In practical life, the transcription of speech into text via automatic speech recognition (ASR) models has become very common due to the essential semantic information contained in speech. However, in speech emotion recognition, multimodal models that combine speech and text significantly outperform speech-only models. For this phenomenon, this paper provides an explanation that the existing speech emotion datasets are insufficient for speech-only models to effectively extract crucial semantic information, thereby affecting generalization capability. Based on this explanation, this paper proposes an efficient speech-only model, called semantic funnel speech emotion recognition (SF-SER) model, which excludes the textual input by introducing and integrating some parameters and structures from the ASR model, and then filters the valuable semantic information by the semantic funnel, thus achieving performance better than the speech+text multimodal model. Finally, experimental results show that the SF-SER model achieves significant performance on both the IEMOCAP and EMODB datasets.

The unmanned aerial vehicle (UAV) equipped with mobile-edge computing (MEC) can act as an air base station to provide computing services for Artificial Intelligence of Things (AIoT) devices in remote areas. However, the computation offloading process poses a risk to users’ privacy due to potential information leaks resulting from interactions between UAVs or migration of data between AIoT devices and UAVs. In this article, we proposed a secure aerial computing network that integrates MEC and blockchain technologies to effectively guarantee privacy and security during computation offloading between AIoT devices and UAVs. Additionally, taking into account task offloading scheduling, radio spectrum resource allocation, and computation resource allocation, a joint optimization problem is formulated to minimize the weighted sum of delay and energy consumption throughout the entire computing process. To tackle this issue, we proposed a block coordinate descent (BCD)-based algorithm to solve the mixed-integer and nonconvex problem. Simulation results demonstrate that the proposed algorithm surpasses other baseline approaches.

In the study of information diffusion in social–physical networks, existing works are usually based on information entropy. These works measure and represent the information attribute characteristics independently for social networks and physical networks, resulting in ineffective interactions and waste of resources. Therefore, to solve the key problem of the mismatch between interaction demands and communication resources, the framework of collaborative diffusion based on value measurement is proposed in social–physical networks, including social–physical interaction, cognitive difference, and mutual trust degree of nodes. Based on parameterizing the relative strength of these influences by confidence and collaborative conservation factors, the collaborative diffusion model based on value measurement is established. Extensive simulations verify the influence of value measurement on the collaboration diffusion process, presented by the evolutions of value entropy, sentiment fragmentation, and diffusion range. In addition, the influence of collaboration on information dissemination is confirmed by the comparison of the change of value entropy and diffusion range. These results can help decision makers better balance the matching problem between interaction demands and available resources, which is beneficial to realize customized information diffusion.

Speech emotion recognition has always been a challenging task due to the difference in emotion expression and perception. Currently, in the supervised speech emotion recognition systems, the soft label overcomes the disadvantage of the hard label losing annotations variability and emotion perception subjectivity, but it only considers the emotion perceptions of a few annotators and thus still brings high statistical error. For this issue, this paper redefines the target and designs a novel loss function (denoted as inter-class difference loss), which enables the network to adaptively learn an emotion distribution in all utterances. This not only restricts the negative class probability less than the positive class probability, but also limits the negative class probability close to zero. To make the speech emotion recognition system more efficient, this paper proposes an end-to-end network, called response residual network (R-ResNet), which incorporates the ResNet for features extraction, together with the emotion response module for data augmentation and variable-length data processing. Finally, the experimental results not only demonstrate the advanced performance of our work, but also confirm that the ambiguous utterances contain emotional characteristics. In addition, another interesting finding is that, on the unbalanced dataset, the batch normalization (BN) after addition performs better than BN before addition.

Abstract In massive multiple‐input multiple‐output (MIMO) systems, most of the existing detection work mainly assumes that the channel is the additive white Gaussian noise (AWGN). However, this assumption is difficult to apply to practical communication scenarios. To this end, this paper proposes a message passing detection (MPD) algorithm with a convolutional neural network (CNN) (denoted as iterative MPD‐CNN structure) under correlated noise channels, which is helpful to solve the issue of detection performance degradation in non‐ideal AWGN channels. Firstly, the MPD algorithm based on the channel hardening phenomenon is used to initially estimate the transmitted signal, and then the CNN is concatenated to remove the estimation error for obtaining more accurate channel noise, which provides a beneficial noise distribution for the MPD algorithm. Finally, the theoretical analysis and simulation results show that the proposed iterative MPD‐CNN structure can improve the detection performance in conditions of correlated noise channels and fewer antennas. Compared with the traditional MPD algorithm, its detection performance is more superior.

Social Internet of Things (SIoT), integrating the social networks and Internet of Things (IoT), leads to heterogeneous interactions of thing to thing, human to human, and human to thing, which in turn generates exploded information. Hence, as the soul of SIoT, information with its interaction and diffusion, records the track of humans and things and contains the hidden value for social administration and people's lives. Therefore, how to characterize the interplay between behavior spreading and information diffusion in SIoT is essential to predict and manage the information. Motivated by this, a more comprehensive understanding of the coupled modeling of social interaction and information diffusion processes in SIoT is conceived first. With the widespread adoption of cloud-edge computing, different nodes have different consciousness on information. Hence, a cloud-edge-aided information diffusion model is proposed for efficient interactions, which incorporates the role of edge in timely processing and feedback. On this basis, a blockchain-based cloud-edge SIoT architecture is proposed for traceability and security of information diffusion. Furthermore, the dynamical analysis of the coupled model in SIoT is provided, which illustrates the outbreak threshold, stability, and scale of information propagation. An interesting finding is that interactive behavior spreading only influences the final size of information propagation, not the spreading threshold. Extensive simulation results and detailed performance analysis verify the theoretical results, which are beneficial to provide traceable dissemination so as to find the most influential node and control the scale of information diffusion.

Human activity recognition plays a fundamental role in smart home systems and contributes to remote health monitoring for the elderly or disabled. However, the shallow architecture and heavy parameters of the current 3-D convolutional networks (3-D ConvNets) still restrict the recognition efficiency and spatiotemporal representations. For these issues, this article proposes a framework called LMFNet, which is mainly composed of a deep attentive 3-D residual network (A3D ResNet) and a multistage fusion strategy. Specifically, LMFNet changes the information flow of transmission in the C3D network and implements a residual learning method for efficient training. Besides, the two-stream-fused spatiotemporal attention 3-D ConvNets (2S-FSTA3DCN) are built based on the A3D ResNet. The experimental results show that the proposed LMFNet can achieve a higher recognition accuracy and satisfactory training efficiency compared with the existing methods.

We study how to achieve the state consensus of a whole multi-agent system after adding some new agent groups dynamically in the original multi-agent system. We analyze the feasibility of dynamically adding agent groups under different forms of network topologies that are currently common, and obtain four feasible schemes in theory, including one scheme that is the best in actual industrial production. Then, we carry out dynamic modeling of multi-agent systems for the best scheme. Impulsive control theory and Lyapunov stability theory are used to analyze the conditions so that the whole multi-agent system with dynamic join characteristics can achieve state consensus. Finally, we provide a numerical example to verify the practicality and validity of the theory.

In massive multiple‐input multiple‐output (MIMO) systems, the channel state information (CSI) feedback enables performance gain in frequency division duplex networks. However, with the increase in the number of antennas, the feedback overhead of CSI will also enhance. To this end, this study addresses the issue of massive MIMO CSI reconstruction using convolutional neural network (CNN), long short‐term memory (LSTM) and attention mechanism, and proposes an efficient network architecture (denoted as CNN‐LSTM‐A). To achieve a compromise between performance and complexity, the proposed method significantly reduces the number of training parameters by utilising a single‐stage network rather than a multiple‐stage network. Finally, simulation results show that the authors method can reduce the feedback overhead of CSI effectively, and achieves better performance in terms of CSI compression and recovery accuracy compared with existing state‐of‐the‐art methods.

This paper proposes an adaptive relay selection method that exploits the social network and establishes a physical domain and social domain-based model named the Social-based device-to-device (D2D) relay selection model to address relay selection failure caused by the diversity of users cooperation willingness and the instability of communication links due to human mobility. Under the premise of tolerable interference caused to cellular users, the proposed approach synthetically considers the desired physical-relation factor and the social-interaction factor for the relay user. In particular, the reliability of the D2D link is obtained by considering the transmission stability of the link, assessed by the user encounter history and cooperation willingness characterized by user social distance, which is set as the criterion of relay selection, combined with the physical channel condition. Simulation results show that compared with the traditional socially blind method, the proposed method can increase the probability of relay selection success and reduce the burden of the cellular network to improve the system performance.

Millimeter wave (mmWave) communication has been a promising technology of future fifth generation (5G) cellular networks. Due to the tremendous propagation loss of mmWave communication, device-to-device (D2D) communications are widely used over directional mmWave networks to improve the network throughput. In this paper, a new time resource sharing scheme is proposed based on Stackelberg game for interference D2D links to further enhance the network throughput. The D2D links causing interference can access to the time resource by paying higher price, while the D2D links causing no interference can also be scheduled in the scheme. Concurrent transmission scheduling between D2D links causing interference is formulated as a non-cooperative game, which achieves a distributed transmission power control solution among the interference D2D links. Moreover, the price strategy can be adjusted by setting the interference threshold such that the transmission quality can be guaranteed. The simulation results show that the proposed scheme can achieve significant network throughput gain compared with traditional concurrent transmission scheme.

To further increase the system capacity in cellular networks, establishing stable D2D (Device-to-Device) links with efficient power allocation is necessary due to the communication interferences. Existing works are mainly focused on the interference control and mitigation at the physical layer. However, the information from social interactions among D2D users are also helpful to improve the system performance. In this paper, we first evaluate the level of social ties and take it as the decision metric for transmission mode selection, which can effectively offload mobile traffic. Then a utility-based maximization game is proposed to reduce interference among D2D pairs. In this game, we use the effective social distance as the penalty coefficient, and perform distributed control of the transmission power for D2D communication. Numerical results demonstrate that the proposed scheme significantly improve the delivery ratio and reduce interference by only sacrificing a small amount of total utility.