Search

Alongside advancements in Artificial Intelligence (AI), significant progress has been made in big data processing, edge/cloud computing, and ubiquitous computing in the past two decades. These advancements catalyzed the development and adoption of Digital Twins (DT) across various domains, serving as virtual replicas of Physical Objects (POs). DTs provide advanced visualization and simulation capabilities, enabling effective estimation, optimization, and forecasting of PO's behaviors. However, the widespread adoption of DTs has introduced various security threats, vulnerabilities, and attacks. Despite ongoing research in DT applications and security, there is a lack of systematic review of the DT security literature across domains and architectural layers. This study fills this gap by systematically reviewing DT research, focusing on three interrelated aspects: DT applications, architectural layers, and security. We explore DT's architectural layers, functional requirements, application, and creation software to identify potential threats, attacks, and vulnerabilities specific to DT layers and application domains. We then systematize our findings under a unified security framework and pinpoint countermeasures against identified security challenges. Furthermore, our study explores DT's role in mitigating existing cyber threats, and we conclude our work by identifying open challenges and potential research directions. [ABSTRACT FROM AUTHOR]

Network-accessible task offloading is for low latency; offline task offloading is for low cost. Offline devices cannot directly access service nodes due to lack of resources. Accordingly, the latter involves more steps and optimization variables such as: where to offload tasks, how to allocate computation resources, how to adjust offloading ratio and transmit power, and such optimization variables and hybrid combination features are highly coupled with each other. In this paper, we first formulate a Mixed Integer Nonlinear Programming Problem (MINLP) for such task offloading under energy and delay constraints. Furthermore, we decompose it into two subproblems so as to efficiently solve the formulated MINLP, and design a low-cost and low-complexity Joint Optimization for Energy Consumption and Task Processing Delay (JOET) algorithm to optimize selection decisions, resource allocation, offloading ratio and transmit power adjustment. We carry out extensive simulation experiments to validate JOET. Simulation results demonstrate that JOET outperforms many representative existing schemes in quickly converge and effective reduction of energy consumption and delay. Specifically, the average energy consumption and the average delay have been reduced by 15.93% and 13.70%, respectively, and the load balancing efficiency has increased by 10.20%. [ABSTRACT FROM AUTHOR]

Modern cloud computing applications have stringent low-latency and high-throughput requirements to meet the increasingly diverse demands from customers. In edge datacenters, the popular load balancing scheme based on packet spraying, i.e., random packet spraying (RPS), makes good use of multiple equal-cost paths to ensure high link utilization and efficient transmission. However, RPS performs poorly under the bursty traffic scenario due to packet loss or even timeout. Therefore, to reduce the tail latency caused by retransmission, we design a coding-based random packet spraying named CRPS. Specifically, the source host transmits forward error correction (FEC) encoded packets and dynamically adjusts the data redundancy based on the packet loss rate. Then the switch randomly sprays the encoded packets across all equal-cost multiple paths to implement parallel transmission. In this way, once enough encoded packets from any parallel paths arrive at the destination host, the original packets can be decoded immediately to address the adverse impact of packet loss. The NS-2 simulation results show that CRPS effectively reduces the probability of timeout and significantly improves the tail flow completion time (FCT) by up to 72% compared with the state-of-the-art multipath transmission schemes. [ABSTRACT FROM AUTHOR]

Task offloading is one of the important research directions in Internet of Vehicles. Mobile edge computing and 5G heterogeneous network frameworks bring new opportunities for task offloading research. Most research works propose resource scheduling schemes and offloading strategies for task offloading within the same period. However, the allocation of resources for offloading tasks at different periods cannot be planned uniformly. The utilization and release of resources in previous periods affect the scheduling of resource allocations in later periods. To address the task offloading problem at different periods, this paper proposes the MPTO-MT algorithm. Specifically, the algorithm first determines whether the task can be offloaded in time at a certain period according to the number of candidate offloading nodes of the task in this period. Secondly, the algorithm selects the offloading node and the task transmission method. In order to select the appropriate offloading node and transmission method, we introduce the service utility and propose three heuristics of selecting the offloading node. Finally, the algorithm updates the resource information of nodes at the next period based on the utilization and release of node resources at the period, and the service time of the nodes. Comparative experiments with other baseline algorithms show that the proposed MPTO-MT algorithm improves the task completion rate by 4.55%, the timely rate of task completion by 3.57%, and shorter average service time. [ABSTRACT FROM AUTHOR]

As an important task of Graph Neural Networks (GNN), graph classification has received increasing attention, as it can be widely used in numerous fields, such as protein prediction and community prediction. The GNN models for graph classification usually require aggregating the structure and feature information of an input graph into a hidden representation vector. Such a technique can be generally referred as graph pooling, which intends to maximize the removal of noise while minimize the loss of important features. Therefore, there exists a trade-off between the noise reduction and information maximization that needs to be balanced for better performance. However, the existing pooling-based GNNs for graph classification depends on the manual setting of the graph-pooling degree, making it difficult to balance the trade-off. To this end, we propose a Feature Pyramid-based Graph Convolutional Neural network for Graph Classification (FPGCN-GC), which constructs multi-scale hierarchical information fusion to reduce information loss, and achieves an adaptive feature fusion through a learnable weighted residual connection and self-attention mechanism. The superior performance of the proposed method is verified on multiple graph classification datasets, thus illustrating the effectiveness and superiority of FPGCN-GC. In addition, we visualize FPGCN-GC by T-SNE to further analyze the underlying reason for its effectiveness. [ABSTRACT FROM AUTHOR]