This paper designs and implements a low-cost realtime sound source localization system based on Raspberry Pi 5 and ReSpeaker four-microphone array. The core innovation lies in an enhanced GCC-PHAT algorithm integrating subband decomposition and adaptive triple- weighting (frequency distribution, signal-to-noise ratio, and coherence), which improves the time-delay estimation accuracy compared with traditional methods. Simulation verification shows that the algorithm achieves an angular resolution of +/- 2.1 degrees under the signal-to-noise ratio conditions from 0dB to 20dB. The system optimizes the state-space model through Kalman filtering to effectively suppress measurement jitter. By using the geometric characteristics of a circular array, it realizes 360 degrees omnidirectional detection based on the sound wave time-of-arrival model. At the test distance of 1.2m, the azimuth error is <=+/- 4.7 degrees, the total cost is controlled within 1,100 yuan, and the processing delay is < 100ms, providing a cost-effective solution for scenarios such as smart homes and service robots.

In this paper, the delay estimation algorithm based on generalized cross-correlation function (GCC) and its weighted GCC-PHAT algorithm are derived in detail, and the Kalman filter algorithm is introduced, and based on which the combination of GCC-PHAT and Kalman filter algorithms is applied in two-dimensional localization of moving sound source targets based on microphone arrays, and the simulations are carried out to simulate the sound source localization with different signal-to-noise ratios by comparing the different algorithms. By comparing these different algorithms, the sound source localization under different signal-to-noise ratios is simulated, and the localization error is analyzed, and it is found that the combination of the GCC-PHAT algorithm and the Kalman filter algorithm makes the performance still good and the localization accuracy high under lower signal-to-noise ratios, and the simulation method provides a good theoretical basis for the practical engineering.

In two-hop MIMO collaborative systems, a nested PARAFAC model is constructed at the destination node to facilitate joint channel and symbol estimation using the BALS algorithm. This paper derives a closed-form expression for the Cramer-Rao Bound (CRB) for the channel by vectorizing and stacking the received signals, and analyzes the computational complexity using the Slepian-Bangs formula. The nested PARAFAC model enhances channel state information (CSI) estimation by integrating multi-dimensional signal information. The CRB closure expression is derived by employing complex matrix derivation rules and properties of the multivariate joint Gaussian distribution. Simulation results illustrate the CRB performance for both hops and highlight the computational complexity involved.

In this paper, a new ultra-wideband microstrip patch antenna with a total size of $32\times 29\times 1.6 \ \text{mm}^{3}$ is proposed. In order to realize a wide operating bandwidth, the antenna is designed based on the traditional square patch by trimming and parasitizing the patch and using a defected ground structure (DGS). The final resonance range of the antenna is from 2.79 GHz to 15.6 GHz with a bandwidth of 139% and a peak gain of up to 3.1 dBi by CST optimization analysis. The study of the novel antenna proposed in this paper in terms of return loss, gain and far-field radiation direction map reveals that the antenna can provide good ultra-broadband performance, as well as the advantages of simple structure, low cost and easy integration and processing.

We present a system for speech separation and speaker recognition, a recognition technology for multi-person mixed speech signals. It’s mainly used for identification of multi-person speech signals, however, it does not pay attention to the recognition of speech content. The proposed system is divided into two parts: speech separation and speaker recognition. For the first part of the task, we propose a Convolutional Neural Network-Gated Recurrent Unit-Attention (CNN-GRU-Attention) model that uses a convolutional neural network to convert the audio signal into a logarithmic spectrogram as input and a GRU to model the timing information. The presented method delivers better generalization ability and speech separation effect. As for the second part of the task, To address the problem that the convolution process generates a large number of channels containing redundant information such as noise and silent segments, the attention mechanism module SENet is introduced to improve the model, giving more weight to the channels containing important information and improving the recognition effect.

To study the influence of electrothermal effect of ESD protection device on the reliability of inte-grated circuit,a theoretical model of ESD protection device thermal resistance is constructed in the paper based on the Conservation of Energy Principle. By analyzing the temperature relativity between electricity parameter and thermology parameter of ESD protection device characterized by self-heating thermal resistance, the quantitative relationship expression of physical parameter and geometric parameter is obtained, which can be used to improve ESD protection device elec-tro-thermal effect. The simulation testing and the real tape-out testing of ESD protection device self-heating thermal resistance based on ggNMOS are shown and compared. It is concluded that self-heating thermal resistance is an effective index for the protection device electrothermal char-acterization. Based on the established model, the self-heating thermal resistance could be decreased to improve the homogeneity of the heat transference. The uneven distribution of device temperature can be reduced, and the overall reliability of chips can be improved.

针对大自由光谱范围的滤波需求,在全通型微环滤波器的基础上耦合三个微环谐振器,设计出一种自由光谱范围加倍的微环滤波器.根据光波导理论,信号流程图理论和梅森公式推导出滤波器输出端的传递函数,在Matlab环境下对滤波器输出光谱进行仿真分析.仿真结果表明,当母子微环滤波器周长成整数倍时,随着比值的增大,自由光谱范围也逐渐增大,当母微环周长为92.4μm,母子微环周长比为L1:L2=6:1时,滤波器输出光谱达到95 nm.

为改善常规马赫-曾德尔干涉仪滤波器的滤波性能,设计了一种"8"字型微环谐振器辅助马赫-曾德尔干涉仪滤波器.根据微环谐振理论、波导理论与梅森公式,推导出滤波器输出端的传递函数,并采用MATLAB软件仿真.通过调整滤波器的结构参数,在滤波器的输出端获得顶部平坦、边沿陡峭的输出光谱.基于SOI材料的热膨胀原理,通过调节温度来改变SOI材料的有效折射率,进而使得滤波器的输出光谱发生漂移,从而使得滤波器具有可调谐性.

As one of the important ways to transmit and receive information, sound contains a lot of information, and the application scope of sound covers all aspects of life, because the transmission of sound has directionality. In this paper, a room impulse response (RIR) model is established based on the Mono and Multichannel Recording Database (SMARD), and the sound features are extracted using the generalized cross-correlation phase transformation (GCC-PHAT) algorithm. Based on this model, the sound source localization based on convolutional neural network (CNN) is studied. With the help of the microphone array, the system collects randomly generated sound source signals in confined space, and then transmits the collected sound source signals to the CNN model for training. Classify the signal using a well-trained model and finally get the location of the sound source. Experimental results show that the CNN-based sound source localization algorithm has high positioning accuracy under different reverberation conditions and different signal-to-noise ratio environments.

随着计算机与通信技术的发展,个人信息泄露的问题日益严重,人们对个人信息安全也越发重视,身份鉴别则是信息安全的重要一环,寻求更加方便快捷且可靠性高的身份验证方式成为了当今许多研究人员的研究重点.语言作为人类交流最重要的工具,每个人的声音都是独一无二的,因此将声音作为身份鉴别的技术引起了研究人员的兴趣.使用声音鉴别身份的原理是提取说话人语音的特征参数,为其建立数学模型,与待测语音进行比对,从而判断出说话人身份.同时随着深度学习技术的不断发展,解决了传统数学模型的过拟合问题,并且可以更好地对说话人特征进行学习.

本文基于车载T-BOX模型进行渗透攻击从而进一步定位车联网的安全风险.文中介绍了车联网智能终端T-BOX的基本内容和对其进行入侵攻击的方法,以及高危漏洞挖掘.同时对车联网的安全加固提出解决方案.

声音作为人们日常生活中信息表达和接收的重要方式之一,包含了大量的有用信息,其应用范围涵盖了生活的方方面面.因此对于声源定位的深入研究依旧具有广泛的现实意义.本文研究了基于神经网络的声源定位技术,首先介绍了三种传统的声源定位算法,并主要研究了基于相位变换的广义互相关的定位算法(GCC-PHAT),随后根据一种新的单声道和多声道录音数据库(SMARD)建立了房间冲激响应(RIR)模型,并通过麦克风阵列对空间随机生成的声源进行采集,结合全连接神经网络模型进行训练,利用训练好的模型对信号进行分类,最终得到声源的方位.实验结果表明,在不同混响条件和信噪比的环境下,基于神经网络的声源定位算法具有较高的定位准确率.

针对传统的马赫-曾德尔干涉仪滤波器余弦状输出光谱不能满足光通信的滤波需求,将2个微环谐振器与马赫-曾德尔干涉仪相耦合,采用热光性能优良的SOI材料作为器件原材料,设计出一种新型的滤波器结构.利用微环谐振理论、信号流程图理论和梅森公式推导出滤波器的传递函数,并对其输出端口的强度进行模拟分析.仿真结果表明:当微环的周长L与马赫-曾德尔干涉仪上下臂的臂长差 Δd满足L1=2L2=Δd时,滤波器的输出光谱为顶部平坦、边沿陡峭型的输出谱.与普通马赫-曾德尔干涉仪滤波器相比,提出的滤波器的滤波性能有了明显改善.

为优化并改进常规马赫-曾德尔干涉仪滤波器的滤波性能,采用2×2耦合器将单个微环谐振器与常规马赫-曾德尔干涉仪的直通臂相耦合,设计了一种改进的微环辅助马赫-曾德尔干涉仪光学滤波器.采用微环谐振理论与信号流程图理论,推导出该滤波器在输出端的传递函数,并对输出端的输出光谱进行模拟分析.结果表明:与常规马赫-曾德尔干涉仪滤波器相比,在这种改进的滤波器结构中,通过引入微环谐振器的相位调节机制,对滤波器结构耦合系数进行合理设置,可以在滤波器输出端获得顶部平坦、边沿滚降明显、形状类似于矩形的输出谱.与常规马赫-曾德尔干涉仪滤波器余弦型的滤波性能相比,这种改进结构滤波器的滤波性能有了较好的提升.

互联网金融的飞速发展,银行对贷款风险的管理重视程度逐渐升高,因此降低客户的不良贷款率,并且判断出存在较大贷款还款违约风险的贷款客户显得格外重要,基于惩罚逻辑回归算法的贷款违约预测模型便应运而生.该模型根据用户基本属性数据以及下载APP种类的数据,实现特征提取并进行数据加权处理,进而利用带惩罚的逻辑回归来进行预测模型构建,提升对于贷款客户是否会违约的判断准确性,进而可以提升银行对于贷款客户的风险评估及管理控制,采用此方法极大地降低了银行的客户贷款隐患.

In order to deal with the problems of background mixing, pedestrian blur and pedestrian multi-scale in pedestrian tunnels, we proposed an improved faster region based convolution neural network (IF-RCNN) pedestrian detection method, which uses deep CNN to automatically extract features from pictures instead of traditional manual design features. In this paper, an improved region proposal network (RPN) structure is proposed to solve the multi-scale problem of pedestrians in tunnel. The anchor size in RPN network is further improved in the face of pedestrian images in tunnels with small total pixels. Meanwhile, feature fusion technology is introduced to the algorithm to output the features of different convolution layers. The image is fused to enhance the detection performance of blurred and occluded pedestrians in tunnel. Experimental results show that IF-RCNN algorithm has better detection performance in tunnel data set and VOC2007 data set.

针对密集波分复用系统中对窄带宽高消光比的滤波需求,设计一种新型的基于跑道型微环谐振器的窄带宽高消光比滤波器.仿真结果表明:滤波器输出端和下载端的输出光谱具有窄带宽、高消光比和高品质因子的优点,能够在密集波分复用系统中发挥良好的选频滤波功能.滤波器结构中,跑道型微环谐振器的直波导长度与自由光谱范围成反比、直波导与谐振器之间的耦合系数与滤波器消光比成反比.

针对密集波分复用系统中,对特定波长光信号进行精确选择与提取的需求,本文将光纤谐振环与马赫-曾德尔干涉仪相结合,设计出一种新型的光学滤波器.采用信号流程图理论和光纤谐振环理论推导出滤波器输出端的传递函数,并进行模拟分析.仿真结果表明:通过引入光纤谐振环的反馈调节机制、优化滤波器结构中的耦合系数,在滤波器输出端获得顶部平坦、通带形状接近方形、品质因子达到2.815×103的输出光谱.与其他类型的光纤环辅助MZI滤波器相比,本文提出的新型光纤环辅助MZI滤波器能够获得更大的品质因子,可以在密集波分复用系统中发挥更好地选频滤波功能.

物联网领域中,人们对于室内移动物体的定位与跟踪位置的服务需求逐渐在各个领域得到提升,随着数据业务和多媒体业务的快速增长与发展,人们在复杂的室内环境中对定位和导航的需求日益增大.基于激光精度高,速度快的特点,激光测距的定位技术已经成为研究与应用的主要方向.本文提出一种基于激光测距系统对室内环境实时扫描从而推算自身位置信息的室内运动物体的定位.

Internet of Things (IoT) is digitizing the world, and indoor positioning is one of the important applications of them. Indoor positioning refers to the realization of positioning in the indoor environment. The recent research on indoor positioning focuses on Wi-Fi-based methods since GPS cannot achieve the desired effect. A core algorithm in those methods is the K nearest neighbor (KNN) search. In this paper, we proposed an improved indoor positioning algorithm named IpKNN with better accuracy and efficiency. The IpKNN mainly includes two parts. The first part is to use the proposed clustering algorithm to classify the data set, which can improve the computational efficiency. The second part is to improve the positioning accuracy by using the proposed KNN algorithm. The proposed algorithm can achieve high precision and low consumption, and the experiment results also proved it.