A cascaded Kalman filter fiber channel damage adaptive equalization processing optimization algorithm，which is based on the multiple input multiple output constant module algorithm（MIMO-CMA）is proposed to address the issues of excessive rotation of state of polarization（RSOP）speed，polarization mode dispersion（PMD），residual chromatic dispersion，polarization dependent loss（PDL），residual carrier frequency offset（CFO），and carrier phase noise（CPN）that seriously affect communication quality in extreme conditions of optical fiber polarized light signals.Firstly，the quadrature phase shift keying（QPSK）signal containing various optical fiber channel impairments is balanced using MIMO-CMA to achieve residual dispersion and preliminary polarization effect related impairments.Then，the Kalman filter is adopted to balance the residual RSOP damage，CFO，and CPN.Simulation results show that the constellation recovery effect of the input signal after the equalization of the optimization algorithm is good，and the trackable RSOP speed can reach over 9Mrad·s^-1，and can effectively reduce the computational complexity of the algorithm.

In order to delve into the relationships of information flow among various brain regions within causal brain networks，a causal network emotion recognition methodology grounded in electroencephalogram（EEG）signals across communities is proposed.Firstly，time-frequency domain features are extracted from the preprocessed EEG signals.The partial directed coherence（PDC）method is adopted to build the casual brain network，and the Infomap community detection algorithm is used to divide the communities of the brain network.Next，a graph representation of the brain network is formulated，in which the causal interactions，quantified by PDC values between different communities，serve as the edge features，while the node features are defined by the weighted average differential entropy computed for each respective community. Finally，this constructed graph data is fed into a graph convolutional neural network for the ultimate task of emotion classification and recognition.Experimental results demonstrate that compared with the conventional full-channel causal emotion recognition approaches，the proposed method decreases the computational complexity by leveraging the directed causal information between the brain sections，and successfully maintains a high level of emotion recognition accuracy.

Wake-up receiver（WuRX）is a critical module for achieving low-power wireless sensor networks.Classical wake-up receivers which adopts envelope detection（ED）as the first stage，suffer from low sensitivity.To address this issue，the traditional single-ended ED in WuRX is extended to a single-ended-to-pseudo-differential topology，thereby improving the conversion gain and output signal-to-noise ratio of the ED.Additionally，the baseband circuit adopts a low-power fully differential structure，which enhances its ability to suppress common-mode noise，and ultimately improves the sensitivity of the WuRX.Furthermore，this design employs a collaborative approach for the internal current source and clock generation circuit，which reduces current consumption through shared paths.The chip is designed using a 65nm complementary metal-oxide-semiconductor（CMOS）process，with a carrier frequency of 109MHz and a data rate of 33.3bps.Simulation results demonstrate that the proposed WuRX achieves a sensitivity of-80dBm under conditions of a false alarm rate below 1 per hour，and a miss detection rate below 0.1%；moreover，the WuRX operates with an overall power consumption of only 5.9nW at a supply voltage of 0.4V.Compared to classical WuRX designs，this work achieves higher sensitivity while maintaining low power consumption.

Aiming at the problems of tight spectrum resources，insufficient computing resources，and easy interception and tampering of information transmission in the Internet of Things（IoT）mobile edge computing（MEC）network，a resource allocation scheme for the IRS-assisted user backscatter communication（BC）technology secure offloading perception MEC network is proposed.By integrating cognitive radio（CR），IRS and BC technologies，an optimization problem targeting the maximization of secure MEC network throughput is constructed，and thejoint optimization solution is carried out by using methods based on block coordinate descent（BCD），Lagrangian duality，and quadratic transformation.Simulation results show that compared with the random IRS phase scheme and the random beamforming scheme，the proposed scheme can increase the task computation amount of secondary users by about 260%and 178%，verifying its effectiveness and feasibility under complex constraint conditions.

To address the challenges of small inter-class differences in pest details，severe field background interference，and imbalanced sample distribution，a complementary feature fusion dual-stream network for pest recognition is proposed.This network combines the local perception capability of convolutional neural networks with the global modeling ability of the Mamba model，capturing and integrating the global and local information of pest images.A hierarchical multiscale perception module is designed to extract multi-scale image features through grouped hierarchical convolution and enhance pest detail information with a detail enhancement perception strategy.An adaptive focusing Mamba module is designed to locate key pest regions using dynamic convolution operators and reduce complex background interference.Additionally，an attention-weighted fusion module is designed to achieve adaptive interaction and optimization of global and local features through a cross-attention mechanism，further improving the accuracy of semantic expression.A balanced loss function is constructed to mitigate the effects of class imbalance in the dataset.The experimental results show that the network achieves an accuracy of 71.19%on the large-scale pest dataset IP102，and an accuracy of 99.36%on the D0 dataset，demonstrating its ability to effectively identify pest species.

To address the issue of excessive hardware overhead that arises when mapping binary decision diagrams（BDD）to memristor crossbar arrays within path-based memristive logic computing frameworks，a memristive logic synthesis framework based on BDD reordering optimization is proposed.The framework pioneers the application of the adaptive restart genetic algorithm（ARGA）to BDD variable order optimization，which generates BDD structures more suitable for mapping to memristor crossbar arrays，while its built-in adaptive restart mechanism ensures the efficiency of this optimization process which further optimizes the number of rows and columns in the mapped crossbar array，thereby effectively reducing hardware area.Evaluations were conducted on 17 benchmark circuits，and the experimental results show that compared with the original memristive logic framework，the proposed method reduces crossbar area by 15%，operation energy consumption by 26%，and latency by 12%.Moreover，in comparison with other memristive logic frameworks such as COMPACT and CONTRA，the proposed method reduces operation energy consumption by 3 and 4 orders of magnitude respectively，and decreases latency by 80%and 97%，respectively.Through the collaborative optimization of BDD structures and memristor array mapping constraints，this research provides an effective approach to enhance the synthesis efficiency of memristive logic circuits.

Aiming at the optimization of 5G network coverage performance in the complex environment of coal mine underground，a 5G optimization scheme based on deep reinforcement learning is proposed.For the 10km main transport roadway scenario，multiple transmission loss factors such as roadway cross-section size，wall roughness，and equipment occlusion are comprehensively considered，and a signal propagation mathematical model integrating the line-of-sight/nonline-of-sight path loss model and the roughness attenuation factor is established.A deep Q-network is adopted as the value-function approximator for the learning agent，transforming the joint online optimization of base-station placement and transmit power into a multi-objective decision problem that maximizes the coverage while minimizing the number of base stations.Adopt a dynamic power-adjustment mechanism，enabling real-time adaptation to abrupt local signal degradations.Experimental results confirm that the scheme achieves a coverage exceeding 95%，while reducing the number of deployed base stations by 28%compared with a conventional static layout，thereby markedly enhancing the underground 5G coverage and lowering the deployment costs and operational power consumption.

To address the issues of missed detection and low detection accuracy in X-ray weld defect detection，an improved YOLOv8-based detection method is proposed.Firstly，the efficient multi-scale attention（EMA）mechanism is improved by replacing the 3×3 convolutional kernel with a 5×5 kernel to expand the receptive field，and replacing the average pooling with the multi-scale pooling to extract multi-scale features.The improved EMA module is embedded into the backbone network to enhance the model's ability to detect defects at various scales.Then the spatial pyramid pooling fast module is improved by introducing adaptive average pooling and max pooling layers，to improve the perception of weld edge information.Finally，in the neck part，Dual convolution is used to replace traditional convolution，to reduce the parameter number of the model.The WIoU（wise intersection over union）loss function is adopted to replace the CIoU（complete intersection over union）loss function to improve the convergence speed of the model. Experimental results show that，compared to YOLOv8n，the proposed algorithm reduces the number of parameters by 4.02%and increases the mean average precision by 5.9%，which is well-suited for X-ray weld defect detection tasks.

To address the challenges of scale variation and dense object distribution in remote sensing imagery caused by varying imaging angles，a novel object detection algorithm is proposed based on multi-branch fusion self-attention（MFS）.A multi-branch module that integrates convolutional and self-attention mechanisms is designed to build a feature extraction network，and the fourth detection head is built for small objects to facilitate multi-scale feature fusion.Meanwhile，the resulting model is pruned by the DepGraph method to achieve a lightweight architecture.Experiments on the DOTA and NWPU VHR-10 datasets demonstrate that the proposed algorithm achieves mean average precision（mAP）scores of 77.7%and 96.5%respectively，outperforming the peer detectors of similar algorithm complexity.Notably，the pruned version maintains a mAP of 72.9%on DOTA，with only 6.64 million parameters.

For the low detection accuracy problem of small targets and low-contrast defects on the surface of latex gloves，an improved YOLOv（you only look once version）8n algorithm for defects detection on the surface of latex gloves is proposed.The receptive field attention convolution module is introduced in the feature extraction network to dynamically adjust the spatial feature weights within the receptive field，and to enhance the network's focus on defect features. The C2f module is redesigned based on the proposed multi-scale convolution，which captures the contextual information from shallow features through multi-scale convolutional kernels，and improves the network's ability to extract shallow features.The context and the spatial feature calibration network are added to the feature fusion network，where feature calibration refines and aligns contextual information and spatial features，and further enhances the representation of defect features.Experimental results show that on the homemade dataset，the mean average precision（mAP）of the improved algorithm reaches 93.2%，which is 3.1%higher than that of YOLOv8n.It effectively improves the surface defect detection accuracy of latex gloves.In addition，on the VisDrone2019Det and steel defect detection datasets，the mAP reaches 36.1%and 79.8%，respectively，which are 1.1%and 2.7%higher than that of YOLOv8n，and further verifies the effectiveness of the improved algorithm.