Satellite networks play a vital role in providing global connectivity to remote areas, including mountains, forests, and regions affected by natural disasters. The primary challenge lies in the limited communication timeframe between satellites and earth stations (ESs) due to the swift motion of satellites, making timely satellite data downloads through ESs challenging. To address this, we propose a method named priority-aware and throughput-optimized intersatellite cooperative data transmission (PACT). PACT optimizes network throughput while maximizing download priorities for ESs by leveraging intersatellite links (ISLs), considering diverse download priorities for different data types. To comprehensively capture constellation characteristics, PACT models low Earth orbit (LEO) constellations using a spatiotemporal graph. Within the graph, data priorities are assigned as edge weights, and the allocation of initial download windows to ESs is achieved by maximum weighted matching. Subsequently, PACT organizes a contact plan through cooperative scheduling leveraging ISLs. A bipartite graph is constructed based on data awaiting download and link allocation to redistribute remaining download windows optimally through maximum matching. This iterative process enhances network throughput and maintains data priority. Performance assessments in the ndnSIM framework, covering diverse load scenarios, demonstrate the efficiency and benefits of PACT, particularly in prioritizing data downloads.

Efficient and stable resource management in ultra-dense networks is essential for interference reduction and quality of service guarantee for large-scale users. Although reinforcement learning is currently the most talked-about method for achieving efficient resource allocation, it still faces significant challenges when dealing with the key adaptive issues, such as strong interference, large-scale users dynamic demands, and differentiated user priority requirements. In the case, this paper proposes a pervasive power allocation method in the framework of deep reinforcement learning. Concretely, we develop a noisy deep Q-network for guaranteeing stable exploration of average rate and accelerating the model convergence, where the parameterized noise is incorporated into the connection weights of neural networks. On the other hand, considering the varying priority levels of different users, we devise a prioritized experience replay mechanism aimed at enhancing the service quality for users with high importance. Simulation results demonstrate that our proposal can surpass the state-of-the-art methods in terms of average rate, convergence, stability, and model complexity, achieving more adaptive power allocation.

Deep learning based object detection has many important real-life applications. Like other deep learning models, object detection models are susceptible to backdoor attacks. The unique characteristics of object detection, such as returning a set of object bounding boxes with labels, pose new challenges to backdoor scanning. Trigger inversion techniques that aim to reverse engineer a trigger to determine if a model is trojaned have to consider which bounding boxes may be attacked, if the attack causes bounding box relocation, and if the attack may even lead to appearance of ‘ghost’ objects invisible to humans. This much larger attack vector makes trigger inversion very challenging. We propose a new trigger inversion technique that leverages a number of critical observations to reduce the search space to an affordable level. Our experiments on 334 benign models and 360 trojaned models with 4 structures and 6 attacks show that our technique can consistently achieve over 0.9 ROC-AUC. In the latest TrojAI competition on object detection, our solution achieved 0.926 ROC-AUC, out-performing the second-best solution by 21.4% (with 0.763 ROC-AUC).

Cellular network traffic analysis can facilitate automated decision-making as a key enabler in the future of intelligent communication systems. The existing methods mainly devote to the temporal or spatial feature acquisitions of network traffic. However, it is difficult to effectively capture the corresponding multidimensional tendency behavior. To tackle this issue, this article utilizes ordinal pattern (OP) translation networks (OPTNs) to simulate the temporal and spatio-temporal trend characteristics of cellular network traffic, aiming to capture multidimensional latent features at both the series and image levels of cellular traffic. Overall, we have proposed an analytical framework comprising trend visualization and OPTN measures. First, the original traffic is modeled as a complex network for trend feature visualization via the OP encoding, OP translation and OPTN generation. Furthermore, we suggest that some graph theoretic metrics quantify the trend features and complexity of cellular traffic and define a trend correlation (TC) metric to obtain conclusions on traffic similarity. The comprehensive analysis and discussion indicate that this OPTN spatio-temporal traffic analysis effectively simulates the complexity, predictability, and correlation of traffic. Additionally, the traffic prediction demonstration based on OPTN feature support achieved a 33.79% improvement in prediction performance compared to the baseline model. These findings underscore the potential of OPTN analysis in enhancing communication automation, including fault detection, predictability assessment, and predictive modeling guidance.

Small object detection has been a research hotspot in the field of computer vision. Especially in complex backgrounds (CBs), SOD faces various challenges, including inconspicuous small object features, object distortion due to CBs interference, and inaccurate object localization due to various noises. So far, many methods have been proposed to improve the SOD content in CBs. In this paper, based on an extensive study of related literature, we first outline the current challenges and some cutting-edge solutions for SOD, and then introduce the complex background interference types present in small object images and the imaging characteristics of different types of images, as well as the characteristics of small objects. Next, the image pre-processing methods are summarized. Based on this, machine learning-based SOD methods and traditional SOD methods are focused on. Finally, the future development direction is given.

Voice user interface (VUI) brings high efficiency and convenience for the applications of Internet of Things (IoT), meanwhile, it can also cause increasingly serious security issues. The word-level voice liveness detection is proved to be the promising solution to thwart spoofing attacks. However, the complex acoustic feature, diversified attacks, and different interaction distance can severely affects the improvement of detection accuracy. To alleviate this issue, we develop a novel pop noise-based word-level voice liveness detection framework. First, a new voice frame selection method is proposed for determining optimal frames, including short time Fourier transform, low-frequency average energy computation, and sequencing. Then, the acoustic features of the selected frames are calculated by the Gammatone frequency cepstral coefficient (GFCC). Finally, based on these features, a newly built joint voice detector, fusing the self-attentional residual network (ResNet), and light gradient boosting machine (LightGBM), can achieve accurate voice classification. On the popular voice spoofing attack data sets, experimental results show that our proposal significantly outperforming the baseline and the state-of-the-arts models, and it is gender dependent. Moreover, our proposal has good generalization ability for far-field replay voice attack, speech synthesis and voice conversion attacks, and partial fake voice attack. Finally, its effectiveness is verified by the ablation study.

Convolutional neural network (CNN)-based object detection methods have aroused widespread interest in the remote sensing images field, which usually achieve satisfactory results. However, there still exist some factors that cause the detection performance to degrade, such as scales variability, back-ground complexity and objects tininess. In this work, we propose a CNN module that combines semantic information with fine-grained information and can replace the basic block in the backbone of object detection methods to enhance performance. More specifically, our module include a double branches for extracting semantic information and fine-grained information, and an Efficient channel attention (ECA) module for adjusting weights in channel-wise. Experimental results on DIOR dataset suggest the superiority of our module.

Although the soft robot is gaining considerable popularity in dexterous and safe manipulation, accurate motion control is still an open problem to be explored. Recent investigations suggest that reinforcement learning (RL) is a promising solution but lacks efficient adaptability for Sim2Real transfer or environment variations. In this paper, we present a deep deterministic policy gradient (DDPG)-based control system for the continuous task-space manipulation of soft robots. Domain randomization is adopted in simulation for fast control-policy initialization, while an offline retraining strategy is utilized to update the controller parameters for incremental learning. The experiments demonstrate that the proposed RL controller can track a moving target accurately (with RMSE of 1.26 mm), and accommodate to external varying load effectively (with ~30% RMSE reduction after retraining). Comparisons among the proposed RL controller and other supervised-learning-based controllers in handling additional tip load were also conducted. The results support that our RL method is appropriate for automatic learning such that there is no need of manual interference for data processing, particularly in cases with external disturbances and actuation redundancy.

Blast furnace gas (BFG) produced from steel industries is generally one of the most important energy supplies in enterprises. Due to a great deal of output, fluctuation, and heterogeneity in data, it is very difficult to provide profound insights into its internal dynamic. In this article, a novel analysis framework is developed for the BFG data processing, considering the recurrence plot (RP) and the recurrence quantification analysis (RQA). The specific aim is to investigate the relationship between BFG output and its potential influencing factors. This framework can be deemed as a uniform and consistent system with functional components of qualitative visualization and quantitative analysis. Concretely, the BFG outputs related to five factors are separately projected to high-dimensional spaces, followed by that their internal dynamics can be embodied through a 2-D recurrence representation of states. Finally, five RQA parameters are used to quantify the influence of these factors on the BFG output. This is the first attempt revealing the relations among the BFG data from the qualitative and quantitative perspectives. The experimental results show that RP can discover the BFG output patterns of laminar state, chaos, and instability over given three states of influencing factors, and the ranked influential degree can be given by a two-stage standard deviation of all considered RQA measures. Besides, we also demonstrate that the temperature of the hot-blast stove is most relevant to the BFG output, while the influence of considered other factors directly depends on the selected series length.

Support vector machine (SVM) is a powerful machine learning (ML) technology and the distinctive generalization ability makes it one of the most popular approximation tools in the field of Internet-of-Things (IoT)-based marine data processing. However, SVM has been criticized for trial and error of parameters, especially, kernel function. How to determine a suitable kernel for SVM in a specific problem has been rather tricky. To give a systematic research of the field, we concentrate on the self-adaptive selection of kernel functions in the framework of SVM for IoT-based marine data prediction. Specifically, we adopt the optimal kernel for obtaining competitive SVM and devises a kernel selection criteria of such high-efficiency models. Experiments are conducted via IoT-based real-world marine data sets of different characteristics. The results demonstrate that our proposed self-adaptive SVM model can autonomously provide a suitable kernel for given marine environmental factor prediction, and outperform the alternative with the linear combination of multiple kernels. Besides, the superior performance is verified from the perspective of statistic analysis.

Network traffic prediction is a great challenge due to complex statistical properties, generally covering the long-range correlations and self-similarity. To address this issue, this article applies an integrated neural computing model to predict network traffic, namely, enhanced echo-state restricted Boltzmann machine (eERBM). In structure, this model possesses the following functional components of feature learning, information compensation, input superposition, and supervised nonlinear approximation. It is motivated by the introduction of information theory in modeling the hybrid architecture of the echo state network and the restricted Boltzmann machine. This is the first attempt that eERBM is applied in network traffic prediction tasks of different origin and characteristics, considering TCP/IP packet and variable-bit-rate video. By performing a theoretical analysis, we show that eERBM achieves superior nonlinear approximation and robustness in comparison to the baseline methods, and effectively preserves the self-similarity of network traffic traces.

Deep neural networks (DNNs) have widely used in various Internet-of-Things (IoT) applications. Pursuing superior performance is always a hot spot in the field of DNN modeling. Recently, feature reuse provides an effective means of achieving favorable nonlinear approximation performance in deep learning. Existing implementations utilizes a multilayer perception (MLP) to act as a functional unit for feature reuse. However, determining connection weight and bias of MLP is a rather intractable problem, since the conventional back-propagation learning approach encounters the limitations of slow convergence and local optimum. To address this issue, this paper develops a novel DNN considering a well-behaved alternative called reservoir computing, i.e., reservoir in network (ResInNet). In this structure, the built-in reservoir has two notable functions. First, it behaves as a bridge between any two restricted Boltzmann machines in the feature learning part of ResInNet, performing a feature abstraction once again. Such reservoir-based feature translation provides excellent starting points for the following nonlinear regression. Second, it serves as a nonlinear approximation, trained by a simple linear regression using the most representative (learned) features. Experimental results over various benchmark datasets show that ResInNet can achieve the superior nonlinear approximation performance in comparison to the baseline models, and produce the excellent dynamic characteristics and memory capacity. Meanwhile, the merits of our approach is further demonstrated in the network traffic prediction related to real-world IoT application.

The joint transmission (JT) technique was mainly presented by Baier et al. (2000). JT used in the downlink channel does not require a training sequence in the downlink in theory. Moreover, the mobile station (MS) does not demand channel estimation. However, JT requires a huge dynamic range and good linearity of the power amplifier of the transmitter because of the high peak to average ratio of transmitted signals of JT. In this paper, a new scheme called pre-distortion based joint transmission is proposed. Two pre-distortion approaches are also presented. One is based on the pre-clip technique, the other on a pre-nonlinear power amplifier. Finally, simulation results show that the pre-distortion based JT scheme has better BER performance and the BER performance of pre-distortion based JT under nonideal power amplifier is better than that of JT under ideal power amplifier. This is to say, maybe it is not worth compensating the deep channel shading.