Orthogonal Time-Frequency Space (OTFS) modulation has gained attention for combating Doppler shifts in high-speed scenarios. However, off-grid effects from fractional Doppler and delay pose challenges for channel estimation (CE). Existing Off-Grid Sparse Bayesian Learning (OGSBL) methods typically ignore temporal channel correlations by relying on single-frame estimation. To address this challenge, a CE scheme is proposed by combining OGSBL with the Extended Kalman Filter (EKF). Hierarchical Bayesian priors are utilized to characterize channel sparsity. Meanwhile, the EKF is leveraged to recursively track time-varying channel states and nonlinear parameters, thereby reformulating the estimation task into a dynamic reconstruction problem for sparse signals. Simulation results demonstrate that the proposed scheme achieves better NMSE and BER performance than existing OGSBL- and BEM-based benchmarks, at the cost of a moderate increase in computational complexity.

This letter proposes a low-complexity off-grid inverse-free sparse Bayesian learning (OG-IFSBL) algorithm for OTFS channel estimation that avoids computationally prohibitive matrix inversions. The primary innovation is extending the inverse-free variational framework to the non-linear off-grid model of OTFS systems, jointly deriving novel closed-form updates for channel gains and off-grid parameters via relaxed evidence lower bound (ELBO) maximization. Built on the efficient discrete Zak transform (DZT) architecture and integrated with a pruning mechanism, our algorithm matches the high estimation accuracy of off-grid SBL benchmarks with significantly reduced computational complexity, validating its suitability for practical high-mobility communication systems.

Orthogonal time frequency space (OTFS) modulation effectively mitigates the Doppler effect in high-speed railway (HSR) train-to-ground communication, leveraging its robustness in time-frequency doubly-selective fading environments. However, current off-grid sparse Bayesian learning (OGSBL) methods based on fixed grids suffer from two primary limitations: insufficient accuracy in frequency shift quantization and the accumulation of errors from Taylor approximations. In response, this paper proposes a non-uniform grid optimization-based OGSBL channel estimation method. Firstly, a non-uniform dynamic grid partitioning strategy based on an exponential growth law is proposed to address the quantization inaccuracy caused by the Doppler effect. This method assigns higher resolution to high Doppler frequency regions while maintaining lower sampling density in low Doppler frequency regions, striking a balance between accuracy and computational complexity. Secondly, a sensing matrix optimization mechanism based on a multi-variable joint update is proposed to reduce Taylor approximation error accumulation. This mechanism facilitates dynamic reconstruction of the sensing matrix, suppressing error accumulation and accelerating convergence through the alternate update of the integer Doppler matrix, offset coupling matrix, and off-grid Doppler coefficients. Simulation results demonstrate that compared to on-grid estimation and conventional OGSBL methods, the proposed solution achieves significant improvement in channel estimation precision and convergence rate.

Aiming at the problem that the variable-step-length class matching tracking algorithm affects the reconstruction accuracy due to the insufficient selection of atoms during the iteration process of, this paper proposes an algorithm from the pre-selection of atoms with Dice coefficient matching and combines the secondary screening and variable-step-length principle. Firstly, the algorithm of adaptive selection of atoms by Dice coefficient matching is used to improve the correlation between the current residuals and the selected atoms by adding the angular characteristics of the signals while retaining the length characteristics of the signals. At the same time, for the presence of certain atoms in the atoms entering the iteration due to the primary selection in the reconstruction process that can still provide a contribution to the original signal, a secondary selection of atoms is carried out using an adaptive weighting strategy, and then the atoms that have already been selected into the support set are screened out using the projection backtracking principle. The balance between reconstruction accuracy and reconstruction time is achieved by adaptively selecting the step size at different stages of the algorithm operation through the noise value and the signal value estimated at each iteration. The reconstruction experiments on one-dimensional signals and two-dimensional images show that the algorithm can effectively improve the reconstruction effect on signals and images.

In high-speed mobile scenarios, fractional delay and fractional Doppler effects are pervasive, disrupting the inherent sparse structure of the Orthogonal Time Frequency Space (OTFS) system in the delay-Doppler (DD) domain. This leads to the spreading of channel response energy, which not only significantly increases the complexity of channel estimation(CE) algorithms, but also reduces the accuracy of CE. In response to this issue, a two-stage CE scheme is proposed based on Sparse Adaptive Matching Pursuit (SAMP) and Off-Grid Sparse Bayesian Learning (OGSBL) algorithms in this paper. In the signal recovery stage, a doubly fractional CE model is established. On this basis, sparse signals are recovered by using the Sparsity Adaptive Matching Pursuit (SAMP) algorithm to achieve rapid channel path location and preliminary estimation, thus making more effective use of the sparsity of the DD domain channel. In the Bayesian optimization stage, the recovered signal is used as the initial input, the CE is further optimized by SBL. Meanwhile, to effectively mitigate the system's Peak-to-Average Power Ratio (PAPR), Zadoff-Chu (ZC) sequences are incorporated into the pilot structure design. Numerical simulations were conducted to compare the PAPR and Normalized Mean Square Error (NMSE) performance across various pilot patterns, confirming the efficacy of the proposed approach. Simulation results demonstrate that the suggested approach enhances CE precision and speeds up the algorithm's convergence.

To address the problem of degradation of multi-focus image fusion performance due to insufficient decision map accuracy, an algorithm is proposed to improve the decision map accuracy by multi-scale dense residual network combined with attention mechanism. Firstly, a multi-scale dilated convolution module is designed to extract information at various scales from the multi-focused images, enabling a wider range of perceptual field feature extraction. Then the feature extraction capability is enhanced using a dense residual network to preserve the middle layer feature information. Additionally, the attention mechanism of shuffle coordinate is employed at the end of the encoding sub-network, focusing on the global context information to enhance the capability of highlighting important feature information in multi-focused images. Finally, the decoding sub-network is constructed by full convolutional layers to reconstruct the features, and the decision map is generated directly by the symbolic function binary mapping to guide the image fusion. The fusion experiments show that the method significantly enhances various indices, including standard deviation, spatial frequency, and mutual information, compared to several classical comparison algorithms.

In order to improve the accuracy of sparse Bayesian learning (SBL) channel estimation in the OTFS system, a hierarchical SBL channel estimation algorithm with adaptive adjusted hyperparameters is adopted. Firstly the algorithm transforms the channel estimation problem into a sparse signal recovery problem based on the sparsity of the delay-Doppler domain channel using the principle of compressed sensing. Secondly the sparse signal is modeled as a three-layer a priori SBL(T-SBL) model. Finally, a particle swarm optimization (PSO) algorithm with high global seeking capability is proposed to optimize the sensing matrix for the effect of the nonorthogonality of the sensing matrix on the algorithm. The simulation results show that the T -SBL algorithm improves the normalized mean square error (NMSE) performance of OTFS system in low SNR environment, and the BER performance of the system can be effectively enhanced when the T -SBL algorithm is used to estimate the channel after optimizing the sensing matrix.

In the Orthogonal Time-Frequency Space (OTFS) modulation system, it is difficult to estimate the Channel State Information (CSI) of physical path corresponding to fractional Doppler channel, and the computation is very complicated. To solve these problems, a channel estimation algorithm PRS-PMF (Pilot Resource Saving-Pulse Matched Filtering) for pulse matching filter which saves pilot resources is proposed. In the algorithm, the embedded auxiliary pilot is employed to obtain the equivalent channel estimation, then the CSI of each path is estimated through the cross-correlation matched filter. Compared with the traditional cross-correlation matched filter channel estimation algorithms, it can reduce the pilot resource occupy and the computational complexity. On this basis, the OTFS system is windowed to reduce the number of integer samples of the main lobe of the window response and reduce the side lobe level, which improves effectively the autocorrelation characteristics of the equivalent channel Doppler response function and thus reduces the interference of other symbols and noise on the estimated symbols.

A compensation algorithm for OTFS receiver was proposed to solve the problem of transmission signal distortion caused by the nonlinear characteristics of power amplifier (PA) in the orthogonal time frequency space (OTFS) system.The channel state information was estimated by Bussgang theorem combined with the complex coefficient polynomial of PA and the average energy of the distortion term.Based on it, the transmitted signal was reconstructed with distortion term and the estimated value was updated for the polynomial normalization.Furthermore, the iterative zero-forcing equalization algorithm was employed to complete adaptive compensation analysis of the nonlinear distortion at the receiver.The results show that the nonlinear influence on transmission signals is reduced, which caused by Saleh and Rapp models, and BER performance of OTFS system is improved effectively.

无关项是"数字电子技术"非完全定义逻辑函数理论教学中的重要内容,无关项包涵约束项和任意项,合理应用无关项对数字逻辑电路实践有重要意义.当前,教学上对这三个概念存在争议和困惑,国内外通行教材讲解并不一致.对约束项、任意项和无关项概念追本溯源进行了定义论证,并举具体实例就概念的理解和区分进行说明.对数字逻辑电路无关项内容的教学有一定的指导作用.

在激烈的就业竞争形势和经济社会发展的迫切需求下,学生是否具备创新创业的能力和素质是衡量人才质量的重要标准之一.从构建专业课程群、实践教学、线下线上教学、评价考核体系、学科竞赛等角度出发,探究在"新工科"建设背景下通信技术课程群与创新创业教学的融合.实践证明,通过教学改革,学生的工程能力显著提升,知识体系更加完善,创新创业能力大大增强.

针对图像识别算法中卷积神经网络运算量大、耗时长、对资源需求高的问题,提出了一种基于现场可编程门阵列的卷积神经网络硬件加速器设计方案.在现有的网络基础上将批量规范化作为训练模型结构的一部分加入卷积层,可有效解决梯度爆炸,加快网络收敛;设计动态定点量化方式,对卷积运算过程中的浮点数定点量化后进行卷积计算,研究不同硬件平台下的加速效果;采用XC7 Z020开发板结合现场可编程门阵列高级综合工具设计并行流水线计算方法的硬件结构.结果表明,该方案有效地节省了查找表和寄存器资源的使用,相比于CPU计算速度提升约10倍.

针对传统多路径负载均衡算法无法有效地感知网络的运行状态、不能综合考虑链路的实时传输状态以及大多数算法缺少自适应性的问题,基于软件定义网络(SDN)的集中控制和全网管控思想,提出一种基于蜘蛛猴优化的SDN自适应多路径负载均衡算法(SMO-LBA).首先,利用数据中心网络的感知能力来获取多路径的实时链路状态信息;然后,利用蜘蛛猴算法的全局探索和局部开采能力将链路空闲率作为每条路径的适应度值,并引入自适应权重对路径进行动态评估及更新;最后,寻找数据中心网络中链路占用率最小的路径,确定其为最优转发路径.选用胖树拓扑在Mininet平台上进行仿真实验,实验结果表明SMO-LBA可提高数据中心网络的吞吐量和平均链路利用率,实现网络自适应负载均衡.

In view of the problem that dual-dispersion channels will reduce the reliability of channel estimation in high-speed mobile environments, a channel estimation algorithm based on compressed sensing is proposed in the input-output model of Orthogonal Time-Frequency-Space (OTFS) modulation system. The maximum Doppler shift and the maximum delay in the channel are employed to determine the size of the pilot transmission matrix in the algorithm. Compared with the traditional Orthogonal Matching Pursuit (OMP) channel estimation algorithms, the pilot resources can be saved in the proposed algorithm while the accuracy of similar channel estimation is guaranteed. Furthermore, the phase rotation of the OTFS modulation symbols is used to improve the rank of the differential matrix. Theoretical analysis and simulation results show that the diversity order of the OTFS system is improved and noise interference is reduced.

针对传统来波方向(direction-of-arrival,DOA)估计在信号相干、低信噪比与噪声非均匀环境下性能差的问题,基于修正后的矩阵分解,提出一种利用凸优化的协方差矩阵最优DOA估计方法.修正后的矩阵分解方法,解相干的同时克服了孔径损失;然后,利用凸优化,重构出无噪声的协方差矩阵;最后,利用最小化搜索计算出DOA.仿真结果表明,所提算法与矩阵分解(matrix decomposition,MD)算法、基于l1范数的奇异值分解(l1-norm singular vector decomposition,l1-SVD)算法以及基于空间平滑的协方差秩最小化估计(spatial smoothing based covari-ance rank minimization,SS-CRM)算法比较,能更好地抑制非均匀噪声,且在低信噪比条件下,依然性能良好.

针对传统机器学习算法在分布式拒绝服务攻击检测中存在检测时间过长、控制器负载过大等缺点,提出基于混沌理论模型下分布式拒绝服务攻击流量预测算法.首先,收集正常数据包和流表信息,当异常流表信息进入系统时,若初始状态之间存在微小差异,初始位置的运动状态轨迹会以指数速率分离;根据Lyapunov指数的取值范围判断进入系统的数据和流信息是否合法,若判断为异常流信息,立即清除.实验结果表明,提出的研究思路对攻击流信息敏感度较高,对分布式拒绝服务攻击的检测率、准确率、误报率相较于传统机器学习算法和统计分析算法有明显的提高.

原DV_Hop算法中存在节点间距离估算的累计误差以及待测节点坐标求解时的误差问题.在平均跳距的计算阶段,信标节点先后以两个通信半径广播自身位置信息,精确了节点间最小跳数值,加入修正因子来校正平均跳距,得到更精确的未知节点坐标.采用基于线性优化惯性权重和线性加权改进的学习因子同步变化的粒子群算法来优化待测节点位置解析误差,降低待测节点的平均定位误差.仿真结果表明,与原有算法相比,该方法可以有效地降低估算距离误差,提高待测节点的定位精度.

In order to solve the problem of low node survival in traditional network energy efficiency optimization method, an energy efficiency optimization method based on improved genetic algorithm is proposed. According to the principle of the improved genetic algorithm, the super-dense heterogeneous network is encoded in the coding space, and the fitness function is determined. By establishing the mathematical model of network energy consumption and using cluster selection algorithm to assign the encoded genetic algorithm operators, the problem of network energy consumption is optimized by using convex optimization. The multi-objective programming function is established, the optimal solution of the optimization problem is obtained by using the prior preference method, and the energy efficiency optimization scheme of the network is obtained. Experimental results show that this method can improve the energy utilization efficiency of the network, and has the advantage of more nodes surviving after optimization

软件定义网络因其特定的网络结构,有集中控制获取与分配全球网络资源等特点.针对软件定义网络中的负载均衡问题,在原有蚁群算法的基础上,提出了一种改进的蚁群优化负载均衡算法,主要思想如下:利用蚁群算法的搜索规则,将链路负载均衡度、流接受率、时延和丢包率作为蚂蚁选择下一节点的影响因素,在多个约束条件下,获得传输的最佳路径.理论分析及仿真结果说明,所提出的算法具有较好的负载平衡能力,而且可以提高网络的服务质量.

针对高校电子信息专业基础实验课程中工程性不足的问题,融合相关课程实验教学内容,通过Matlab软件与FPGA硬件设计了综合性实验.以FIR滤波器和ASK调制综合实验为例,对实验内容理论分析、Matlab验证、FPGA实现等设计与实验过程做了详细解析.通过此类综合实验学生能够参与理论分析、软件仿真、硬件实现的整个系统开发流程,有利于培养学生综合运用专业知识解决复杂工程问题的能力.