Search

In mobile edge computing environment, intelligent inference services driven by DNN are highly sensitive to latency. Recently, collaborative inference between User Devices and Edge Servers (ESs) based on Deep Neural Networks (DNN) partition has achieved success in service acceleration. However, most of the existing collaborative acceleration schemes are partitioned for a single DNN inference task, which cannot quickly make partition decisions for a set of concurrent inference tasks, and often sacrifice inference accuracy. In addition, due to the limited resources of ESs, there is resource competition among concurrent requests, which makes the partitioned tasks cannot be offloaded to ESs in time for processing. Therefore, designing an efficient offloading scheme becomes essential. The task offloading schemes based on deep reinforcement learning can solve complex decision-making problems in high-dimensional state space, but they have problems such as insufficient sample diversity and easily falling into local optimum. In this paper, a Collaborative Inference Acceleration Scheme integrating DNN Partitioning and Task Offloading (CIAS-PnO) is proposed. First, while ensuring inference accuracy, the Collaborative DNN Layer Partitioning (CDLP) algorithm is designed with the goal of optimal latency. CDLP can reduce the problem scale of concurrent inference tasks partition by pruning operation and determine the partition decisions in time. Then, the Distributed Soft Actor-Critic (SAC)-based Partition Task Offloading algorithm (DSACO) is designed. DSACO supports SAC Agents to explore samples in parallel and share learning experiences, and uses the automatic entropy adjustment mechanism to improve the exploration efficiency of Agents, so as to avoid falling into local optimum and achieve efficient offloading of partition tasks. Experimental results on DNN benchmarks show that compared with the baseline acceleration schemes, CIAS-PnO achieves more than 19.8% acceleration performance improvement, and has higher convergence performance and task success rate. [ABSTRACT FROM AUTHOR]

Erasure-code-based storage systems suffer from problems such as long repair time and low I/O performance, resulting in high repair costs. For many years, researchers have focused on reducing the cost of repairing erasure-code-based storage systems. In this study, we discuss the demerits of node selecting, data transferring and data repair in erasure-code-based storage systems. Based on the network topology and node structure, we propose SDNC-Repair, a cooperative data repair strategy based on erasure code for SDS (Software Defined Storage), and describe its framework. Then, we propose a data source selection algorithm that senses the available network bandwidth between nodes and a data flow scheduling algorithm in SDNC-Repair. Additionally, we propose a data repair method based on node collaboration and data aggregation. Experiments illustrate that the proposed method has better repair performance under different data granularities. Compared to the conventional repair method, although the SDNC-Repair is more constrained by the cross-rack bandwidth, it improves system throughput effectively and significantly reduces data repair time in scenarios where multiple nodes fail and bandwidth is limited. [ABSTRACT FROM AUTHOR]

Motivation K -mers along with their frequency have served as an elementary building block for error correction, repeat detection, multiple sequence alignment, genome assembly, etc. attracting intensive studies in k -mer counting. However, the output of k -mer counters itself is large; very often, it is too large to fit into main memory, leading to highly narrowed usability. Results We introduce a novel idea of encoding k -mers as well as their frequency, achieving good memory saving and retrieval efficiency. Specifically, we propose a Bloom filter-like data structure to encode counted k -mers by coupled-bit arrays—one for k -mer representation and the other for frequency encoding. Experiments on five real datasets show that the average memory-saving ratio on all 31-mers is as high as 13.81 as compared with raw input, with 7 hash functions. At the same time, the retrieval time complexity is well controlled (effectively constant), and the false-positive rate is decreased by two orders of magnitude. Availability and implementation The source codes of our algorithm are available at github.com/lzhLab/kmcEx. Supplementary information Supplementary data are available at Bioinformatics online. [ABSTRACT FROM AUTHOR]

The article discusses the study on the server side SOAP Processing approach. It states that the approach is aimed at improving the performance of web services. It mentions that the web services not only acquire XML's advantages but likewise its poor performance, thus makes it an unsatisfactory choice in a number of high- performance applications. It notes that various cache implementations are embedded in SOAP processor and likewise employ an adaptive cache mechanism. It suggests that the approach can accomplish a huge gain in performance by integrating the SOAP Processor to the SOAP engine.

As the requirement for real-time data analysis increases, edge computing is being implemented to leverage the resources of edge devices to reduce system response times and decrease the latency. However, due to the resource constraints of edge clouds, edge servers are more prone to failures than other systems. Therefore, guaranteeing the reliability of services in edge clouds is critical. In this paper, we propose a fault-tolerant adaptive scheduling mechanism with dynamic quality of service (QoS) awareness (FASDQ), which extends the primary/backup (PB) model by applying QoS on demand to task copies. The aim of the method is to reduce the latency and achieve reliable service for tasks by changing the execution time of task copies. This paper also proposes a container resource-adaptive adjustment mechanism, which adjusts the timing of resources when the available resources cannot meet the task copy requirements. Finally, this paper reports the results of simulation experiments on the EdgeCloudSim platform to evaluate the difference in performance between FASDQ and other methods. The results show that the mechanism effectively reduces the execution time of task copies and outperforms other methods in terms of reliability and general resource utilization. [ABSTRACT FROM AUTHOR]