Traffic measurement is key to many important network functions. Supporting real-time queries at the individual flow level over networkwide traffic represents a major challenge that has not been successfully addressed ...
详细信息
ISBN:
(数字)9781665471770
ISBN:
(纸本)9781665471770
Traffic measurement is key to many important network functions. Supporting real-time queries at the individual flow level over networkwide traffic represents a major challenge that has not been successfully addressed yet. This paper provides the first solutions in supporting real-time networkwide queries and allowing a local network function (for performance, security or management purpose) to make queries at any measurement point at any time on any flow's networkwide statistics, while the packets of the flow may traverse different paths in the network, some of which may not come across the point where the query is made. Our trace-based experiments demonstrate that the proposed solutions significantly outperform the baseline solutions derived from the existing techniques.
Mobility management is crucial for urban planning, and accurate road condition information is necessary for safe and efficient transportation. In the context of this study, mobility refers to the quality of the road a...
详细信息
ISBN:
(纸本)9798350333398
Mobility management is crucial for urban planning, and accurate road condition information is necessary for safe and efficient transportation. In the context of this study, mobility refers to the quality of the road and the level of service provided to road users. To this end and to advance the state-of-the-art in improving mobility management strategies, we propose a new capability for our ongoing work on the Decentralized Road Traffic Monitoring (DRTM) system: road conditions assessment, aimed at enhancing mobility management for Abu Dhabi city. Our DRTM system uses smartphone cameras, Vehicular Ad-hoc Networks (VANETs) on vehicle-to-vehicle (V2V) mode, and federated learning to extract and classify road features, such as potholes, cracks, and accidents, impacting overall mobility. The system comprises a smartphone app and a decentralized network of participants, providing real-time insights about road conditions. Our experimental results demonstrate the system's effectiveness in providing precise and up-to-date information about road conditions, with increased scalability and efficiency compared to centralized methods. By utilizing federated learning, the system ensures data security and privacy.
The particle Markov-chain Monte Carlo (PMCMC) method is a stochastic algorithm that combines Particle Filters (PFs) and Markov-chain Monte Carlo (MCMC) techniques. This approach is widely used in Bayesian inference fo...
详细信息
ISBN:
(纸本)9798350350920
The particle Markov-chain Monte Carlo (PMCMC) method is a stochastic algorithm that combines Particle Filters (PFs) and Markov-chain Monte Carlo (MCMC) techniques. This approach is widely used in Bayesian inference for high-dimensional state spaces and nonlinear, non-Gaussian dynamic systems. However, current PMCMC accelerators face significant challenges due to their intensive computational complexity and the intricate particle routing, limiting their application in real-time scenarios. To address these challenges, we propose a novel distributed PMCMC method that leverages parallel computing to enhance hardware execution speed. Additionally, our method introduces a particle exchange scheme that not only resolves the accuracy issues caused by particle routing in distributed PMCMC but also achieves faster computing speed. Our design is implemented on a Xilinx Kintex-7 xc7k480t FPGA device. Experimental results demonstrate that our accelerator is nearly 65x faster than CPU performance, and provides speedups up to 5x compared to existing FPGA-based accelerators.
Device-free sensing with wifi signals has gained great attention from the community to enhance building automation systems. In such systems, energy efficiency can be improved through predictive forecasting of occupanc...
详细信息
Electricity load calculation of high-speed train group is one critical step for traction power system design. The energy-saving oriented electricity load calculation of high-speed train group is time-consuming. A dist...
详细信息
In this paper, develop a power control scheme for underwater wireless sensor networks that takes into account the impact of communication ranges on network performance, including energy utilization and delay overhead....
详细信息
In recent years, distributedcomputing has wit-nessed widespread applications across numerous organizations. Predicting workload and computing resource data can facilitate proactive service operation management, leadi...
详细信息
Efficient load balancing is essential for distributedsystems to have outstanding performance and scalability. This study compares the load balancing strategies of two popular distributed application frameworks, gRPC ...
详细信息
This paper examines the overall performance of autonomous and adaptive communications systems in Wi-Fi sensor networks. Self-sustaining communications systems are those capable of adjusting their radio configurations ...
详细信息
Community detection is a fundamental operation in graph mining, and by uncovering hidden structures and patterns within complex systems it helps solve fundamental problems pertaining to social networks, such as inform...
详细信息
ISBN:
(纸本)9798350364613;9798350364606
Community detection is a fundamental operation in graph mining, and by uncovering hidden structures and patterns within complex systems it helps solve fundamental problems pertaining to social networks, such as information diffusion, epidemics, and recommender systems. Scaling graph algorithms for massive networks becomes challenging on modern distributed-memory multi-GPU (Graphics Processing Unit) systems due to limitations such as irregular memory access patterns, load imbalances, higher communication-computation ratios, and cross-platform support. We present a novel algorithm HiPDPL-GPU (distributed Parallel Louvain) to address these challenges. We conduct experiments involving different partitioning techniques to achieve an optimized performance of HiPDPL-GPU on the two largest supercomputers: Frontier and Summit. Remarkably, HiPDPL-GPU processes a graph with 4.2 billion edges in less than 3 minutes using 1024 GPUs. Qualitatively, the performance of HiPDPL-GPU is similar or better compared to other state-of-the-art CPU- and GPU-based implementations. While prior GPU implementations have predominantly employed CUDA, our first-of-its-kind implementation for community detection is cross-platform, accommodating both AMD and NVIDIA GPUs.
暂无评论