1 / 20

Yaxin Cao, Ka-Cheong Leung, Member, IEEE, and Victor O. K. Li, Fellow, IEEE

Bandwidth-Guaranteed Fair Scheduling with Effective Excess Bandwidth Allocation for Wireless Networks. Yaxin Cao, Ka-Cheong Leung, Member, IEEE, and Victor O. K. Li, Fellow, IEEE IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS, Vol. 7, No. 6, JUNE 2008 報告者：李宗穎. Outline. Introduction

jiro
Download Presentation

Yaxin Cao, Ka-Cheong Leung, Member, IEEE, and Victor O. K. Li, Fellow, IEEE

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. Bandwidth-Guaranteed Fair Scheduling with Effective Excess Bandwidth Allocation for Wireless Networks Yaxin Cao, Ka-Cheong Leung, Member, IEEE, and Victor O. K. Li, Fellow, IEEE IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS, Vol. 7, No. 6, JUNE 2008 報告者：李宗穎

  2. Outline • Introduction • System Model • Proposed Algorithm : BGFS-EBA • Performance Evaluation • Conclusion

  3. Problem • Packet transmission in wireline networks enjoy very low error rate, but wireless channels are more error-prone and suffer from interference, fading, and shadowing.

  4. Related Work • channel-condition independent packet fair queueing (CIFQ) • error-free fair queueing reference system • associating compensation rate and penalty rate with a flow’s allocated service rate and guaranteeing any flow sent on an error-free link with the minimal service rate

  5. System Model • uplink and downlink flows are scheduled independently • centralized scheduling • the quality of an error-prone link between two states, good or bad • In bad states, the packet can be split into m low-rate packets

  6. The scope of paper work • Provide long-term fairness and goodput guarantees for flows with error-free links or occasional link errors • Achieve high wireless channel utilization • Minimize packet loss • Provide delay bound for flows with error-free links or sporadic link errors • Achieve low power consumption in mobile hosts • Achieve medium algorithm complexity

  7. Algorithm Description • Flow state • leading, lagging, or satisfied • CIF-Q • error-free start-time fair queueing (SFQ) system, which is dependent on the traffic load • BGFS-EBA • load-independent function ri t, for all i = 1, 2, . . .,N

  8. Scheduling process phase 1 The scheduler always selects a virtual packet with the smallest deadline Deadline calculation of a virtual flow

  9. Scheduling process phase 2 • L and riare the packet size and the target rate for flow i • the scheduler only needs to maintain one deadline for each flow i, di, by which the HOL virtual packet should be served

  10. Normalized goodput gap • To determine how much Flow i is leading and lagging its target rate • normalized goodput gap, which is defined as gi(t) : data of flow i transmitted successfully up to time t ri : the flow i goodput Gi(t) : represents how much a flow is leading and lagging compared to its target goodput

  11. Scheduling operation for BGFS-EBA (1/2) • The packet will give up its service opportunity to some other flow if all of the following conditions are satisfied : • Flow i is leading, i.e. Gi > 0 • gi+ L ≥ ri・ (di − τi), where diis the current deadline (after an update in the first phase) of a virtual packet for flow i, and τiis the start time of the current backlog period for flow i • There exists at least one flow lagging and goodstate

  12. Scheduling operation for BGFS-EBA (2/2) • virtual flow is picked but its real packet queue is empty, paper will search other non-empty flow with following conditions : • Any flow with a link in the good state and its normalized goodput gap below its goodput threshold • Any flow with a link in the good state and a negative normalized goodput gap • Any flow with its normalized goodput gap below its goodput threshold • Any flow with a link in the good state • Any backlogged flow

  13. Multi-rate Time-Varying Channel • BGFS-EBA can be easily extended to handle traffic flows transmitted over a multi-rate wireless communication channel • R = R1> R2> ・ ・ ・ > RC > 0 • Any flow, Flow j, with S(j) ≤ c and a negative normalized goodput gap, where c ≤ C is a system parameter for the generalized algorithm • Any flow, Flow j, with S(j) > c and its normalized goodput gap below its goodput threshold • Any backlogged flow

  14. Simulation Setup

  15. Minimum Transmission Bandwidth and GoodputGuarantees • Flow 1 : Good/Bad state 0.01s/0.09s • Flow 2 : Good/Bad state 0.09s/0.01s

  16. Outcome Fairness Flow 1,3,4 : 200kbps Flow 2,5 : 100kbps

  17. Distribution of Excess Bandwidth Flow 1,3,4 : 200kbps Flow 2 : 100kbps

  18. Packet Delay

  19. Discussion on Goodput Threshold • it would be advantageous to set a higher goodput threshold for a flow with a better link • high-priority flow, its goodput threshold can be set to zero or close to zero • goodput threshold may also be changed dynamically according to the link quality and traffic load

  20. Conclusions • To strike a balance between effort-fair and outcome-fair, BGFS-EBA permits a flow leading its target goodput to give up its current service opportunity to a lagging flow

More Related