1 / 17

An Integrated Load Balancing Scheme for Future Wireless Networks

An Integrated Load Balancing Scheme for Future Wireless Networks. Instructor: 陳仁暉 Student: 連挺鈞 Wireless Pervasive Computing, 2009 ISWPC GLOBECOM Workshops, 2008 IEEE. Outline. Introduction Related Work Contribution Proposed Scheme Scheme Propagation Scheme Network Flow Simulation

gari
Télécharger la présentation

An Integrated Load Balancing Scheme for Future Wireless Networks

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. An Integrated Load Balancing Scheme for Future Wireless Networks Instructor: 陳仁暉 Student: 連挺鈞 Wireless Pervasive Computing, 2009 ISWPC GLOBECOM Workshops, 2008 IEEE

  2. Outline • Introduction • Related Work • Contribution • Proposed Scheme • Scheme Propagation • Scheme Network Flow • Simulation • Result • Conclusion

  3. Introduction • Focus on delay sensitive traffic • Excessive handover latency and packet loss • Overloading AP suffers from high delay • Can’t guarantee real-time services under heavy loads • Dynamic network • Suffers high packet delay and packet loss • Goal • Throughput and QoS fairness

  4. Related Work (1/2) • Metrics • Num. of active connections, gross load, packet loss and throughput • Circuit-Switched • Bianchi et al [7] • WLAN can be improved by using additional packet level • Bazzi et al [8] • CAC denying incoming calls when low resources • Disadvantage • Not adaptive to dynamic network condition

  5. Related Work (2/2) • Centralized (Balachandran et al. [9]) • Centralized admission control servers • Load information of all APs • Disadvantage • Increase signaling overhead • Creates bottleneck • Prone to single point of failure • Decentralized (Velayos et al. [10]) • Using throughput per AP as load metric • Disadvantage • Can only reassociate with underloaded AP

  6. Contribution • Guarantee service QoS during handover • Enabling seamless handover • Operate network in unsaturated mode with soft admission control • Select best target network • Network-assisted discovery compatibility with IEEE 802.21 media independent handover infrastructure • Single transceiver stations, horizontal and vertical handovers

  7. Propose Scheme (1/2) • Bootstrap approximation • Est. short-term stationary dynamic QoS parameters in AP • Sequential Bayesian estimation • Estimate probability density function • Station select best AP according to delay est. • Obviate detection and scanning • Latency reduction • Protects existing connection • Soft admission control • Important for multimedia traffic

  8. Propose Scheme (2/2) • Shaded block • Network entry • Unshaded block • Terminal entry • Trigger • Initial access • Choose best network according to packet delay requirement • Handover • Packet loss rate exceeds 2% for VoIP • Soft admission control monitors source and target AP • Stability period: 10 beacon intervals

  9. Scheme Propagation • Measure report • Access point controller (APC) • Using location based broadcast • Monitor source AP • Advantage • Monitoring AP w/o APs exchanging information • Eliminate scanning • Consists of two-way handshake • Soft admission control: 1ms • Avg. channel switch time: 12 ms • Authentication delay: < 1ms • Avg. reassociation delay: 15.37ms

  10. Scheme Network Flow

  11. Simulation (Topology) • Basic settings • Codec delay: 40ms • Packetization delay: 20ms • Backbone network delay: 30ms • Wireless network delay: < 60ms • 802.11b(11Mbps), 802.11g(54Mbps) • At least 1 legacy link to 11g • OPNET Modeler 14.0 with Wireless Module • No hidden terminal • MSDU life time: voice(50ms), video(100ms), data(1s) • No mobility

  12. Simulation (Process) • Initial (Unbalanced load) • 2 FTPs, 2 videos • 7 G.711 stations in BS1 • 7 G.711 stations in BS2 • 900s • Additional 1 FTP, 1 video • 5 G.711 from BS1 stopped • 5 G.711 from BS2 start

  13. Result (Settings) • Examine QoS performance • Delay and packet loss of AP • Quantify effect of load balancing • xi: total throughput or delay of access point I • n: number of access points after redistribution • Balanced index • 0: APs extremely unbalanced, n  inf • 1: APs same throughput or delay, 1/n

  14. Result (Packet loss rate) • iLB: integrated load balancing • 802.11b/g: DCF • 802.11e: EDCA • Focus on avg. downlink delay and packet loss rate

  15. Result (Throughput and delay) • Balance index • Throughput • DCF and EDCA: 0.86 • iLB: 0.96 • Delay • DCF and EDCA: 0.56, 0.58 • iLB: 0.81

  16. Conclusion (1/2) • This paper provides a thorough investigation • System model, propagation, network flow • Simulation on known model (OPNET) • Simulation model • Mathematical approach • Focus on one topic • Delay and packet loss over vertical handover • Specific on the system settings

  17. Conclusion (2/2) • Disadvantage • Still have lots of overhead from APC • No comparison with other load balancing scheme • No robust data flow • Similarity • APC as to ABS • APs as to femto BS • Difference • In current scheme all MSs are connected to AP • Difference CAC design • Different focus (MAC overhead) • Only horizontal handover

More Related