A QoS-Enabled VPN Architecture Supporting Real-Time Micro-Flow Bandwidth Guarantee

• Main Authors: Wei-Nung Lee, 李偉農
• Other Authors: Yau-Hwang Kuo
• Format: Others
• Language: en_US
• Published: 2003
• Online Access: http://ndltd.ncl.edu.tw/handle/32745150703388047988

碩士 === 國立成功大學 === 資訊工程學系碩博士班 === 91 === Virtual private network (VPN) techniques emulate leased lines over Internet are frequently used to secure communications between remote sites. However, there is a demand to guarantee quality of service (QoS) related to VPN. In this thesis, we propose a QoS-enabled VPN architecture with a new service level agreement (SLA) approach. Within this architecture, a customer can specify his QoS requirement in detail such as VPN total bandwidth and real-time micro-flow bandwidth requirements in SLA. Our QoS-enabled VPN architecture is based on VPN hose model over differentiated services (DiffServ) architecture. It logically partition capacity at edge routers to various classes of (real-time and non real-time) micro-flow aggregations of VPNs. First, micro-flows’ state information is dealt at admission control module in edge routers in order to provide fine granularity of QoS control. Second, a dynamic bandwidth prediction technique, normalized least mean square error (NLMS) linear predictor, is used to share resources efficiently between VPN real-time service classes and VPN best-effort service class within each VPN. Our architecture put the responsibility of managing DiffServ domain’s network resources in a bandwidth broker (BB). Simulation networks are set up by network simulator 2 (NS-2) and performances of proposed QoS-enabled VPN architecture, CE-based and current provider provisioned VPN architectures have been evaluated according to five metrics.