The Journal of Supercomputing最新文献

High-performance computing relies heavily on parallel and distributed systems, which promptes us to establish both qualitative and quantitative criteria to assess the fault tolerability and vulnerability of the system’s underlying interconnection networks. Consider the scenario in which large-scale link failures split the interconnection network into several components and each processor has multiple good neighboring processors. In this scenario, the fault tolerability of the system can be measured by g-good-neighbor r-component edge-connectivity, denoted by (lambda _{g,r}(G)), which is defined as the minimum number of edges whose removal results in a disconnected network with at least r connected components and each vertex has at least g good neighbors. It combines the strategies of g-good-neighbor edge-connectivity and component edge-connectivity. In this paper, the g-good-neighbor ((r+1))-component edge-connectivity of 3-ary n-cube is investigated. This work is the first attempt enhancing link fault tolerability for 3-ary n-cube under double constraints in the presence of the large-scale faulty links, which breaks down the inherent idea that poses one limitation on the resulting network. In addition, our results cover the work of Xu et al. (IEEE Trans Reliab, 71(3):1230–1240, 2022) and Li et al. (J Parallel Distrib Comput, 27:104886, 2024). Finally, the compared results reveal that the g-good-neighbor ((r+1))-component edge-connectivity is almost r times the size of g-good-neighbor edge-connectivity and much larger than ((r+1))-component edge-connectivity in 3-ary n-cube.

The exponential growth of IoT devices and the surge in the data volume, coupled with the rise of latency-sensitive applications, has led to a heightened interest in fog computing to meet user demands. In this context, the service provisioning problem consists of dynamically selecting desirable fog computing nodes and routing user traffic to these nodes. Given that the fog computing layer is composed of heterogeneous nodes, which vary in resource capacity, availability, and power sources, the service provisioning problem becomes challenging. Existing solutions, often using classical optimization approaches or heuristic algorithms due to the NP-hardness of the problem, have struggled to address the issue effectively, particularly in accounting for the heterogeneity of fog nodes and uncertainty of the ad hoc fog nodes. These techniques show exponential computation times and deal only with small network scales. To overcome these issues, we are motivated to replace these approaches with deep reinforcement learning (DRL) techniques, specifically employing the proximal policy optimization (PPO) algorithm to understand the dynamic behavior of the environment. The main objective of the proposed DRL-based dynamic service provisioning (DDSP) algorithm is minimizing service provisioning costs while considering service delay constraints, the uncertainty of ad hoc fog nodes, and the heterogeneity of both ad hoc and dedicated fog nodes. Extensive simulations demonstrate that our approach provides a near-optimal solution with high efficiency. Notably, our proposed algorithm selects more stable fog nodes for service provisioning and successfully minimizes cost even with uncertainty regarding ad hoc fog nodes, compared to heuristic algorithms.

The widespread adoption of utility-based real-time applications has placed the necessity of widescale deployment of edge computing infrastructure. Crowdsourced edge computing is deemed a suitable way out. On the other hand, a collection of today’s powerful smart mobile devices (SMDs) can cumulatively offer high-performance computing. The public-owned SMDs are utilized opportunistically to form a dynamic ad-hoc computing grid known as mobile crowd computing (MCC). This paper aspires to establish a proof-of-concept for the feasibility and use of MCC as a sustainable edge computing solution (MCC-edge). A typical smart HVAC system of an office building has been considered for the experiment case. We aim to process the HVAC data in real-time using the MCC-edge setup within the building for auto adjustment of the AC controller and error notifications. To maintain the ideal comfort level of the occupants, we present an extensive calculation using the dew point and heat index of the room. A high-level layered architecture of the MCC-edge for HVAC is presented along with a general framework of the MCC-edge. We report the module-wise design and implementation procedures with exhaustive details. The performance of MCC-edge is statistically compared with the commercial edge and cloud computing solutions in terms of cost, energy consumption, latency, and environmental impact, showing a significant advantage over the two. Every procedural detail of each module's design, development, and implementation is meticulously presented, which would aid interested readers and researchers in rebuilding such an application.

Camouflaged object detection (COD) aims to accurately segment camouflaged objects blending into the environment and is a challenging task. Most existing deep learning-based COD methods do not explicitly enhance the region information of camouflaged objects, nor do they use the region information for mask calibration. To solve this issue, we propose a novel mask stepwise calibration network (MSCNet) for camouflaged object detection, which achieves high-precision detection of camouflaged objects. Specifically, MSCNet consists of a region information enhancement encoder and a mask stepwise calibration decoder. In the encoder, we first utilize a PVT backbone network to extract different levels of features from camouflaged images. Then, we design a region information enhancement module to enhance the region information of camouflaged objects while suppressing the interference of background information by mining, embedding, and aggregating the region information in different levels of features. In the decoder, we first design a coarse mask generation module to generate coarse prediction masks of camouflaged objects by directly cross-fusing different levels of features extracted by the backbone. In addition, we also design a mask calibration module to calibrate coarse prediction masks of camouflaged objects using the region information of different levels of camouflaged objects as a guide. Extensive experimental results on four benchmark datasets show that our method effectively identifies camouflaged objects and surpasses most state-of-the-art COD methods.

Edge computing has become quintessential in commercial, healthcare, and industrial applications. It enables real-time data processing at the edge device, thus reducing the data traffic to the cloud and increasing the processing time efficiency. As an edge device, modern System-on-Chips (SoCs) provide scalability, security, and development in an integrated platform. Intellectual Property (IP) core reuse is a boon in SoCs that bridges the gap between integrated circuit design and fabrication. Such edge devices modeled by vendors are bound to ensure high security to avoid piracy. The proposed architecture provides a two-step authentication utilizing a Finite State Machine (FSM) with a secured key obtained from the newly structured Physical Unclonable Function (PUF) within the same edge device, with the primary goal of verifying several heterogeneous IPs to achieve the least power and energy. Two PUF designs, Anderson Arbiter PUF (AA-PUF) and Balanced AA-PUF, have been proposed for two different placements taking advantage of SoC-based architecture. The PUF characteristics have been experimentally validated with and without majority voting and demonstrate their proximity close to the desired value in ZedBoard. The proposed design is a strong PUF with less than 15% area overhead and power dissipation of 1.982 W for a 64-bit response. The experimental validation has evaluated that the power and energy consumptions are 2.56 W and 2.17 J for 52 heterogeneous IPs.

Traffic flow prediction plays a crucial role in intelligent transportation systems (ITS), offering applications across diverse domains. However, current deep learning models face significant challenges. Real-world traffic conditions, especially during peak hours, exhibit complex spatio-temporal dynamics and intricate nonlinear relationships. Existing studies often overlook variations in traffic flow across different time periods, locations, and scenarios, resulting in prediction models lacking robustness and accuracy across diverse contexts. Furthermore, simplistic models struggle to accurately forecast traffic flow during peak periods, as they typically focus on isolated features such as traffic speed, flow rate, or occupancy rate, neglecting crucial interdependencies with other relevant factors. This paper introduces a novel approach, the peak hour embedding-based multi-feature spatio-temporal coupled traffic flow prediction model (PE-MFSTC), to address these challenges. The PE-MFSTC model incorporates peak time embedding within a multirelational synchronization graph attention network structure. The peak time-based embedding involves mapping daily, weekly, and morning/evening peak periods into low-dimensional time representations, facilitating the extraction of nonlinear spatio-temporal features. The network framework employs a multirelational synchronized graph attention network, integrating multiple traffic features and spatio-temporal sequences for learning. Additionally, a spatio-temporal dynamic fusion module (STDFM) is introduced to model correlations and dynamically adjust node weights, enhancing the model’s sensitivity. Experimental evaluations on four real-world public datasets consistently demonstrate the superior performance of the PE-MFSTC model over seven state-of-the-art deep learning models. These results highlight the efficacy of the proposed model in addressing the complexities of traffic flow prediction across various scenarios.

Communities in a network are groups of nodes that are more strongly connected to each other. This article proposes a novel method for community detection in dynamic networks, focusing on influential nodes and overlapping communities. The method, named community detection based on adaptive multi-centrality aggregation (CDAMA), tackles two key challenges identifying influential nodes and overlapping communities. CDAMA introduces the Adaptive multi-centrality aggregation (AMCA) approach to identify influential nodes. AMCA integrates multiple centrality measures. The adaptive overlap control and merging (AOC-CM) approach addresses overlapping communities. AOC-CM utilizes structural, temporal, and semantic factors to strategically merge communities while preserving those with minimal overlap. CDAMA consists of five phases: receiving network snapshots, selecting influential nodes, launching communities, checking overlap and merging communities, and updating communities. Evaluation on three benchmark datasets demonstrates that CDAMA outperforms existing state-of-th-art methods in terms of Newman modularity, Modularity with split penalty and density modularity and Execution time. This suggests CDAMA is a valuable tool for tasks like viral marketing, information diffusion analysis, and network resilience studies.

In this paper, the approximate parallel prefix adders with minimizing hardware and timing complexities are proposed. Moreover, an approximate multiplier based on these adders is designed. The approximate structures include two approximate Sklansky adders, one approximate Ladner-Fischer adder, and one approximate Kogge-Stone adder. The proposed adders are free from carry rippling. The main strategy for approximate design is primarily based on rearranging and deleting sub-blocks and secondary reducing the critical path delay and area in the adders. In this case, we have a trade-off between accuracy, delay, and area. The proposed approximate multiplier has a serial structure that is designed based on using one approximate parallel prefix adder. The proposed approximate adders and multiplier are compared from hardware and accuracy point of view such as gate count, delay, area delay product, error rate, mean error distance, mean relative error distance, and normalized error distance. The efficacy of proposed structures in image processing applications such as image smoothing (low-pass filter) and image multiplication is performed using MATLAB. The results show the proposed approximate structures are comparable in terms of area, delay, PSNR, and mean structural similarity index metric parameters with other works.

Recognition of microglia activation state is required in the research of problems such as brain neurological diseases. In this paper, a novel recognition network based on YOLOv5 is proposed for microglia activation state recognition. Firstly, the decoupled head is integrated into the head network, and secondly, novel feature extraction modules containing DenseNet are introduced: the DenseNet-C2f module and the DenseNet-SimCSPSPPF module. Subsequently, Wise-IoU is employed as the loss function, and the parameters therein are discussed. The network performance was evaluated using the microglia dataset. The experimental results show that the average precision of the enhanced network increases from 59.6 to 65.6%. In addition, the recall was improved from 56.3 to 71.5%. These improvements resulted in more efficient detection performance, which better meets the requirements of the medical field for identifying microglia activation states.

Wireless rechargeable sensor network (WRSN) is one of the most important networks in today’s world for extending network lifetime. To ensure a collision-free contention in WRSN, this paper proposes the Contention Window Optimization using the Cuttlefish Algorithm (CWO-CA) and addresses existing issues such as the capture effect, the fairness problem, and the queue hike. For safe channel access, the fitness function of CWO-CA is proposed using the retransmission count, queue size, and residual energy of the sensor node. The proposed algorithm divides the Contention Window interval into two halves based on the fitness function, and the nodes are assigned in either half to guarantee service differentiation. The fitness function among groups of the proposed algorithm constitutes the best fitness value. By selecting the best fitness value, the proposed algorithm facilitates adaptive contention management by dynamically balancing resource utilization of the network with the optimal result. The nodal behavior of CWO-CA has been modeled using a discrete Markov model and also simulated to measure throughput, packet delivery ratio, packet loss ratio, average queue size, residual energy, and delay. These results confirm that the proposed CWO-CA algorithm attained better performance than other existing algorithms.