NAND flash chips have been innovated from two-dimension (2D) design which is based on planar NAND cells to three-dimension (3D) design which is based on vertical NAND cells. Two types of NAND flash technologies–charge-trap (CT) and floating-gate (FG) are presented in this paper to introduce NAND flash designs in detail. The physical characteristics of CT-based and FG-based 3D NAND flashes are analyzed. Moreover, the advantages and disadvantages of these two technologies in architecture, manufacture, interference and reliability are studied and compared.

This paper proposes a novel high-power supply rejection ratio (high-PSRR) high-order curvature-compensated CMOS bandgap voltage reference (BGR) in SMIC 0.18 μm CMOS process. Three kinds of current are added to a conventional BGR in order to improve the temperature drift within wider temperature range, which include a piecewise-curvature- corrected current in high temperature range, a piecewise-curvature-corrected current in low temperature range and a proportional-to-absolute-temperature current. The high-PSRR characteristic of the proposed BGR is achieved by adopting the technique of pre-regulator. Simulation results shows that the temperature coefficient of the proposed BGR with pre-regulator is /°C from 55 °C to 125 °C with a 1.8 V power supply voltage. The proposed BGR with pre-regulator achieves PSRR of 123.51 dB, 123.52 dB, 88.5 dB and 50.23 dB at 1 Hz, 100 Hz, 100 kHz and 1 MHz respectively.

Attacks on web servers are part of the most serious threats in network security fields. Analyzing logs of web attacks is an effective approach for malicious behavior identification. Traditionally, machine learning models based on labeled data are popular identification methods. Some deep learning models are also recently introduced for analyzing logs based on web logs classification. However, it is limited to the amount of labeled data in model training. Web logs with labels which mark specific categories of data are difficult to obtain. Consequently, it is necessary to follow the problem about data generation with a focus on learning similar feature representations from the original data and improve the accuracy of classification model. In this paper, a novel framework is proposed, which differs in two important aspects: one is that long short-term memory (LSTM) is incorporated into generative adversarial networks (GANs) to generate the logs of web attack. The other is that a data augment model is proposed by adding logs of web attack generated by GANs to the original dataset and improved the performance of the classification model. The results experimentally demonstrate the effectiveness of the proposed method. It improved the classification accuracy from 89.04% to 95.04%.

It is the premise of accessing and controlling cloud environment to establish the mutual trust relationship between users and clouds. How to identify the credible degree of the user identity and behavior becomes the core problem? This paper proposes a user abnormal behavior analysis method based on neural network clustering to resolve the problems of over-fitting and flooding the feature information, which exists in the process of traditional clustering analysis and calculating similarity. Firstly, singular value decomposition (SVD) is applied to reduce dimension and de-noise for massive data, where Map-Reduce parallel processing is used to accelerate the computation speed, and neural network model is used for softening points. Secondly, information entropy is added to hidden layer of neural network model to calculate the weight of each attribute. Finally, weight factor is used to calculate the similarity to make the cluster more accuracy. For the problem of analyzing the mobile cloud user behaviors, the experimental results show that the scheme has higher detection speed and clustering accuracy than traditional schemes. The proposed method is more suitable for the mobile cloud environment.

To meet the demands of large-scale user access with computation-intensive and delay-sensitive applications,
combining ultra-dense networks (UDNs) and mobile edge computing (MEC)are considered as important solutions.
In the MEC enabled UDNs, one of the most important issues is computation offloading. Although a number of work
have been done toward this issue, the problem of dynamic computation offloading in time-varying environment,
especially the dynamic computation offloading problem for multi-user, has not been fully considered. Therefore, in
order to fill this gap, the dynamic computation offloading problem in time-varying environment for multi-user is
considered in this paper. By considering the dynamic changes of channel state and users queue state, the dynamic
computation offloading problem for multi-user is formulated as a stochastic game, which aims to optimize the delay
and packet loss rate of users. To find the optimal solution of the formulated optimization problem, Nash Q-l earning
(NQLN) algorithm is proposed which can be quickly converged to a Nash equilibrium solution. Finally, extensive
simulation results are presented to demonstrate the superiority of NQLN algorithm. It is shown that NQLN algorithm
has better optimization performance than the benchmark schemes.

Membership inference attacks on machine learning models have drawn significant attention. While current research primarily utilizes shadow modeling techniques, which require knowledge of the target model and training data, practical scenarios involve black-box access to the target model with no available information. Limited training data further complicate the implementation of these attacks. In this paper, we experimentally compare common data enhancement schemes and propose a data synthesis framework based on the variational autoencoder generative adversarial network (VAE-GAN) to extend the training data for shadow models. Meanwhile, this paper proposes a shadow model training algorithm based on adversarial training to improve the shadow model's ability to mimic the predicted behavior of the target model when the target model's information is unknown. By conducting attack experiments on different models under the black-box access setting, this paper verifies the effectiveness of the VAE-GAN-based data synthesis framework for improving the accuracy of membership inference attack. Furthermore, we verify that the shadow model, trained by using the adversarial training approach, effectively improves the degree of mimicking the predicted behavior of the target model. Compared with existing research methods, the method proposed in this paper achieves a 2% improvement in attack accuracy and delivers better attack performance.

It is of great value and significance to model the interests of microblog user in terms of business and sociology. This paper presents a framework for mining and analyzing personal interests from microblog text with a new algorithm which integrates term frequency-inverse document frequency (TF-IDF) with TextRank. Firstly, we build a three-tier category system of user interest based on Wikipedia. In order to obtain the keywords of interest, we preprocess the posts, comments and reposts in different categories to select the keywords which appear both in the category system and microblogs. We then assign weight to each category and calculate the weight of keyword to get TF-IDF factors. Finally we score the ranking of each keyword by the TextRank algorithm with TF-IDF factors. Experiments on real Sina microblog data demonstrate that the precision of our approach significantly outperforms other existing methods.

At present, there is an urgent need for blockchain interoperability technology to realize interconnection between various blockchains, data communication and value transfer between blockchains, so as to break the ‘ value silo’ phenomenon of each blockchain. Firstly, it lists what people understand about the concept of interoperability. Secondly, it gives the key technical issues of cross-chain, including cross-chain mechanism, interoperability, eventual consistency, and universality. Then, the implementation of each cross-chain key technology is analyzed, including Hash-locking, two-way peg, notary schemes, relay chain scheme, cross-chain protocol, and global identity system. Immediately after that, five typical cross-chain systems are introduced and comparative analysis is made. In addition, two examples of cross-chain programmability and their analysis are given. Finally, the current state of cross-chain technology is summarized from two aspects: key technology implementation and cross-chain application enforcement. The cross-chain technology as a whole has formed a centralized fixed mechanism, as well as a trend of modular design, and some of the solutions to mature applications were established in the relevant standards organizations, and the cross-chain technology architecture tends to be unified, which is expected to accelerate the evolution of the open cross-chain network that supports the real needs of the interconnection of all chains.

This paper attempts to present an interactive color natural images segmentation method. This method extracts the feature of images by using the nonlinear compact structure tensor (NCST) and then uses GrabCut method to obtain the segmentation. This method not only realizes the non-parametric fusion of texture information and color information, but also improves the efficiency of the calculation. Then, the improved GrabCut algorithm is used to evaluate the foreground target segmentation. In order to calculate the simplicity and efficiency, this paper also extends the Gaussian mixture model (GMM) constructed base on the GrabCut to the tensor space, and uses the Kullback-Leibler (KL) divergence instead of the usual Riemannian geometry. Lastly, an iteration convergence criterion is proposed to reduce the time of the iteration of GrabCut algorithm dramatically with satisfied segmentation accuracy. After conducting a large number of experiments on synthetic texture images and natural images, the results demonstrate that this method has a more accurate segmentation effect.

In convolutional neural networks ( CNNs), the floating-point computation in the traditional convolutional layer is enormous, and the execution speed of the network is limited by intensive computing, which makes it challenging to meet the real-time response requirements of complex applications. This work is based on the principle that the time domain convolution result equals the frequency domain point multiplication result to reduce the amount of floating- point calculations for convolution. The input feature map and the convolution kernel are converted to the frequency domain by the fast Fourier transform( FFT), and the corresponding point multiplication is performed. Then the frequency domain result is converted back to the time domain, and the output result of the convolution is obtained. In the shared CNN, the input feature map is much larger than the convolution kernel, resulting in many invalid operations. The overlap addition method is proposed to reduce invalid calculations and speed up network execution better. This work designs a hardware accelerator for frequency domain convolution and verifies its efficiency on the Xilinx Zynq UltraScale + MPSoC ZCU102 board. Comparing the calculation time of visual geometry group 16 ( VGG16 ) under the ImageNet dataset faster than the traditional time domain convolution, the hardware acceleration of frequency domain convolution is 8. 5 times.

In frequency division duplex ( FDD) massive multiple-input multiple-output ( MIMO) systems, a bidirectional positional attention network ( BPANet) was proposed to address the high computational complexity and low accuracy of existing deep learning-based channel state information ( CSI) feedback methods. Specifically, a bidirectional position attention module ( BPAM) was designed in the BPANet to improve the network performance. The BPAM captures the distribution characteristics of the CSI matrix by integrating channel and spatial dimension information, thereby enhancing the feature representation of the CSI matrix. Furthermore, channel attention is decomposed into two one-dimensional (1D) feature encoding processes effectively reducing computational costs. Simulation results demonstrate that, compared with the existing representative method complex input lightweight neural network ( CLNet), BPANet reduces computational complexity by an average of 19. 4% and improves accuracy by an average of 7. 1% . Additionally, it performs better in terms of running time delay and cosine similarity.

The bionics-based swarm intelligence optimization algorithm is a typical natural heuristic algorithm whose goal is to find the global optimal solution of the optimization problem. It simulates the group behavior of various animals and uses the information exchange and cooperation between individuals to achieve optimal goals through simple and effective interaction with experienced and intelligent individuals. This paper first introduces the principles of various swarm intelligent optimization algorithms. Then, the typical application of these swarm intelligence optimization algorithms in various fields is listed. After that, the advantages and defects of all swarm intelligence optimization algorithms are summarized. Next, the improvement strategies of various swarm intelligence optimization algorithms are explained. Finally, the future development of various swarm intelligence optimization algorithms is prospected.

Lattice-based hierarchical identity-based broadcast encryption ( H-IBBE) schemes have broad application prospects in the quantum era,because it reduces the burden of private key generator (PKG) and is suitable for one-to-many communication. However, previous lattice-based H-IBBE schemes are mostly constructed in the random oracle model with more complex trapdoor delegation process and have lower practical application. A lattice-based H-IBBE is proposed in the fixed dimension under the standard model, which mainly consists of binary tree encryption (BTE) system, MP12 trapdoor function and ABB10b trapdoor delegation algorithm. First, this paper uses BTE system to eliminate the random oracle so that the scheme can be implemented under the standard model, and it also uses MP12 trapdoor function to reduce trapdoor generation complexity and obtains a safe and efficient trapdoor matrix; Second, this paper uses ABB10b trapdoor delegation algorithm to delegate user爷s private key, and the trapdoor matrices' dimensions are the same before and after the trapdoor delegation. Comparative analysis shows that trapdoor delegation process reduces complexity, and the size of cipher-text and trapdoor matrix does not increase with deeper trapdoor delegation process. This paper achieves indistinguishability of cipher-texts under a selective chosen-cipher-text and chosen-identity attack (INDr-sID-CCA) security in the standard model based on learning with errors (LWE) hard assumption.

In the 6th generation mobile communication system (6G) era, a large number of delay-sensitive and computation-intensive applications impose great pressure on resource-constrained Internet of things (IoT) devices. Aerial edge computing is envisioned as a promising and cost-effective solution, especially in hostile environments without terrestrial infrastructures. Therefore, this paper focuses on integrating aerial edge computing into 6G for providing ubiquitous computing services for IoT devices. This paper first presents the layered network architecture of aerial edge computing for 6G. The benefits, potential applications, and design challenges are also discussed in detail. Next, several key techniques like unmanned aerial vehicle (UAV) deployment, operation mode, offloading mode, caching policy, and resource management are highlighted to present how to integrated aerial edge computing into 6G. Then, the joint UAV deployment optimization and computation offloading method is designed to minimize the computing delay for a typical aerial edge computing network. Numerical results reveal the significant delay reduction of the proposed method compared with the other benchmark methods. Finally, several open issues for aerial edge computing in 6G are elaborated to provide some guidance for future research.

In order to achieve the lateral control of the intelligent vehicle, use the bi-cognitive model based on cloud model and cloud reasoning, solve the decision problem of the qualitative and quantitative of the lateral control of the intelligent vehicle. Obtaining a number of experiment data by driving a vehicle, classify the data according to the concept of data and fix the input and output variables of the cloud controller, design the control rules of the cloud controller of intelligent vehicle, and clouded and fix the parameter of cloud controller: expectation, entropy and hyper entropy. In order to verify the effectiveness of the cloud controller, joint simulation platform based on Matlab/Simulink/CarSim is established. Experimental analysis shows that: driver’s lateral controller based on cloud model is able to achieve tracking of the desired angle, and achieve good control effect, it also verifies that a series of mental activities such as feeling, cognition, calculation, decision and so on are fuzzy and uncertain.

As one of the critical technologies for the 6th generation mobile communication system (6G) mobile communication systems, artificial intelligence (AI) technology will provide complete automation for connecting the virtual and physical worlds. In order to construct the future ubiquitous intelligent network, people are beginning to rethink how mobile communication systems transmit and exploit intelligent information. This paper proposes a new communication paradigm, called the Intellicise communication system: model-driven semantic communication. Intellicise communication system is built on top of the traditional communication system and innovatively adds a new feature dimension on top of the traditional source coding, which enables the communication system to evolve from the traditional transmission of bit to the transmission of “model”. Like the semantic base (Seb) for semantic communication, the model is considered as the new feature obtained from the joint source-channel coding. The sink node can re-construct the original signal based on the received model and the encoded sequence. In addition, the performance evaluation metrics and the implementation details of the Intellicise communication system are discussed in this paper. Finally, preliminary results of model-driven image transmission in the Intellicise communication system are presented.

In order to meet various challenges in the Internet of things (IoT), such as identity authentication, privacy preservation of distributed data and network security, the integration of blockchain and IoT became a new trend in recent years. As the key supporting technology of blockchain, the consensus algorithm is a hotspot of distributed system research. At present, the research direction of the consensus algorithm is mainly focused on improving throughput and reducing delay. However, when blockchain is applied to IoT scenario, the storage capacity of lightweight IoT devices is limited, and the normal operations of blockchain system cannot be guaranteed. To solve this problem, an improved version of Raft (Imp Raft) based on Raft and the storage compression consensus (SCC) algorithm is proposed, where initialization process and compression process are added into the flow of Raft. Moreover, the data validation process aims to ensure that blockchain data cannot be tampered with. It is obtained from experiments and analysis that the new proposed algorithm can effectively reduce the size of the blockchain and the storage burden of lightweight IoT devices.