Biblio

Information privacy and security has become a necessity in the rapid growth of computer technology. A new algorithm for image encryption is proposed in this paper; using hash functions, chaotic map and two levels of diffusion process. The initialization key for chaos map is generated with the help of two hash functions. The initial seed for these hash functions is the sum of rows, columns and pixels across the diagonal of the plain image. Firstly, the image is scrambled using quantization unit. In the first level of diffusion process, the pixel values of the scrambled image are XOR with the normalized chaotic map. Odd pixel value is XOR with an even bit of chaotic map and even pixel is XOR with an odd bit of chaotic map. To achieve strong encryption, the image undergoes a second level of diffusion process where it is XOR with the map a finite number of times. After every round, the pixel array is circular shifted three times to achieve a strong encrypted image. The experimental and comparative analysis done with state of the art techniques on the proposed image encryption algorithm shows that it is strong enough to resist statistical and differential attacks present in the communication channel.

We investigate the fuzzy adaptive compensation control problem for nonlinear cyber-physical system with false data injection attack over digital communication links. The fuzzy logic system is first introduced to approximate uncertain nonlinear functions. And the time-varying sliding mode surface is designed. Secondly, for the actual require-ment of data transmission, three uniform quantizers are designed to quantify system state and sliding mode surface and control input signal, respectively. Then, the adaptive fuzzy laws are designed, which can effectively compensate for FDI attack and the quantization errors. Furthermore, the system stability and the reachability of sliding surface are strictly guaranteed by using adaptive fuzzy laws. Finally, we use an example to verify the effectiveness of the method.

We present a ternary source coding scheme in this paper, which is a special class of low density generator matrix (LDGM) codes. We prove that a ternary linear block LDGM code, whose generator matrix is randomly generated with each element independent and identically distributed, is universal for source coding in terms of the symbol-error rate (SER). To circumvent the high-complex maximum likelihood decoding, we introduce a special class of convolutional LDGM codes, called block Markov superposition transmission of repetition (BMST-R) codes, which are iteratively decodable by a sliding window algorithm. Then the presented BMST-R codes are applied to construct a tandem scheme for Gaussian source compression, where a dead-zone quantizer is introduced before the ternary source coding. The main advantages of this scheme are its universality and flexibility. The dead-zone quantizer can choose a proper quantization level according to the distortion requirement, while the LDGM codes can adapt the code rate to approach the entropy of the quantized sequence. Numerical results show that the proposed scheme performs well for ternary sources over a wide range of code rates and that the distortion introduced by quantization dominates provided that the code rate is slightly greater than the discrete entropy.

The cutting-edge biometric recognition systems extract distinctive feature vectors of biometric samples using deep neural networks to measure the amount of (dis-)similarity between two biometric samples. Studies have shown that personal information (e.g., health condition, ethnicity, etc.) can be inferred, and biometric samples can be reconstructed from those feature vectors, making their protection an urgent necessity. State-of-the-art biometrics protection solutions are based on homomorphic encryption (HE) to perform recognition over encrypted feature vectors, hiding the features and their processing while releasing the outcome only. However, this comes at the cost of those solutions' efficiency due to the inefficiency of HE-based solutions with a large number of multiplications; for (dis-)similarity measures, this number is proportional to the vector's dimension. In this paper, we tackle the HE performance bottleneck by freeing the two common (dis-)similarity measures, the cosine similarity and the squared Euclidean distance, from multiplications. Assuming normalized feature vectors, our approach pre-computes and organizes those (dis-)similarity measures into lookup tables. This transforms their computation into simple table-lookups and summation only. We study quantization parameters for the values in the lookup tables and evaluate performances on both synthetic and facial feature vectors for which we achieve a recognition performance identical to the non-tabularized baseline systems. We then assess their efficiency under HE and record runtimes between 28.95ms and 59.35ms for the three security levels, demonstrating their enhanced speed.

Efficient large-scale biometric identification is a challenging open problem in biometrics today. Adding biometric information protection by cryptographic techniques increases the computational workload even further. Therefore, this paper proposes an efficient and improved use of coefficient packing for homomorphically protected biometric templates, allowing for the evaluation of multiple biometric comparisons at the cost of one. In combination with feature dimensionality reduction, the proposed technique facilitates a quadratic computational workload reduction for biometric identification, while long-term protection of the sensitive biometric data is maintained throughout the system. In previous works on using coefficient packing, only a linear speed-up was reported. In an experimental evaluation on a public face database, efficient identification in the encrypted domain is achieved on off-the-shelf hardware with no loss in recognition performance. In particular, the proposed improved use of coefficient packing allows for a computational workload reduction down to 1.6% of a conventional homomorphically protected identification system without improved packing.

In this paper, we investigate a networked control problem in the presence of Denial-of-Service (DoS) attacks, which prevent transmissions over the communication network. The communication between the process and controller is also subject to bit rate constraints. For mitigating the influences of DoS attacks and bit rate constraints, we develop a variable bit rate (VBR) encoding-decoding protocol and quantized controller to stabilize the control system. We show that the system’s resilience against DoS under VBR is preserved comparing with those under constant bit rate (CBR) quantized control, with fewer bits transmitted especially when the attack levels are low. The proposed VBR quantized control framework in this paper is general enough such that the results of CBR quantized control under DoS and moreover the results of minimum bit rate in the absence of DoS can be recovered.

Efficient low-power accelerators for Convolutional Neural Networks (CNNs) largely benefit from quantization and approximation, which are typically applied layer-wise for efficient hardware implementation. In this work, we present a novel strategy for efficient combination of these concepts at a deeper level, which is at each channel or kernel. We first apply layer-wise, low bit-width, linear quantization and truncation-based approximate multipliers to the CNN computation. Then, based on a state-of-the-art resiliency analysis, we are able to apply a kernel-wise approximation and quantization scheme with negligible accuracy losses, without further retraining. Our proposed strategy is implemented in a specialized framework for fast design space exploration. This optimization leads to a boost in estimated power savings of up to 34% in residual CNN architectures for image classification, compared to the base quantized architecture.

In recent years, Deep Neural Networks (DNNs) have made inroads into a number of applications involving pattern recognition - from facial recognition to self-driving cars. Some of these applications, such as self-driving cars, have real-time requirements, where specialized DNN hardware accelerators help meet those requirements. Since DNN execution time is dominated by convolution, Multiply-and-Accumulate (MAC) units are at the heart of these accelerators. As hardware accelerators push the performance limits with strict power constraints, reliability is often compromised. In particular, power-constrained DNN accelerators are more vulnerable to transient and intermittent hardware faults due to particle hits, manufacturing variations, and fluctuations in power supply voltage and temperature. Methods such as hardware replication have been used to deal with these reliability problems in the past. Unfortunately, the duplication approach is untenable in a power constrained environment. This paper introduces a low-cost error-resiliency scheme that targets MAC units employed in conventional DNN accelerators. We evaluate the reliability improvements from the proposed architecture using a set of 6 CNNs over varying bit error rates (BER) and demonstrate that our proposed solution can achieve more than 99% of fault coverage with a 5-bits arithmetic code, complying with the ASIL-D level of ISO26262 standards with a negligible area and power overhead. Additionally, we evaluate the proposed detection mechanism coupled with a word masking correction scheme, demonstrating no loss of accuracy up to a BER of 10-2.

Quantized deep neural networks (QDNNs) are necessary for low-power, high throughput, and embedded applications. Previous studies mostly focused on developing optimization methods for the quantization of given models. However, quantization sensitivity depends on the model architecture. Also, the characteristics of weight and activation quantization are quite different. This study proposes a holistic approach for the optimization of QDNNs, which contains QDNN training methods as well as quantization-friendly architecture design. Synthesized data is used to visualize the effects of weight and activation quantization. The results indicate that deeper models are more prone to activation quantization, while wider models improve the resiliency to both weight and activation quantization.

In this paper, we present a flexible Variable Precision Computation Array (VPCA) component for different accelerators, which leverages a sparsification scheme for activations and a low bits serial-parallel combination computation unit for improving the efficiency and resiliency of accelerators. The VPCA can dynamically decompose the width of activation/weights (from 32bit to 3bit in different accelerators) into 2-bits serial computation units while the 2bits computing units can be combined in parallel computing for high throughput. We propose an on-the-fly compressing and calculating strategy SLE-CLC (single lane encoding, cross lane calculation), which could further improve performance of 2-bit parallel computing. The experiments results on image classification datasets show VPCA can outperforms DaDianNao, Stripes, Loom-2bit by 4.67×, 2.42×, 1.52× without other overhead on convolution layers.

Security of physical layer key generation is based on the randomness and reciprocity of wireless fading channel, which has attracted more and more attention in recent years. This paper proposes a rate adaptive key agreement scheme and utilizes the received signal strength (RSS) of the channel between two wireless devices to generate the key. In conventional information reconciliation process, the bit inconsistency rate is usually eliminated by using the filter method, which increases the possibility of exposing the generated key bit string. Building on the strengths of existing secret key extraction approaches, this paper develops a scheme that uses Reed-Solomon (RS) codes, one of forward error correction channel codes, for information reconciliation. Owing to strong error correction performance of RS codes, the proposed scheme can solve the problem of inconsistent key bit string in the process of channel sensing. At the same time, the composition of RS codes can help the scheme realize rate adaptation well due to the construction principle of error correction code, which can freely control the code rate and achieve the reconciliation method of different key bit string length. Through experiments, we find that when the number of inconsistent key bits is not greater than the maximum error correction number of RS codes, it can well meet the purpose of reconciliation.

The physical layer secret key generation exploiting wireless channel reciprocity has attracted considerable attention in the past two decades. On-going research have demonstrated its viability in various radio frequency (RF) systems. Most of existing work rely on quantization technique to convert channel measurements into digital binaries that are suitable for secret key generation. However, non-simultaneous packet exchanges in time division duplex systems and noise effects in practice usually create random channel measurements between two users, leading to inconsistent quantization results and mismatched secret bits. While significant efforts were spent in recent research to mitigate such non-reciprocity, no efficient method has been found yet. Unlike existing quantization-based approaches, we take a different viewpoint and perform the secret key agreement by solving a bipartite graph matching problem. Specifically, an efficient dual-permutation secret key generation method, DP-SKG, is developed to match the randomly permuted channel measurements between a pair of users by minimizing their discrepancy holistically. DP-SKG allows two users to generate the same secret key based on the permutation order of channel measurements despite the non-reciprocity over wireless channels. Extensive experimental results show that DP-SKG could achieve error-free key agreement on received signal strength (RSS) with a low cost under various scenarios.

Common randomness of channels offers the possibility to create cryptographic keys without the need for a key exchange procedure. Channel reciprocity for TDD (time-division duplexing) systems has been used for this purpose many times. FDD (frequency-division duplexing) systems, however, were long considered to not provide any usable symmetry. However, since the scattering transmission parameters S\textbackslashtextlessinf\textbackslashtextgreater12\textbackslashtextless/inf\textbackslashtextgreater and S\textbackslashtextlessinf\textbackslashtextgreater21\textbackslashtextless/inf\textbackslashtextgreater would ideally be the same due to reciprocity, when using neighboring frequency ranges for both directions, they would just follow a continuous curve when putting them next to each other. To not rely on absolute phase, we use phase differences between antennas and apply a polynomial curve fitting, thereafter, quantize the midpoint between the two frequency ranges with the two measurement directions. This is shown to work even with some spacing between the two bands. For key reconciliation, we force the measurement point from one direction to be in the midpoint of the quantization interval by a grid shift (or likewise measurement data shift). Since the histogram over the quantization intervals does not follow a uniform distribution, some source coding / hashing will be necessary. The key disagreement rate toward an eavesdropper was found to be close to 0.5. Additionally, when using an antenna array, a random permutation of antenna measurements can even further improve the protection against eavesdropping.

Automated analysis of facial expressions is one of the most interesting and challenging problems in many areas such as human-computer interaction. Facial images are affected by many factors, such as intensity, pose and facial expressions. These factors make facial expression recognition problem a challenge. The aim of this paper is to propose a new method based on the pyramid local binary pattern (PLBP) and the pyramid local phase quantization (PLPQ), which are the extension of the local binary pattern (LBP) and the local phase quantization (LPQ) as two methods for extracting texture features. LBP operator is used to extract LBP feature in the spatial domain and LPQ operator is used to extract LPQ feature in the frequency domain. The combination of features in spatial and frequency domains can provide important information in both domains. In this paper, PLBP and PLPQ operators are separately used to extract features. Then, these features are combined to create a new feature vector. The advantage of pyramid transform domain is that it can recognize facial expressions efficiently and with high accuracy even for very low-resolution facial images. The proposed method is verified on the CK+ facial expression database. The proposed method achieves the recognition rate of 99.85% on CK+ database.

This note addresses the problem of distributed control for a class of nonlinear multi-agent systems over a communication graph. In many real practical systems, owing to communication limits and the vulnerability of communication networks to be overheard and modified by the adversary, consideration of communication delays and cyber-attacks in designing of the controller is important. To consider these challenges, in the presented approach, a distributed controller for a group of one-link flexible joint manipulators is provided which are connected via data delaying communication network in the presence of cyber-attacks. Sufficient conditions are provided to guarantee that the closed-loop system is stable with prescribed disturbance attenuation, and the parameter of the control law can be obtained by solving a set of linear matrix inequities (LMIs). Eventually, simulations results of four single-link manipulators are provided to demonstrate the performance of the introduced method.

In the course of the research, the research of discriminatory methods of handling video information resource based on the JPEG platform was carried out. This research showed a high interest of the scientific world in identifying important data at different phases of handling. However, the discriminatory handling of the video information resource after the quantization phase is not well understood. Based on the research data, the goal is to find possible ways to operation a video information resource based on a JPEG platform in order to identify important data in a telecommunications system. At the same time, the proposed strategies must provide the required pace of dynamic picture grade and hiding in the context of limited bandwidth. The fulfillment of the condition with limited bandwidth is achieved through the use of a lossless compression algorism based on arithmetic coding. The purpose of the study is considered to be achieved if the following requirements are met:1.Reduction of the volume of dynamic pictures by 30% compared to the initial amount;2.The quality pace is confirmed by an estimate of the peak signal-to-noise ratio for an authorized user, which is Ψauthor ≥ 20 dB;3.The pace of hiding is confirmed by an estimate of the peak signal-to-noise ratio for unauthorized access, which is Ψunauthor ≤ 9 dBThe first strategy is to use encryption tables. The advantage of this strategy is its high hiding strength.The second strategy is the important matrix method. The advantage of this strategy is higher performance.Thus, the goal of the study on the development of possible ways of handling a video information resource based on a JPEG platform in order to identify important data in a telecommunication system with the given requirements is achieved.

This paper proposes an efficient face template protection system based on homomorphic encryption. By developing a message packing method suitable for the calculation of the squared Euclidean distance, the proposed system computes the squared Euclidean distance between facial features by a single homomorphic multiplication. Our experimental results show the transaction time of the proposed system is about 14 times faster than that of the existing face template protection system based on homomorphic encryption presented in BIOSIG2020.

In recent years, hashing algorithm has been widely researched and has made considerable progress in large-scale image retrieval tasks due to its advantages of convenient storage and fast calculation efficiency. Nowadays most researchers use deep convolutional neural networks (CNNs) to perform feature learning and hash coding learning at the same time for image retrieval and the deep hashing methods based on deep CNNs perform much better than the traditional manual feature hashing methods. But most methods are designed to handle simple binary similarity and decrease quantization error, ignoring that the features of similar images and hashing codes generated are not compact enough. In order to enhance the performance of CNNs-based hashing algorithms for large scale image retrieval, this paper proposes a new deep-supervised hashing algorithm in which a novel channel attention mechanism is added and the loss function is elaborately redesigned to generate compact binary codes. It experimentally proves that, compared with the existing hashing methods, this method has better performance on two large scale image datasets CIFAR-10 and NUS-WIDE.

Recently, vehicular communications have been the focus of industry, research and development fields. There are many benefits of vehicular communications. It improves traffic management and put derivers in better control of their vehicles. Privacy and security protection are collective accountability in which all parties need to actively engage and collaborate to afford safe and secure communication environments. The primary objective of this paper is to exploit the RSS characteristic of physical layer, in order to generate a secret key that can securely be exchanged between legitimated communication vehicles. In this paper, secret key extraction from wireless channel will be the main focus of the countermeasures against VANET security attacks. The technique produces a high rate of bits stream while drop less amount of information. Information reconciliation is then used to remove dissimilarity of two initially extracted keys, to increase the uncertainty associated to the extracted bits. Five values are defined as quantization thresholds for the captured probes. These values are derived statistically, adaptively and randomly according to the readings obtained from the received signal strength.

Based on the third generation neural network spiking neural network, this paper optimizes and improves a classification and coding method, and proposes an image recognition method. Firstly, the read image is converted into a spike sequence, and then the spike sequence is encoded in groups and sent to the neurons in the spike neural network. After learning and training for many times, the quantization standard code is obtained. In this process, the spike sequence transformation matrix and dynamic weight matrix are obtained, and the unclassified data are output through the same matrix for image recognition and classification. Simulation results show that the above methods can get correct coding and preliminary recognition classification, and the spiking neural network can be applied.

Reconfigurable Intelligent Surface (RIS) is one of the latest technologies in bringing a certain amount of control to the rather unpredictable and uncontrollable wireless channel. In this paper, RIS is introduced in a bidirectional system with two source nodes and a Decode and Forward (DF) relay node. It is assumed that there is no direct path between the source nodes. The relay node receives information from source nodes simultaneously. The Physical Layer Network Coding (PLNC) is applied at the relay node to assist in the exchange of information between the source nodes. Analytical expressions are derived for the average probability of errors at the source nodes and relay node of the proposed RIS-assisted bidirectional relay system. The Bit Error Rate (BER) performance is analyzed using both simulation and analytical forms. It is observed that RIS-assisted PLNC based bidirectional relay system performs better than the conventional PLNC based bidirectional system.

The performance prediction of user equipment (UE) metrics has many applications in the 5G era and beyond. For instance, throughput prediction can improve carrier selection, adaptive video streaming's quality of experience (QoE), and traffic latency. Many studies suggest distributed learning algorithms (e.g., federated learning (FL)) for this purpose. However, in a multi-tier design, features are measured in different tiers, e.g., UE tier, and gNodeB (gNB) tier. On one hand, neglecting the measurements in one tier results in inaccurate predictions. On the other hand, transmitting the data from one tier to another improves the prediction performance at the expense of increasing network overhead and privacy risks. In this paper, we propose cascaded FL to enhance UE throughput prediction with minimum network footprint and privacy ramifications (if any). The idea is to introduce feedback to conventional FL, in multi-tier architectures. Although we use cascaded FL for UE prediction tasks, the idea is rather general and can be used for many prediction problems in multi-tier architectures, such as cellular networks. We evaluate the performance of cascaded FL by detailed and 3GPP compliant simulations of London's city center. Our simulations show that the proposed cascaded FL can achieve up to 54% improvement over conventional FL in the normalized gain, at the cost of 1.8 MB (without quantization) and no cost with quantization.

Reconfigurable Intelligent Surface (RIS) is a new paradigm that enables the reconfiguration of the wireless environment. Based on this feature, RIS can be employed to facilitate Physical-layer Key Generation (PKG). However, this technique could also be exploited by the attacker to destroy the key generation process via manipulating the channel features at the legitimate user side. Specifically, this paper proposes a new RIS-assisted Manipulating attack (RISM) that reduces the wireless channel reciprocity by rapidly changing the RIS reflection coefficient in the uplink and downlink channel probing step in orthogonal frequency division multiplexing (OFDM) systems. The vulnerability of traditional key generation technology based on channel frequency response (CFR) under this attack is analyzed. Then, we propose a slewing rate detection method based on path separation. The attacked path is removed from the time domain and a flexible quantization method is employed to maximize the Key Generation Rate (KGR). The simulation results show that under RISM attack, when the ratio of the attack path variance to the total path variance is 0.17, the Bit Disagreement Rate (BDR) of the CFR-based method is greater than 0.25, and the KGR is close to zero. In addition, the proposed detection method can successfully detect the attacked path for SNR above 0 dB in the case of 16 rounds of probing and the KGR is 35 bits/channel use at 23.04MHz bandwidth.

Wireless security is confronted with the complexity of the secret key distribution process, which is difficult to implement on an Ad Hoc network without a key management infrastructure. The symmetric key extraction mechanism from a response channel in a wireless environment is a very promising alternative solution with the simplicity of the key distribution process. Various mechanisms have been proposed for extracting the symmetric key, but many mechanisms produce low rates of the symmetric key due to the high bit differences that occur. This led to the fact that the reconciliation phase was unable to make corrections, as a result of which many key bits were lost, and the time required to obtain a symmetric key was increased. In this paper, we propose the use of an interval approach that divides the response channel into segments at specific intervals to reduce the key bit difference and increase the key rates. The results of tests conducted in the wireless environment show that the use of these mechanisms can increase the rate of the keys up to 35% compared to existing mechanisms.

In recent years, the integration of encryption schemes into cyber-physical systems (CPS) has attracted much attention to improve the confidentiality of sensor signals and control input signals sent over the network. However, in principle an adversary can still modify the sensor signals and the control input signals, even though he does not know the concrete values of the signals. In this paper, we shall first show that a standard encryption scheme can not prevent some sophisticated attacks such as covert attacks, which remain invisible in the CPS with encrypted communication and a conventional diagnosis system. To cope with this problem, an improved encryption scheme is proposed to mask the communication and to cancel the influence of the attack signal out of the system. The basic idea is to swap the plaintext and the generated random value in the somewhat homomorphic encryption scheme to prevent a direct access of the adversary to the transmitted plaintext. It will be shown that the CPS with the improved encryption scheme is resilient to covert attacks. The proposed encryption scheme and the CPS structure are finally illustrated through the well-established quadruple-tank process.