, and Mobile Radio Communications (PIMRC), Montreal, QC, Canada, 2017, pp. 1-5, doi: 10.1109 / PIMRC.2017.8292335.

[0030] 11. M. Bouanen, F. Gagnon, G. Kaddoum, D. Couillard and C. Thibeault, "An LPI design for secure OFDM systems," MILCOM 2012 - 2012 IEEE Military Communications Conference, Orlando, FL, USA, 2012, pp. 1-6, doi: 10.1109 / MILCOM.2012.6415833.

[0031] 12. H. Wei, B. Zheng and X. Hou, "Compressive channel sensing based on random pilot for physical layer communication security," 2013 22nd Wireless and Optical Communication Conference, Chongqing, China, 2013, pp. 693-698, doi: 10.1109 / WOCC.2013.6676463.

[0032] 13. M. Soltani, T. Bayka§ and H. Arslan, "Achieving secure communication through pilot manipulation," 2015 IEEE 26th Annual International Symposium on Personal, Indoor, and Mobile Radio Communications (PIMRC), Hong Kong, China, 2015, pp. 527-531 , doi: 10.1109 / PIMRC.2015.7343356.

[0033] 14. M. Sakai, H. Lin and K. Yamashita, "Intrinsic Interference Based Physical Layer Encryption for OFDM / OQAM," in IEEE Communications Letters, vol. 21 , no. 5, pp. 1059-1062, May 2017, doi: 10.1109 / LCOMM.2017.2654442.

[0034] 15. H. Qin et aL, "Power Allocation and Time-Domain Artificial Noise Design for Wiretap OFDM with Discrete Inputs," in IEEE Transactions on Wireless Communications, vol. 12, no.

[0035] 6, pp. 2717-2729, June 2013, doi: 10.1109 / TCQMM.2013.050713.120730.

[0036] 16. T. Akitaya, S. Asano and T. Saba, "Time-domain artificial noise generation technique using time-domain and frequency-domain processing for physical layer security in MIMO- OFDM systems," 2014 IEEE International Conference on Communications Workshops (ICC), Sydney, NSW, Australia, 2014, pp. 807-812, doi: 10.1109 / ICCW.2014.6881299.

[0037] 17. Yuan Wu and Boon Poh Ng, "Speech scrambling with Hadamard transform in frequency domain," 6th International Conference on Signal Processing, 2012., Beijing, China, 2012, pp. 1560-1563 vol.2, doi: 10.1109 / ICQSP.2012.1180094. 18. D. C. Tseng and J. H. Chiu, "An OFDM speech scrambler without residual intelligibility," TENCON 2017 - 2017 IEEE Region 10 Conference, Taipei, Taiwan, 2017, pp. 1-4, doi: 10.1109 / TENCON.2017.4428903.

[0038] BRIEF DESCRIPTION OF THE INVENTION

[0039] The present invention relates to a method to eliminate the above-mentioned disadvantages and bring new advantages to the relevant technical field.

[0040] An object of the invention is to prevent internal eavesdroppers and external eaves droppers from leaking data of a transmission from a base station to a target user in Orthogonal Frequency Division Multiplexing (OFDM)-based Integrated Sensing and Communication (ISAC) systems while maintaining target user’s receiving performance and sensing user’s sensing performance.

[0041] Another object of the invention is to provide protection from eavesdropping even in the low varying channel conditions.

[0042] To achieve all the objects mentioned above and that will emerge from the following detailed description, the present invention relates to a method realized by an Orthogonal Frequency Division Multiplexing (OFDM) based Integrated Sensing and Communication (ISAC) system comprising at least one base station and at least one user equipment. Accordingly, comprising the steps of; performing channel probing between the base station and the user equipment; if a channel mismatch detected, performing reconciliation between the base station and the user equipment; determining a channel between the base station and the user equipment; by the base station, accessing to a pre-trained generative adversarial network (GAN) model which the GAN model is configured to generate adversarial channels from an inputted channel, wherein the GAN model is stored on both the base station and the user equipment; inputting, by the base station, determined channel to the GAN model and adversarial channels from GAN model; generating sequences based on adversarial channels; arranging order of OFDM symbols of a message to be transmitted according to generated sequences; transmitting, by the base station, arranged OFDM symbols to user equipment; receiving, by the user equipment, OFDM symbols; inputting, by the user equipment, determined channel to the GAN model and acquiring adversarial channels; generating sequences based on adversarial channels; rearranging OFDM symbols based on generated sequences for acquiring the message. Thus, enabling the generation of high-entropy channels from low-entropy channels, which is crucial for systems that rely on channel-based physical layer security.

[0043] A possible embodiment of the invention is characterized in that updating, by the base station, at least one weight of the GAN model based on determined channel before inputting the channel to the GAN model; updating, by the user equipment, at least one weight of the GAN model based on determined channel before inputting the channel to the GAN model. This further increases the entropy.

[0044] BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 is a drawing illustrating top schematic view of the system.

[0046] Figure 2 is a drawing illustrating the flow chart of the method.

[0047] Figure 3 is a drawing illustrating components of a base station and user equipment in the system.

[0048] REFERENCE NUMBERS GIVEN IN THE FIGURE

[0049] 100 Base station

[0050] 201 User equipment

[0051] 202 Sensing user equipment

[0052] 203 Eavesdropper user equipment

[0053] 211 Parallel to serial conversion unit

[0054] 212 Serial to parallel conversion unit

[0055] 213 FFT unit

[0056] 214 IFFT unit

[0057] 215 Demodulator

[0058] 216 Modulator

[0059] 217 Scrambler

[0060] 218 De-scrambler

[0061] 219 CP adder

[0062] 220 CP remover

[0063] 221 Up conversion unit 222 Down conversion unit

[0064] DETAILED DESCRIPTION OF THE INVENTION

[0065] In this detailed description, the subject matter is explained with references to examples without forming any restrictive effect only in order to make the subject more understandable.

[0066] Invention is a method for increasing physical layer security of Orthogonal Frequency Division Multiplexing (OFDM) based Integrated Sensing and Communication (ISAC) system.

[0067] Referring to figure 1 , the system comprises at least a base station (100) and at least a user equipment (201 ) capable of communicate using OFDM. The method aims to prevent sensing user equipment (202) or eavesdropping user equipment (203) from eavesdropping to data sent to the user equipment (201 ). Rather than using the probed channel directly as in conventional physical layer security method, this is realized by re-arranging OFDM symbols utilizing Generative Adversarial Network (GAN) model which uses channel between base station (100) and user equipment (201 ) to generate plurality of high entropy adversarial channels sequences.

[0068] Subject matter method is realized by the base station (100) and the user equipment (201 ) of the ISAC system. Referring to figure 2 the method comprises the following steps:

[0069] Base station (100) and user equipment (201 ) perform channel probing. The user equipment (201 ) and base station (100) observe a similar channel. However, due to timing misalignment, slight channel discrepancies may arise, resulting in a degree of channel mismatch.

[0070] Base station (100) and user equipment (201 ) perform reconciliation if a channel mismatch is detected. Reconciliation is well known in the art and refers to the process of aligning or resolving any discrepancies in the communication or operational parameters between the two entities in a wireless network.

[0071] Base station (100) and user equipment (201 ) determine a channel between them.

[0072] Base station (100) accesses a pre-trained generative adversarial network (GAN) model, which is configured to generate adversarial channels from an inputted channel. The GAN model is stored on both the base station (100) and the user equipment (201 ). Thus, both use it to generate the same adversarial channels to use them in securing the communication.

[0073] Base station (100) inputs the determined channel to the GAN model and generates sequences based on adversarial channels from the GAN model.

[0074] Base station (100) arranges order of OFDM symbols of a message to be sent according to generated sequences. Then the base station (100) transmits arranged OFDM symbols to user equipment (201 ).

[0075] The user equipment (201 ) receives OFDM symbols. It inputs the determined channel to the GAN model and acquires adversarial channels. The user equipment (201 ) then generates sequences based on adversarial channels. The user equipment (201 ) rearranges OFDM symbols based on the generated sequences for acquiring the message. Thus, eavesdropping user equipment (203) and sensing user equipment (202) are unable to equalize and leak the data from the transmission.

[0076] In a second embodiment, the base station (100) and user equipment (201 ) update the GAN model’s weights based on the determined channel. This allows them to have a unique GAN model for communication, which cannot easily be replicated by eavesdroppers.

[0077] GAN models are well known in the art. Generative Adversarial Networks (GANs) are a class of deep neural networks designed to generate synthetic data or augment existing datasets. They consist of two competing neural networks: a Generator and a Discriminator [Z. Pan, W. Yu, X. Yi, A. Khan, F. Yuan and Y. Zheng, "Recent Progress on Generative Adversarial Networks (GANs): A Survey," in IEEE Access, vol. 7, pp. 36322-36333, 2019, doi: 10.1109 / ACCESS.2019.2905015.], [A. Creswell, T. White, V. Dumoulin, K. Arulkumaran, B. Sengupta and A. A. Bharath, "Generative Adversarial Networks: An Overview," in IEEE Signal Processing Magazine, vol. 35, no. 1 , pp. 53-65, Jan. 2018, doi: 10.1109 / MSP.2017.2765202.]. The Generator's role is to create new data samples, while the Discriminator evaluates these samples, determining whether they are 'real' (from the original training data) or 'fake' (generated by the Generator).

[0078] Once the Generator is adequately trained to fool the Discriminator, a GAN can produce realistic synthetic data. A key characteristic of GANs is that they are deterministic in nature; once trained, the weights and biases of the neural network layers are fixed. This means that providing the same input to a trained GAN will consistently yield the same output. Multiple GANs can be trained simultaneously to produce various synthetic outputs from a single input, allowing for diverse data generation. Thus, a base station (100) and a user equipment (201 ) receive the same adversarial channel outputs when they input the same channel to the GAN.

[0079] GAN models are pre-trained. The network may utilize the channel sets that it processes from different users' activities to train GANs. In simulations, the channel dataset can be constructed using specific channel models or distributions corresponding to the environment's propagation characteristics and richness.

[0080] In the training process of GANs, the available dataset is split into multiple training subsets. The number of training subsets corresponds to the number of adversarial channels that are needed. To guarantee adversarial channel richness and diversification, ensuring that GANs produce different high-entropy adversarial channels from the same input low-entropy channel, different filters extracting different features from the input channel need to be specified and implemented.

[0081] Since GANs are deterministic neural networks, once trained, the weights and biases of the neural network layers are fixed. The network may keep trained weights and share them with user equipment (201 ) and base stations (100). Considering that different users will observe different channels, all users may have the same weights without the need for retraining the GANs based on specific user channel observation history. In critical applications, it may be required for users to have different weights such that even if the channels are similar or correlated at a time instant, having different weights will result in different adversarial channels.

[0082] To this end, instead of retraining the whole GAN structure with the limited amount of user channel history, the weights obtained from the general training can be updated based on the user's dataset. The user channel dataset used to update the weights should be exactly the same, which means they are extracted from different reconciliation processes between the network and the corresponding user. Assuming that the network and the user have the reconciled channel dataset, the network and the specific user update the GANs toward the same weights, as in the second embodiment.

[0083] Since only generator’s weights are transferred and updated, the storage space needed in each user equipment (201 ) and base station (100) is significantly reduced.

[0084] In figure 1 , pilots are used for two purposes: sensing and channel estimation during conventional communication or data transmission. Base station (100) exchanges data with user equipment (201 ) over a wireless channel using OFDM. The pilots, used for channel estimation, are also accessible to the sensing user equipment (202), who is a legitimate sensing / localization user or device. To ensure proper sensing / localization functionality, the base station (100) cannot hide the pilots, preventing the base station (100) from hindering sensing user equipment (202) equalization and reconstruction of transmitted data. Any user who receives the signal transmitted by base station (100) must perform descrambling to extract the data. The descrambling sequence corresponds to the scrambling sequence but is used in reverse operation. In this scenario, since the base station (100) and user equipment (201 ) have observed the same channel during the probing and reconciliation process, both will generate the same sequences using the proposed methods. Base station (100) will scramble the data while the user equipment (201 ) will descramble it. However, the sensing user equipment (202) will also attempt to generate the sequence, as it has the GANs’ weights and can implement the proposed methods. However, because it observes a different channel, its sequences will differ from those of the base station (100) and user equipment (201 ), preventing it from properly descrambling the data unless it breaks the sequences. Since different OFDM symbols are scrambled with different sequences, the sensing user equipment (202) must break through multiple sequences, complicating this data reconstruction task.

[0085] Compared to the sensing user equipment (202), eavesdropper user equipment (203) is not recognized by the system and is not authorized for both communication and sensing functionalities. The scrambling mechanism protects communication data from this eavesdropper user equipment (203). However, eavesdropper user equipment (203) can still perform sensing since they are aware of the pilots and their locations.

[0086] In this case, two scrambling sequence generation processes can be employed: one for the data and another for the pilots. The second process is mainly to conceal pilot signals from unauthorized users within the network. In this, the network (base station (100)) shares segments of individual channels among users, ensuring that all authorized users have access to the same information. This aligns with the concept of group key generation in Physical Layer Security (PLS). The shared channel information can then be utilized to generate pilot scrambling sequences. As a result, all users with access to the common channel information can leverage the proposed Generative Adversarial Networks (GANs)-assisted methods to generate identical pilot positions. All the scrambling sequence generation processes and usage presented in the previous section on 'Securing Communication Data' remain the same. The difference is that we need to input common information to generate common pilot scrambling sequences. Since external eavesdroppers lack this common information, they will be unable to generate the pilot position sequences even if they have deployed GANs.

[0087] Referring to figure 3 an exemplary base station (100) may comprise a serial-to-parallel conversion unit (212) for converting incoming serial data streams into parallel data streams for processing, a modulator (216) for mapping data onto specific modulation schemes to encode information, a scrambler (217) rearranging OFDM symbols based on determined sequences, an IFFT (Inverse Fast Fourier Transform) unit (214) for transforming frequency-domain data into time-domain signals to generate OFDM waveforms, a parallel-to-serial conversion unit (211 ) for converting the parallel processed data back into a serial format for transmission, a CP (Cyclic Prefix) adder (219) for appending a cyclic prefix to mitigate inter-symbol interference (ISI) and ensure robust communication, and an up-conversion unit (221 ) for converting the baseband signal to a higher frequency suitable for transmission over the wireless medium.

[0088] An exemplary user equipment (201 ) may comprise a parallel-to-serial conversion unit (211 ) for converting incoming parallel data streams into a serial format for further processing, a demodulator (215) for extracting encoded information from the received signal by reversing the modulation process, a de-scrambler (218) for restoring the original data patterns by reversing the scrambling operation using determined sequences, an FFT (Fast Fourier Transform) unit (213) for converting time-domain signals into frequency-domain data to facilitate OFDM symbol detection, a serial-to-parallel conversion unit (212) for converting the processed serial data into parallel streams for additional processing, a CP (Cyclic Prefix) remover (220) for eliminating the cyclic prefix from the received signal to restore the original data frame, and a down-conversion unit (222) for translating the high-frequency received signal to a lower frequency suitable for baseband processing.

[0089] The scope of protection of the invention is specified in the attached claims and cannot be limited to those explained for sampling purposes in this detailed description. It is evident that a person skilled in the art may exhibit similar embodiments in light of the above-mentioned facts without drifting apart from the main theme of the invention.

Claims

CLAIMS1. A method realized by an Orthogonal Frequency Division Multiplexing (OFDM) based Integrated Sensing and Communication (ISAC) system comprising at least one base station (100) and at least one user equipment (201 ) characterized in that comprising the steps of; performing channel probing between the base station (100) and the user equipment (201 ); if a channel mismatch detected, performing reconciliation between the base station (100) and the user equipment (201 ); determining a channel between the base station (100) and the user equipment (201 ); by the base station, accessing to a pre-trained generative adversarial network (GAN) model which the GAN model is configured to generate adversarial channels from an inputted channel, wherein the GAN model is stored on both the base station (100) and the user equipment (201 ); inputting, by the base station, determined channel to the GAN model and adversarial channels from GAN model; generating sequences based on adversarial channels; arranging order of OFDM symbols of a message to be transmitted according to generated sequences; transmitting, by the base station, arranged OFDM symbols to user equipment (201 ); receiving, by the user equipment (201 ), OFDM symbols; inputting, by the user equipment (201 ), determined channel to the GAN model and acquiring adversarial channels; generating sequences based on adversarial channels; rearranging OFDM symbols based on generated sequences for acquiring the message.

2. The method according to claim 1 , characterized in that updating, by the base station, at least one weight of the GAN model based on determined channel before inputting the channel to the GAN model; updating, by the user equipment (201 ), at least one weight of the GAN model based on determined channel before inputting the channel to the GAN model.

Images