Presentation is loading. Please wait.

Presentation is loading. Please wait.

Doc.: IEEE 802.11-07/2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 1 Performance Evaluation of Express Forwarding for a Single-Channel.

Similar presentations


Presentation on theme: "Doc.: IEEE 802.11-07/2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 1 Performance Evaluation of Express Forwarding for a Single-Channel."— Presentation transcript:

1 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 1 Performance Evaluation of Express Forwarding for a Single-Channel Mesh Notice: This document has been prepared to assist IEEE It is offered as a basis for discussion and is not binding on the contributing individual(s) or organization(s). The material in this document is subject to change in form and content after further study. The contributor(s) reserve(s) the right to add, amend or withdraw material contained herein. Release: The contributor grants a free, irrevocable license to the IEEE to incorporate material contained in this contribution, and any modifications thereof, in the creation of an IEEE Standards publication; to copyright in the IEEEs name any IEEE Standards publication even though it may include portions of this contribution; and at the IEEEs sole discretion to permit others to reproduce in whole or in part the resulting IEEE Standards publication. The contributor also acknowledges and accepts that this contribution may be made public by IEEE Patent Policy and Procedures: The contributor is familiar with the IEEE 802 Patent Policy and Procedures, including the statement "IEEE standards may include the known use of patent(s), including patent applications, provided the IEEE receives assurance from the patent holder or applicant with respect to patents essential for compliance with both mandatory and optional portions of the standard." Early disclosure to the Working Group of patent information that might be relevant to the standard is essential to reduce the possibility for delays in the development process and increase the likelihood that the draft publication will be approved for publication. Please notify the Chair as early as possible, in written or electronic form, if patented technology (or technology under patent application) might be incorporated into a draft standard being developed within the IEEE Working Group. If you have questions, contact the IEEE Patent Committee Administrator at.http:// Date: Authors: NameAddressCompanyPhone Mathilde Benveniste 233 Mt Airy Road Basking Ridge, NJ 07920, US Avaya Labs- Research org Kaustubh Sinkar 233 Mt Airy Road Basking Ridge, NJ 07920, US Avaya Labs- Research m

2 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 2 Performance Evaluation of Express Forwarding for a Single-Channel Mesh Mathilde Benveniste Kaustubh Sinkar Avaya Labs - Research

3 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 3 Introduction VoIP cannot meet QoS requirements on a wireless mesh unless there is a way to reduce delay/jitter –End-to-end delay and jitter can be too high in a single-channel mesh because of multi-hop transmissions Delay/jitter determines the delay experienced by the end-user receiving QoS traffic –Frames are kept in a jitter buffer on receiving device for smooth delivery Reducing the worst-case delay causes all frames of a QoS traffic stream to experience lower delay –A shorter jitter buffer is needed This presentation shows the performance of Express Forwarding and Express Retransmission –These optional features help multi-hop QoS traffic get through the mesh fast with minimum impact on other traffic

4 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 4 Delay Budget Recommended one-way total delay (ITU G.114) – 150 ms Delay introduced in IP network and end system – 110 ms –IP network delay – sum of transmission and queuing delays traveling thru IP network (~50 ms) –End-system delay – sum of the encoding (20 ms), decoding (small), jitter buffer (~40 ms), and other data handling delays The target for maximum latency in a wireless mesh should be 40 ms* * In TGe, a target of 10 ms was used for WLAN delay in top-priority ACs

5 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 5 Express Forwarding – Review Ref Doc 11-07/2452, 2453 Express forwarding reduces the end-to-end delay of selected frames by granting forwarding nodes immediate access to the channel Criteria for frames to qualify for express forwarding are: –Time sensitive QoS [TSQ] frames – e.g. VO/VI –Frames on paths traversing more than a specified number of hops –Other Time critical frames – do not yield priority to express forwarded frames; such frames are –Top priority management frames –Top priority frames experiencing longer delay than a specified limit

6 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 6 Express Forwarding Illustration Express Forwarding Illustration Ref Doc 11-07/2452, 2453 The Duration field is set at a value longer than usual when a TSQ frame is transmitted to a forwarding node of a multi-hop path; DT0 added The forwarding nodes, 2 and 3, adjust the Duration value on the received frame by subtracting an increment DTI when setting their NAV The non-forwarding neighbor nodes (e.g. 5) sets NAV by Duration field NAV setting at all other neighbor nodes NAV setting at receiving node Channel time ACK Value in Duration field Frame DT0 3-hop path DTI ANIMATED

7 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 7 Express Retransmission Express Retransmission Ref Doc 11-07/2452, 2453 Retransmission typically involves backoff using a wider contention window With express retransmission, a frame is retransmitted by dispensing with backoff and transmitting within DT0 following ACKtimeout An express forwarded frame is less likely to collide on retransmission because of its prioritization Only the first retransmission attempt receives priority treatment Prevents two express forwarded frames from colliding repeatedly Channel time ACKtimeout TSQ Frame DT0 DTI TSQ Retransmission NAV setting at all other neighbor nodes NAV setting at receiving node Value in Duration field of TSQ frame

8 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 8 Performance Evaluation

9 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 9 Scenarios b Mesh (4 hops) –Light load b Mesh (6 hops) –Heavy load, concentration near portal g Mesh (5 and 2 hops) –Long data range –Multiple flows thru portal a Mesh (5 and 2 hops) –Short data range –Heavy load, multiple flows thru portal All scenarios consider only high-priority traffic (VOIP and Video)

10 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 10 Parameters Network802.11a802.11g802.11b Slot Time9 usec 20 usec Sifs Time16 usec10 usec Phy_CWmin15 Phy_CWmax1023 PLCP overhead control20 usec 96 usec PLCP overhead data20 usec 96 usec Control Data Rate24 Mbps 5.5 Mbps Difs Timesifs + 2*slot_time = 28 usec sifs + 2*slot_time = 50 usec Eifs_timedifs + sifs + 24 Mbps difs + sifs Mbps aifsn222 Aifs [ac] aifsn[ac] * slot_time + sifs_time = 28 usec aifsn[ac] * slot_time + sifs_time = 50 usec ACK tx rate24 Mbps 5.5 Mbps DATA tx rate54 Mbps 11 Mbps Cwmax1023 Cwmin15

11 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 11 Scenario 1: b Mesh (4 hops) Examines the effectiveness of Express Forwarding in an lightly loaded 11b single channel network Multi-hop path of 4 hops Surrounded by peer-to-peer mesh or independent WLAN flows on same channel Low total traffic load (3.9 Mbps), symmetrically distributed along the multi-hop path

12 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 12 Scenario 1: b Mesh (4 hops) Traffic description VIDEO (L): Low Resolution, 1.4 Mbps payload size: 1464 bytes, inter-arrival 8 ms VOIP : G711, 0.16 Mbps payload size: 200 bytes, inter-arrival 20 ms P VOIP VIDEO (L) Network configuration Data range: 25 m, Ack range: 31 m 4-hop path, next-hop neighbors dont hear each other Physical layer rates 11 Mbps 5.5 Mbps TOTAL LOAD: 3.9 Mbps

13 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 13 Scenario 1: Mean Delays Express Fwd Disabled Express Fwd Enabled Express Fwd With Express Rtx Node2 > Node Node3 > Node Node5 > Node Node7 > Node Node9 > Node Node10 > Node Node16 > Portal Portal > Node_ With Express Forwarding 4-hop VOIP ETE delay 50 ms + Express Retransmission 4-hop VOIP mean delay 33 ms Lower delays for all other flows Without Express Forwarding 4-hop VOIP mean delay 75 ms

14 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 14 Scenario 1: 4-hop Delays (CDF) Uplink Delay Downlink Delay

15 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 15 Scenario 1: Dropped Frames* With Express Forwarding Fewer frames are dropped by all nodes * Frames are dropped if retransmitted 7 times unsuccessfully

16 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 16 Scenario 1 – Summary of Results Express Forwarding benefits Reduces ETE delay for 4-hop VOIP flow by over 50% when combined with Express Retransmission All other flows (either on the mesh or in neighboring BSS) also enjoy delay reduction –Prioritization reduces contention – fewer collisions Fewer frames are dropped as a result of fewer retransmissions

17 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 17 Scenario 2: b Mesh (6 hops) Examines the effectiveness of Express Forwarding in more heavily loaded 11b single channel network, with increased hop count Multi-hop path of 6 hops Surrounded by peer-to-peer mesh or independent WLAN flows on same channel Total traffic load (5.7 Mbps), asymmetrically distributed along the multi-hop path, is more concentrated around portal P

18 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 18 Scenario 2: b Mesh (6 hops) P VIDEO (L) VOIP Traffic description VIDEO (L): Low Resolution, 1.4 Mbps payload size: 1464 bytes, inter-arrival 8 ms VOIP : G711, 0.16 Mbps payload size: 200 bytes, inter-arrival 20 ms Network configuration Data range: 25 m 6-hop path, reduced distance between neighbors Physical layer rates 11 Mbps 5.5 Mbps TOTAL LOAD: 5.7 Mbps

19 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 19 Scenario 2: Mean Delays Express Fwd Disabled Express Fwd Enabled Express Fwd With Exp Rtx Node2 > Node Node3 > Node Node5 > Node Node6 > Node Node7 > Node Node9 > Node Node10 > Node Node16 > Portal Portal > Node With Express Forwarding 6-hop VOIP mean delay 58 ms + Express Retransmission 6-hop VOIP mean delay 39 ms Lower delays for all other flows Without Express Forwarding 6-hop VOIP mean delay 164 ms

20 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 20 Scenario 2: 6-hop Delays (CDF) Uplink Delay Downlink Delay (sec)

21 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 21 Scenario 2: Dropped Frames With Express Forwarding Fewer frames are dropped by all nodes

22 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 22 Scenario 2 – Summary of Results Express Forwarding benefits Express forwarding reduces ETE delay for 6-hop VOIP flow by over 75% when combined with Express Retransmission All other flows (either on the mesh or in neighboring BSS) also enjoy delay reduction –Prioritization reduces contention – fewer collisions Fewer frames are dropped as a result of fewer retransmissions Traffic concentration near the portal Delays of other flows near the portal are longer –Express forwarding causes these delays to be lower The uplink and downlink delays of the multi-hop flow display small asymmetry –While the same hops are traversed in both directions, the traffic near the portal disadvantages the first downlink hop more as compared to the traffic near the first uplink hop –Note: Access on the first hop of a multi-hop flow is not prioritized

23 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 23 Scenario 3: g Mesh (5 hops) Examines the effectiveness of Express Forwarding in a long transmit range (391 meters) 11g single channel network Multi-hop paths of 5 and 2 hops Surrounded by peer-to-peer mesh or independent WLAN flows on same channel Total traffic load (17 Mbps), asymmetrically distributed along the multi-hop path, is more concentrated around portal P Multiple VOIP flows and Video go thru portal P

24 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 24 P VIDEO (L) VIDEO (H) VIDEO (L) VOIP Scenario 3: g Mesh (5 hops) Traffic description VIDEO (L): Low Resolution, 1.4 Mbps payload size: 1464 bytes, inter-arrival 8 ms VIDEO (H): High Resolution, 4.2 Mbps payload size: 1464 bytes, inter-arrival 2.83 ms VOIP : G711, 0.16 Mbps payload size: 200 bytes, inter-arrival 20 ms Network configuration TX RANGE: 391 m 5-hop path, next-hop neighbors dont hear each other Physical layer rates 54 Mbps 24 Mbps TOTAL LOAD: 17 Mbps

25 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 25 Scenario 3: Mean Delays Without Express Forwarding 5-hop VOIP delay 91 ms 2-hop VOIP delay 43 ms Flow 25-6 is unstable Flow delay 100 ms With Express Forwarding 5-hop VOIP delay 4 ms 2-hop VOIP delay 3 ms Flow 25-6 delay 4 ms Flow delay 3 ms + Express Retransmission 5-hop VOIP delay 3 ms 2-hop VOIP delay <3 ms Flow 25-6 delay 3 ms Flow delay <3 ms Substantially lower delays for all other flows

26 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 26 Scenario 3: Mean Delays (ms) Flows Express Forwarding Disabled Express Forwarding Enabled Express Forwarding With Express Re-TX Without Multi-hop Flows Portal->Node_ NA Portal->Node_ Portal->Node_ Portal->Node_ NA Portal->Node_ Node_5->Portal NA Node_7->Node_ Node_8->Node_ Node_11->Node_ Node_12->Node_ Node_13->Node_ Node_16->Portal Node_17->Portal Node_18->Portal NA Node_19->Node_ Node_24->Node_ Node_25->Node_

27 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 27 Scenario 3: 5-hop Delays (CDF) P -> Node5 Delay Node5 -> P Delay

28 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 28 Scenario 3: 2-hop Delays (CDF) P -> Node18 Delay Node18 -> P Delay (sec)

29 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 29 (sec) Scenario 3: Video (H) Delays (CDF) Node11 -> Node6 Delay Node25 -> Node6 Delay (sec)

30 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 30 Scenario 3: Dropped Frames There are practically no dropped frames with express forwarding

31 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 31 Scenario 3 – Summary of Results Express Forwarding benefits ETE delay for 5-hop and 2-hop VOIP flows is reduced by 95% All other flows (either on the mesh or in neighboring BSS) also enjoy substantial delay reduction –The long delays experienced by the Video (H) flows disappear when multi- hop flows are express forwarded Long Ack range Synergy with longer Ack range (47 % longer than Data range) magnifies the benefit of express forwarding, with result few retransmissions –Low retransmissions lead to few dropped frames With few retransmissions, Express Retransmission provides little further delay reduction Multi-hop flows Express forwarding removes the deleterious effect of multi-hop flows on other traffic

32 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 32 P VIDEO (L) VIDEO (H) VIDEO (L) VOIP Multi-hop effect in single-channel mesh If the ACK from 2 (or 3) causes collision at Node6, retransmission of frame from Node25 will wait till multi-hop TX P->Node5 completes The sooner the latter completes, the sooner the transmission Node25->Node6 will complete Express forwarding reduces time of transmission P->Node5, thus shortens delay for flow Node25- >Node6 With Express Forwarding, the ACK from 2 prevents collision by Node25 for the remainder of multi-hop transmission P->Node5 Multi-hop flow from P to Node5 ANIMATED

33 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 33 Scenario 3.1: Multi-hop flows removed P VIDEO (L) VIDEO (H) VIDEO (L) VOIP Traffic description The same as Scenario 3, with multi-hop flows (P->Node5 and P->Node18) removed Network configuration TX RANGE: 391 m 5-hop path, next-hop neighbors dont hear each other Physical layer rates 54 Mbps 24 Mbps X X TOTAL LOAD: 16 Mbps

34 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 34 Scenario 3.1: Mean Delays Delays are substantially shorter without the multi-hop flows Express Forwarding reduces the negative effect of multi-hop flows on other traffic

35 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 35 Scenario 3.1: Mean Delays (ms) Flows Express Forwarding Disabled Express Forwarding Enabled Express Forwarding With Express Re-TX Without Multi-hop Flows Portal->Node_ NA Portal->Node_ Portal->Node_ Portal->Node_ NA Portal->Node_ Node_5->Portal NA Node_7->Node_ Node_8->Node_ Node_11->Node_ Node_12->Node_ Node_13->Node_ Node_16->Portal Node_17->Portal Node_18->Portal NA Node_19->Node_ Node_24->Node_ Node_25->Node_

36 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 36 Scenario 4: a Mesh (5 hops) Examines the effectiveness of Express Forwarding in a short transmit range (25 meters) 11a single channel network Multi-hop paths of 5 and 2 hops Surrounded by peer-to-peer mesh or independent WLAN flows on same channel Total traffic load (20 Mbps), asymmetrically distributed along the multi-hop path, is more concentrated around portal P Multiple VOIP flows and Video go thru portal P

37 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 37 Scenario 4: a Mesh (5 hops) P VOIP VIDEO (L) VIDEO (H) VIDEO (L) VOIP TOTAL LOAD: 20 Mbps 25 Traffic description VIDEO (L): Low Resolution, 1.4 Mbps payload size: 1464 bytes, inter-arrival 8 ms VIDEO (H): High Resolution, 4.2 Mbps payload size: 1464 bytes, inter-arrival 2.83 ms VOIP : G711, 0.16 Mbps payload size: 200 bytes, inter-arrival 20 ms Network configuration TX RANGE: 25 m 2 multi-hop paths, next-hop neighbors dont hear each other Physical layer rates 54 Mbps 24 Mbps

38 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 38 Scenario 4: Mean Delays Without Express Forwarding 5-hop VOIP delay 179 ms 2-hop VOIP delay 79 ms P->16 VOIP delay 76 ms P->22 Video delay 45 ms With Express Forwarding 5-hop VOIP delay 19 ms 2-hop VOIP delay 6 ms P->16 VOIP delay 7 ms P->22 Video delay 6 ms + Express Retransmission 5-hop VOIP delay 10 ms 2-hop VOIP delay 4 ms P->16 VOIP delay 5 ms P->22 Video delay 4 ms Substantially lower delays for all flows

39 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 39 Express Forwarding Disabled Express Forwarding Enabled Express Forwarding With Express Re-TX Portal > Node Portal > Node Portal > Node Portal > Node Portal > Node Node5 > Portal Node7 > Node Node8 > Node Node11 > Node Node12 > Node Node13 > Node Node14 > Node Node15 > Node Node16 > Portal Node17 > Portal Node18 > Portal Node19 > Node Node21 > Node Node21 > Node Node24 > Node Node25 > Node Node25 > Node

40 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 40 Scenario 4: 5-hop Delays (CDF) Downlink Delay Uplink Delay

41 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 41 Scenario 4: 2-hop Delays (CDF) Downlink Delay Uplink Delay

42 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 42 Scenario 4: Dropped Frames Fewer dropped frames with express forwarding

43 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 43 Scenario 4 – Summary of Results Express Forwarding benefits ETE delay for 5-hop and 2-hop VOIP flows is reduced by 94% All other flows (either on the mesh or in neighboring BSS) also enjoy substantial delay reduction –The long delays experienced by the Video (H) flows disappear when multi- hop flows are express forwarded Traffic concentration at Portal The flows going thru the portal experience very long delays Express forwarding on the multi-hop flows reduces delays substantially on all flows thru the portal, both multi-hop and single- hop flows

44 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 44 Conclusion Express Forwarding reduces end-to-end delay of the multi-hop flows substantially Express Forwarding causes fewer frames to be dropped Express Retransmission reduces retransmissions and end-to-end delay further Other (not express forwarded) traffic also benefits substantially from Express Forwarding Express Forwarding and Express Retransmission enable multi-hop QoS traffic to get through a single-channel mesh fast with minimum impact on other traffic

45 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 45 Back up

46 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 46 Excessive latency in meshes Hidden terminal collisions between two transmissions are likely to repeat Single channel mesh A and D cannot hear each other C cannot receive when A transmits B cannot receive when D transmits Retransmission attempts likely to fail Increased delay for successful transmission A B Tx X C D (((((((( (((((((( X Interference

47 doc.: IEEE /2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 47 Express retransmission illustration Express retransmission (X-RTX) enables a multi-hop transmission to complete faster Express forwarding and X-RTX on single-channel mesh The transmissions from A->B and F->E lead to hidden terminal collisions Express retransmission enables the TSQ frame (A->B) to succeed upon retransmission The ACKs sent by B and C protect the frame as it is forwarded on A B E F (((((((( Tx Interference X X TSQ (((((((( Interference DC X-RTX TSQ ANIMATED


Download ppt "Doc.: IEEE 802.11-07/2454r0 Submission September 2007 M. Benveniste (Avaya Labs)Slide 1 Performance Evaluation of Express Forwarding for a Single-Channel."

Similar presentations


Ads by Google