In UAV-assisted Mobile Edge Computing (MEC), conventional efficiency-first paradigms often cause severe service degradation for disadvantaged users by neglecting fairness. This paper tackles this critical challenge by proposing a joint optimization framework grounded in the max-min fairness criterion. We formulate a complex non-convex mixed-integer nonlinear program (MINLP) to minimize the maximum weighted cost across all users by holistically optimizing UAV trajectories, task offloading, and multi-dimensional resources. To solve this NP-hard problem, we design a novel hierarchical algorithm, AGJO-SCA, that synergizes an adaptive metaheuristic for global search with convex optimization for local resource refinement. Compared to an efficiency-driven baseline, our scheme dramatically reduces the worst-case user's cost by 73% with only a negligible impact on average system efficiency. Crucially, our fairness-centric design demonstrates superior reliability, achieving a task violation rate of only 8.5% under a stringent 1.0 s delay constraint-a scenario where traditional methods largely fail. This work provides a robust and equitable solution for developing dependable next-generation air-ground integrated networks.

Mobile Edge Intelligence is a promising computing paradigm for mobile users to access Artificial Intelligence (AI) services. It seamlessly integrates AI online inference processes with Mobile Edge Computing (MEC), delivering low-latency services through application offloading. However, previous works often overlooked the need to pre-deploy relevant AI models on the edge server and disregarded the impact of model deployment on service performance. Furthermore, even with proper model deployment, service efficiency still depends heavily on the resource allocation strategies of the edge server. To this end, we propose a hybrid Deep Reinforcement Learning (DRL) approach, termed SA2CNN, which jointly optimizes discrete decisions on AI model caching and application offloading, along with continuous allocation of bandwidth and frequency resources, to minimize user energy cost and task latency. We first formulate the aforementioned challenges into a Markov decision process, then decompose it into two low-complexity sub-problems: the Discrete Destination Selection (DDS) problem and the Continuous Resources Allocation (CRA) problem. DRL is responsible for outputting the DDS actions, while CRA sub-problem results are solved by convex optimization. Furthermore, we decouple the caching and offloading decisions across time slots to eliminate the impact of model deployment on task performance, and employ a deep convolutional neural network to effectively learn the underlying temporal dependencies. Simulations show that our approach reduces service cost by 52.1% to 68.1% compared to the baselines, exhibiting significant performance enhancements.

Object localization has been the focus of research in FGVC(Fine-Grained Visual Categorization). With the aim of improving the accuracy and precision of object localization in multi-branch networks, as well as the robustness and universality of object localization methods, our study mainly focus on how to combines coordinate attention and feature activation map for target localization. The model in this paper is a three-branch model including raw branch, object branch and part branch. The images are fed directly into the raw branch. CAOLM is used to localize and crop objects in the image to generate the input for the object branch. APPM is used to propose part regions at different scales. The three classes of input images undergo end-to-end weakly supervised learning through different branches of the network. The model expands the receptive field to capture multi-scale features by SB-ASPP. It can fuse the feature maps obtained from the raw branch and the object branch with SBBlock, and the complete features of the raw branch are used to supplement the missing information of the object branch. Extensive experimental results on CUB-200-2011, FGVC-Aircraft and Stanford Cars datasets show that our method has the best classification performance on FGVC-Aircraft and also has competitive performance on other datasets. Few parameters and fast inference speed are also the advantages of our model.

With the widespread application of new technologies, fine-grained authorization requires a large number of access control policies. However, the existing policy retrieval method applied to a large-scale policy environment has the problem of low retrieval efficiency. Therefore, this paper proposes an attribute access control policy retrieval method based on the binary sequence. This method uses binary identification and binary code to express access control requests and policies. When the policy is retrieved, the appropriate group is selected through the logical operation of the access control request and the policy binary identification. Within the group, the binary code of the access control request is matched with the binary code of all rules to find suitable rules, thereby reducing the number of matching attribute-value pairs in the rule and improving the efficiency of policy retrieval. Experimental results show that the policy retrieval method proposed in this paper has higher retrieval efficiency.

With the widespread application of cloud computing sharing technology, the demand for cross-domain interaction is also increasing. However, due to the uncertainty of interaction behaviour and the difference of network service quality, the risk of cross-domain interaction cannot be accurately evaluated. Therefore, this paper proposes a risk situation evaluation model based on interdomain interactions. The model collects interactive credentials such as the frequency, credibility, and time-effectiveness of the user-submitted evaluations. At the same time, it collects the evaluation of quality of service provided by the network security domain. Then, we set up a risk evaluation equation based on the interaction credentials to implement the risk evaluation of cross-domain interaction behaviour. Finally, we apply MATLAB platform to simulate the evolution process of evaluation. The experimental results show that, compared with other models, the evaluation method proposed in this paper improves the accuracy of the evaluation results and meets the security requirements of multidomain interaction.

In recent years, evolutionary game theory has been gradually applied to analyze and predict network attack and defense for maintaining cybersecurity. The traditional deterministic game model cannot accurately describe the process of actual network attack and defense due to changing in the set of attack-defense strategies and external factors (such as the operating environment of the system). In this paper, we construct a stochastic evolutionary game model by the stochastic differential equation with Markov property. The evolutionary equilibrium solution of the model is found and the stability of the model is proved according to the knowledge of the stochastic differential equation. And we apply the explicit Euler numerical method to analyze the evolution of the strategy selection of the players for different problem situations. The simulation results show that the stochastic evolutionary game model proposed in this paper can get a steady state and obtain the optimal defense strategy under the action of the stochastic disturbance factor. In addition, compared with other kinds of literature, we can conclude that the return on security investment of this model is better, and the strategy selection of the attackers and defenders in our model is more suitable for actual network attack and defense.

Sensor-cloud is a developing technology and popular paradigm for various applications. It integrates wireless sensor into a cloud computing environment. On the one hand, the cloud offers extensive data storage and analytical and processing capabilities not available in sensor nodes. On the other hand, data distribution (such as time synchronization and configuration files) is always an important topic in such sensor-cloud systems, which leads to a rapid increase in energy consumption by sensors. In this paper, we aim to reduce the energy consumption of data dissemination in sensor-cloud systems and study the optimization of energy consumption with time-varying channel quality when multiple nodes use the same channel to transmit data. Suppose that there is a certain probability that the nodes send data for competing channel. And then, they decide to distribute data in terms of channel quality for saving energy after getting the channel successfully whether or not. Firstly, we construct the maximization problem of average energy efficiency for distributing data with delay demand. Then, this maximization problem transferred an optimal stopping problem which generates the optimal stopping rule. At last, the thresholds of the optimal transmission rate in each period are solved by using the optimal stopping theory, and the optimal energy efficiency for data distribution is achieved. Simulation results indicate that the strategy proposed in this paper can to some extent improve average energy efficiency and delivery ratio and enhance energy optimization effect and network performance compared with other strategies.

Considering the widespread use of mobile devices and the increased performance requirements of mobile users, shifting the complex computing and storage requirements of mobile terminals to the cloud is an effective way to solve the limitation of mobile terminals, which has led to the rapid development of mobile cloud computing. How to reduce and balance the energy consumption of mobile terminals and clouds in data transmission, as well as improve energy efficiency and user experience, is one of the problems that green cloud computing needs to solve. This paper focuses on energy optimization in the data transmission process of mobile cloud computing. Considering that the data generation rate is variable, because of the instability of the wireless connection, combined with the transmission delay requirement, a strategy based on the optimal stopping theory to minimize the average transmission energy of the unit data is proposed. By constructing a data transmission queue model with multiple applications, an admission rule that is superior to the top candidates is proposed by using secretary problem of selecting candidates with the lowest average absolute ranking. Then, it is proved that the rule has the best candidate. Finally, experimental results show that the proposed optimization strategy has lower average energy per unit of data, higher energy efficiency, and better average scheduling period.

M2M (Machine-to-Machine) communication networks have received more attention in recent years. Performance analysis is helpful for the design and evaluation of network deployment. In this paper, we mainly focus on performance analysis on M2M communication networks. We develop a performance analysis framework for M2M communication networks based on stochastic network calculus. We use Poisson process model to describe arrival process, and derive stochastic arrival curve and stochastic service curve. Then we provide some performance bounds for output, backlog and delay in M2M communication networks. The output characterization of a network node on M2M communication networks is characterized by stochastic arrival curve and stochastic service curve's deconvolution in a time interval, but backlog bound is characterized by stochastic arrival curve and stochastic service curve's deconvolution at a time. Delay bound will become much bigger with the increasing of the arrival traffic. For M2M communication networks with different sizes, the delay bound of the whole network is relatively low. On the other hand, network node has a small amount of delay at the beginning of traffic arrival, as parts of arrival traffic served delay will become gradually zero.

In this paper, following our recent developed concept of subnet model in mesh networks, we continue to investigate the characterizations of probabilistic fault tolerance for the mesh networks with faulty node. We consider two fault models: each node has deterministic or stochastic failure probability, then we study the fault tolerance of mesh networks based on our novel technique - subnet model. We derive lower bounds on the connectivity probability for mesh networks. Our study shows that mesh networks of practical size can tolerate a large number of faulty nodes thus are reliable enough for multicomputer systems under deterministic or stochastic node failure probability. Comparing with deterministic node failure probability, stochastic model is close to realistic case.

In this paper, we propose the expected schedulability to characterize the schedulability for the stochastic tasks on a single real-time computer processor. Our results show that the stochastic model has more flexibility to characterize the real-time tasks than the deterministic model. The expected schedulability is related to the real-time t at tasks arrival and may show that a sub-set of task would be scheduled at any given real-time interval. The numerical analysis based on our theoretic results is consistent with the simulation analysis. Both numerical and simulation results show that the tasks would be scheduled for real-time tasks in realtime systems if the traffic load is less than a specific value which is more than 69%, which was provided in some deterministic situations. This observation implies that the expected schedulability based on the stochastic model would provide a bigger threshold for the real-time tasks to be scheduled.

Distributed adaptive routing, which routes through alternative paths in presence of faulty network components, is an important subject in the research of fault tolerant multicomputer systems. In this paper, we study the performance of routing scheme under the model in which each network node has independent failure probability. We concentrate on two different routing scheme on the mesh-connected multicomputer systems. We develop new techniques that enable us to derive formally proven success probability for these routing schemes, and compare and analyze their performance. The formal study shows that distributed adaptive routing schemes have the clear advantage over centralized adaptive routing schemes, not only for the well-known facts that distributed routing schemes require no global knowledge of network faults and computationally more efficient, but also because the distributed routing schemes have higher success probability and are more robust to node failure probability and to network size.

Reliability and availability are critical when faults appear in the design of large multicomputer systems. On the other hand, it is very difficult to predict the reliability and availability of multicomputer systems. In this paper, we study the reliability and availability of large multicomputer systems under a more realistic model in which each network node has an independent failure probability. We mainly consider the reliability and availability of large mesh-connected multicomputer systems. The metric is connectivity probability of networks. In a previous work (J. Chen and T. Wang, Proc. 14th Int. Conf. Parallel and Distr. Comp. and Sys., pp. 606-611, 2002), we proved that if the node failure probability is fixed, then the connectivity probability of mesh networks can be arbitrarily small when the network size is sufficiently large. Thus, it is practically important for multicomputer system manufacturers to determine the upper bound for node failure probability, when the probability of network connectivity and the network size are given. We develop another novel technique to formally derive lower bounds on the connectivity probability for mesh networks. Our study shows that mesh networks of practical size can tolerate a large number of faulty nodes and thus are reliable enough for multicomputer systems. For example, we formally prove that as long as the node failure probability is bounded by 0.09% (note that according to current VLSI technology, building network nodes with failure probability under 0.09% is achievable), mesh networks of up to a million nodes remain connected with a probability larger than 99%. The results for mesh network reliability and availability are obtained by formal and thorough mathematical proofs.