1 / 23

Distributed Delay Estimation and Call Admission Control in IEEE 802.11 WLANs

Distributed Delay Estimation and Call Admission Control in IEEE 802.11 WLANs. Kenta Yasukawa, Andrea G. Forte and Henning Schulzrinne Ericsson Research Japan, Department of Computer Science Columbia University IEEE ICC 2009 proceedings. 報告者:李宗穎. Outline. Introduction

Télécharger la présentation

Distributed Delay Estimation and Call Admission Control in IEEE 802.11 WLANs

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. Distributed Delay Estimation and Call Admission Control in IEEE 802.11 WLANs Kenta Yasukawa, Andrea G. Forte and Henning Schulzrinne Ericsson Research Japan, Department of Computer Science Columbia University IEEE ICC 2009 proceedings 報告者:李宗穎

  2. Outline • Introduction • Delay Estimation using Time between Idle Time • Call Admission Control using TBIT • Evaluation • Conclusion

  3. Introduction • After the number of concurrent voice calls in the BSS surpasses the AP capacity • high delay, high collision rate • even worse in scenarios where both voice traffic and data traffic

  4. Mobile-Station-Based CAC • NOT requiring any probing of the medium • any STA in the BSS can monitor the shared medium and calculate the TBIT • NOT requiring any changes in the infrastructure and the protocol (client-based) • deployment feasibility ( it is not realistic to assume a new CAC mechanism will be widely deployed on existing APs )

  5. Time Between Idle Time (TBIT) • “idle time” as a period during which the medium is idle and long enough to represent a transmission opportunity • if it takes “too long” for the AP/STA to empty its queue, it means that the packets stay in the queue for a longer time • the time between idle times gives us a direct estimate of the level of congestion

  6. Delay Estimation using TBIT • the wireless medium is shared and every STA can hear packets that other STAs and the AP send and receive (STAs can still “hear” ACK in hidden nodes) The queuing delay for the new packet is t2− t1 and it is maximized when t0 = t1 The TBIT is equal to t2− t0 and represents a direct measure of the maximum queuing delay at the AP at the time of the observation

  7. Idle Time Threshold Ith • if the AP has more packets to send, it has to wait for DCF inter frame space (DIFS) plus the time needed to decrement a random backoff counter to zero Expected Backoff Time

  8. Expected Backoff Time (1/2) • The Ith second term is the expected backoff time, paper defines it as : • We consider N equal to its upper bound, that is, equal to CW and we set CW to its lower bound, that is, to CWmin.

  9. Expected Backoff Time (2/2) • CW is equal to CWmin for more than 95% of the time, and paper selected (N = CW = CWmin) • Ithis big enough not to count as transmission opportunities those idle periods due to backoff, and small enough not to make our method too conservative

  10. CAC using TBIT • Let P and λ be its packet size and packet rate, respectively, and let Ttx(P) be the time needed to send a packet whose size is P • if we denote the frequency of idle times that are longer than Ttx(P) by μ, we can say that the new call will not cause congestion if μ is larger than λ

  11. Obviously the inverse of TBIT • Paper can inverse of TBIT and is obtained in the same way as in the delay estimation. The only change we need to apply here is to set the idle time threshold as Ttx(P), that is:

  12. How to get P and λ • both caller and callee usually negotiate which codec to be used in the call setup phase • SIP is used codecs are negotiated in the initial INVITE−200 OK handshake • By substituting the codec information obtained through the negotiation, it is possible for a STA to check if λ is larger than μ

  13. Common Network Setup Network Simulator 2 (ns-2) IEEE 802.11B

  14. Implementation and Testbed • The TBIT listener was based on Java and libpcap version 0.9.8 • Paper used three T42 and one R51 IBM Thinkpad laptops to make the same topology as Fig. 3 (Linux OS)

  15. Delay Estimation Experiments • The PC connected to the AP by Ethernet sent packets with time-stamps to the TBIT listener and the TBIT listener plotted the actual one-way delay • The other laptop PC was used to generate background traffic so to increase the queuing delay at the AP • The clocks of all the PCs were synchronized by using the NTP every second (the clocks to much less than 1ms)

  16. Delay estimation using the Java-based TBIT listener

  17. Limitation of TBIT listener • The limitation of our current implementation is that, TBIT is computed by capturing packets, and packets can be lost due to errors and collisions • This problem can be avoided by accessing the status of the medium directly from the wireless card (Such implementation will be reserved for future study)

  18. CAC Experiments • Paper emulated G.711 CBR VoIP calls by creating a simple program which sends out packets with a given constant interval • The idle time threshold Ithwas set to 1650μs according to TBIT equation

  19. CAC using TBIT for 802.11b (G.711 CBR at 2 Mb/s)

  20. Simulation Study • Paper did simulations assuming both homogeneous and heterogeneous scenarios, • only one kind of VoIP codec is used • different kinds of VoIP codecs are mixed together

  21. VoIP capacity and frequency of idle times (G.711 VBR)

  22. Contour of 90thpercentile delays in ms and CAC decisions made using TBIT (G.711 VBR + G.723.1 CBR)

  23. Conclusions • This paper introduce the concept of time between idle times (TBIT) and show how this can be used to estimate queuing delay and make CAC decisions in IEEE 802.11 networks

More Related