A5009003415- Cloud Computing and Resource Management
- Software-Defined Networks and 5G
- Network Traffic and Congestion Control
- Interconnection Networks and Systems
- Data Stream Mining Techniques
- Peer-to-Peer Network Technologies
- Video Coding and Compression Technologies
- Image and Video Quality Assessment
- Caching and Content Delivery
- IoT and Edge/Fog Computing
- Energy Efficient Wireless Sensor Networks
Hainan University
2023-2024
Central South University
2019-2022
Changsha University
2021-2022
Modern data center networks provide multiple paths between host pairs to guarantee high aggregated network bandwidth and transmission reliability. However, suffer from various uncertainties such as highly dynamic traffic heterogeneous devices. These easily lead asymmetry cause significant packet reordering. Unfortunately, due lacking the explicit reordering feedback, existing sender-based receiver-based solutions hardly adapt asymmetric long-tailed flow completion time well throughput loss....
To improve the quality of experience for worldwide users, an increasing number service providers deploy their services on geographically dispersed data centers, which are connected by wide area network (WAN). In cross-datacenter networks, however, intra- and inter-datacenter parts have different characteristics, including switch buffer depth, round-trip time bandwidth. Besides, most intra-DC flows belong to interactive that require low delay while inter-DC typically need achieve high...
Abstract Modern datacenters provide a wide variety of application services, which generate mix delay-sensitive short flows and throughput-oriented long flows, transmitting in the multi-path datacenter network. Though existing load balancing designs successfully make full use available parallel paths attain high bisection network bandwidth, they reroute regardless their dissimilar performance requirements. The suffer from problems large queuing delay packet reordering, while fail to obtain...
Abstract Modern data center topologies often take the form of a multi-rooted tree with rich parallel paths to provide high bandwidth. However, various path diversities caused by traffic dynamics, link failures and heterogeneous switching equipment widely exist in production datacenter network. Therefore, multi-path load balancer should be robust these diversities. Although prior fine-grained schemes such as RPS Presto make best use available paths, they are prone experiencing packet...
Abstract Modern data center networks typically adopt multi-rooted tree topologies such leaf-spine and fat-tree to provide high bisection bandwidth. Load balancing is critical achieve low latency throughput. Although the per-packet schemes as Random Packet Spraying (RPS) can network utilization near-optimal tail in symmetric topologies, they are prone cause significant packet reordering degrade performance. Moreover, some coding-based proposed alleviate problem of loss. Unfortunately, these...
To deliver high quality video across different network conditions, the chunks are explicitly fetched by client or proactively pushed server in Dynamic Adaptive Streaming over HTTP (DASH) system. Unfortunately, on one hand, fetch mechanism suffers from bandwidth wastage due to its stop-and-wait fashion when delay becomes large. On other push performs poorly because of inflexibility bitrate switching under fluctuating bandwidth. address these inefficiencies, we propose a pipeline-based chunk...
Abstract Modern datacenters provide a wide variety of application services, which generate mix delay-sensitive short flows and throughput-oriented long flows, transmitting in the multi-path datacenter network. Though existing load balancing designs successfully make full use available parallel paths attain high bisection network bandwidth, they reroute regardless their dissimilar performance requirements. The suffer from problems large queuing delay packet reordering, while fail to obtain...
In the data center environment, meeting dissimilar requirements of transmission performance for long and short flows is crucial guaranteeing service quality. It well known that optimizing network key. Abundant research studies have reported designing an effective load-balancing scheme can boost by enhancing path utilization multirooted networks. Unfortunately, existing schemes fail to provide satisfying flow-level since they choose ignore constant changing status blindly assign paths...
Abstract Modern data centers have widely deployed lots of cluster computing applications such as MapReduce and Spark. Since the coflow/task abstraction can exactly express requirements applications, various task-based solutions been proposed to improve application-level performance. However, most require modification obtain task information, making them impractical in many scenarios. In this paper, we propose a Bayesian decision-based Task Prediction mechanism named BTP identify predict...