Doc.: IEEE 802.11-00/398 Submission October 2000 Wim Diepstraten, LucentSlide 1 Initial D-QoS Proposal Maarten Hoeben-Intersil/NWN Menzo Wentink-Intersil/NWN.

Slides:



Advertisements
Similar presentations
Doc.: IEEE /387r1 Submission November 2000 W.-P. Ying, M. Nakahara, S. Ho, NextComm, Inc.Slide 1 A Scheduling Scheme for Level-2 Enhanced PCF.
Advertisements

Doc.: IEEE /399 Submission November 2000 Wim Diepstraten, LucentSlide 1 Baseline D-QoS Proposal Greg Chesson-Atheros Wim Diepstraten- Lucent Technologies.
Doc.: IEEE /351 Submission October 2000 Maarten Hoeben, Menzo Wentink, IntersilSlide 1 Enhance D-QoS through Virtual DCF Maarten Hoeben, Menzo.
Doc.: IEEE /080r1 Submission January 2001 Jie Liang, Texas InstrumentsSlide 1 Jie Liang Texas Instruments Incorporated TI Blvd. Dallas,
Medium Access Issues David Holmer
Contents IEEE MAC layer operation Basic CSMA/CA operation
Ethernet – CSMA/CD Review
Doc.: IEEE /0604r1 Submission May 2014 Slide 1 Modeling and Evaluating Variable Bit rate Video Steaming for ax Date: Authors:
Achieving Quality of Service in Wireless Networks A simulation comparison of MAC layer protocols. CS444N Presentation By: Priyank Garg Rushabh Doshi.
An Adaptive Energy-Efficient MAC Protocol for Wireless Sensor Network
Explicit and Implicit Pipelining in Wireless MAC Nitin Vaidya University of Illinois at Urbana-Champaign Joint work with Xue Yang, UIUC.
1 Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks INFOCOM 2001 Michael Barry, Andrew T. Campbell Andras Veres.
Presented by Scott Kristjanson CMPT-820 Multimedia Systems Instructor: Dr. Mohamed Hefeeda 1 Cross-Layer Wireless Multimedia.
Contention Window Optimization for IEEE DCF Access Control D. J. Deng, C. H. Ke, H. H. Chen, and Y. M. Huang IEEE Transaction on Wireless Communication.
802.11n MAC layer simulation Submitted by: Niv Tokman Aya Mire Oren Gur-Arie.
1 QoS Schemes for IEEE Wireless LAN – An Evaluation by Anders Lindgren, Andreas Almquist and Olov Schelen Presented by Tony Sung, 10 th Feburary.
Submission doc.: IEEE /1454r0 November 2014 Jarkko Kneckt (Nokia)Slide ax Power Save Discussion Date: Authors:
Voice Traffic Performance over Wireless LAN using the Point Coordination Function Wei Supervisor: Prof. Sven-Gustav Häggman Instructor: Researcher Michael.
Page 1 January 2002 doc.: IEEE 802.RR-02/018A-d5 IEEE 802 IEEE 802 proposal relating to DFS and JPT5G proposal.
A Multichain Backoff Mechanism for IEEE WLANs Alkesh Patel & Hemant Patel ECE 695 – Leading Discussion By : Shiang- Rung Ye and Yu-Chee Tseng.
Opersating Mode DCF: distributed coordination function
PLANETE group, INRIA Sophia-Antipolis July 1, 2003 Adaptive Channel allocation for QoS Enhancement in IEEE Wireless LANs Presented by: Mohammad.
1 Real-Time Traffic over the IEEE Medium Access Control Layer Tian He J. Sobrinho and A. krishnakumar.
Wireless Medium Access. Multi-transmitter Interference Problem  Similar to multi-path or noise  Two transmitting stations will constructively/destructively.
1 Dynamic Adaption of DCF and PCF mode of IEEE WLAN Abhishek Goliya Guided By: Prof. Sridhar Iyer Dr. Leena-Chandran Wadia MTech Dissertation.
Company LOGO Provision of Multimedia Services in based Networks Colin Roby CMSC 681 Fall 2007.
Distributed QoS model for IEEE doc.: IEEE /267 September 2000 Jan Kruys, Harold Teunissen, Lucent TechnologiesSlide 1 Distributed QoS model.
Voice Capacity analysis over Introducing VoIP and WLans IEEE based Wireless Local Area Networks (WLANs) are becoming popular While WLANs.
IEEE Wireless LAN Standard. Medium Access Control-CSMA/CA IEEE defines two MAC sublayers Distributed coordination function (DCF) Point coordination.
Demand Based Bandwidth Assignment MAC Protocol for Wireless LANs K.Murugan, B.Dushyanth, E.Gunasekaran S.Arivuthokai, RS.Bhuvaneswaran, S.Shanmugavel.
IEEE EDCF: a QoS Solution for WLAN Javier del Prado 1, Sunghyun Choi 2 and Sai Shankar 1 1 Philips Research USA - Briarcliff Manor, NY 2 Seoul National.
Covilhã, 30 June Atílio Gameiro Page 1 The information in this document is provided as is and no guarantee or warranty is given that the information is.
Doc.: IEEE /065r0 Submission January 2001 Brockmann, Hoeben, Wentink (Intersil) g MAC Analysis Menzo Wentink Ron Brockmann.
Vertical Optimization Of Data Transmission For Mobile Wireless Terminals MICHAEL METHFESSEL, KAI F. DOMBROWSKI, PETER LANGENDORFER, HORST FRANKENFELDT,
Copyright 2008 Kenneth M. Chipps Ph.D. Controlling Flow Last Update
An Energy Efficient MAC Protocol for Wireless LANs, E.-S. Jung and N.H. Vaidya, INFOCOM 2002, June 2002 吳豐州.
Quality of Service Schemes for IEEE Wireless LANs-An Evaluation 主講人 : 黃政偉.
SubmissionJoe Kwak, InterDigital1 BSS Load: AP Loading Metric for QOS Joe Kwak InterDigital doc: IEEE /0079r0January 2005.
doc.: IEEE /183r0 Submission March 2002 David Beberman, Corporate Wave Net, Inc.Slide 1 Single Burst Contention Resolution “Wireless Collision.
Doc.: IEEE /457 Submission December 2000 Mathilde Benveniste, AT&T Labs - ResearchSlide 1 An Enhanced-DCF Proposal Based on ‘Tiered Contention’
1 Ethernet CSE 3213 Fall February Introduction Rapid changes in technology designs Broader use of LANs New schemes for high-speed LANs High-speed.
Doc.: IEEE /1181r0 Submission October 2004 He et alSlide 1 Proposal for Fast Inter-BBS Transitions Xiaoning He Paragon Wireless, Inc. Sunnyvale,
Design and Implementation of a Reservation-based MAC Protocol for Voice/Data over IEEE Ad-Hoc Wireless Networks Shiann-Tsong Sheu, Tzu-Fang Sheu.
Doc.: IEEE /361 Submission October 2000 Wim Diepstraten, LucentSlide 1 Distributed QoS resolution Greg Chesson-Altheros Wim Diepstraten- Lucent.
Submission doc.: IEEE /599r1 November 2001 M. Benveniste -- AT&T Labs, ResearchSlide 1 ‘Cyclic Prioritized Multiple Access (CPMA): An Access Mechanism.
COE-541 LAN / MAN Simulation & Performance Evaluation of CSMA/CA
November 2000 Jin-Meng Ho, Texas InstrumentsSlide 1 doc.: IEEE /367 Submission p-DCF for Prioritized MAC Service Jin-Meng Ho, Sid Schrum, and.
Copyright © 2003 OPNET Technologies, Inc. Confidential, not for distribution to third parties. Wireless LANs Session
Doc.:IEEE /517r0 Submission August 2002 IBM Research Slide 1 Some Clarifications to IEEE e, Draft 3.2, August 2002 H.L. Truong and G. Vannuccini.
Submission doc.: IEEE /596r1 November 2001 M. Benveniste -- AT&T Labs, ResearchSlide 1 ‘Neighborhood Capture’ in Wireless LANs Mathilde Benveniste.
Doc.: IEEE /465r0 Submission Wim Diepstraten, Agere Systems July 2002 Slide 1 WiSP Wireless Sidelink Protocol Wim Diepstraten Gerrit Hiddink Agere.
Media Access Methods MAC Functionality CSMA/CA with ACK
EA C451 (Internetworking Technologies)
Topics discussed in this section:
Scheduled DCF vs. Virtual DCF
RTP: A Transport Protocol for Real-Time Applications
Topics in Distributed Wireless Medium Access Control
Multi-Channel MAC for Ad Hoc Networks: Handling Multi-Channel Hidden Terminals Using A Single Transceiver Jungmin So and Nitin Vaidya Modified and Presented.
A Scheduling Scheme for Level-2 Enhanced PCF MAC Service
GAPA - Efficient, More Reliable Multicast
Provision of Multimedia Services in based Networks
Simulation Results for QoS, pDCF, VDCF, Backoff/Retry
GAPA - Efficient, More Reliable Multicast
HCF medium access rules
DL MU MIMO Error Handling and Simulation Results
PCF Enhancements and Contention Free Bursts
HCF medium access rules
Infocom 2004 Speaker : Bo-Chun Wang
Wireless MAC Multimedia Extensions Albert Banchs, Witold Pokorski
Presentation transcript:

doc.: IEEE /398 Submission October 2000 Wim Diepstraten, LucentSlide 1 Initial D-QoS Proposal Maarten Hoeben-Intersil/NWN Menzo Wentink-Intersil/NWN Wim Diepstraten- Lucent Technologies WCND Harold Teunissen-Lucent Technologies Prepared by:Wim Diepstraten

doc.: IEEE /398 Submission October 2000 Wim Diepstraten, LucentSlide 2 QoS Requirements The reigning network paradigm is IP - not ATM or circuit switching IP performance is inherently variable Multi-media applications for IP based networks include - or will include - pacing mechanisms to adapt to available capacity –e.g. the new MPEG4 standard allows continuous adaptation of the encoder to available capacity Wireless QoS should take advantage of these developments and focus on: –maintaining [reasonable] performance in the presence of interference –maintaining delivery rates rather than tight delay constraints per packet –accommodate rate adaptation

doc.: IEEE /398 Submission October 2000 Wim Diepstraten, LucentSlide 3 Assumptions The Enhanced DCF proposal is part of a total QoS proposal which is layered as follows: –QoS Level 0: No QoS using DCF & PCF as in –QoS Level 1: Priority based QoS delivery using enhanced DCF –QoS Level 2: Priority based QoS delivery using enhanced DCF & PCF –QoS Level 3: Parameterized QoS delivery using enhanced DCF & PCF Each successive level does support the full functionality of the level below. Use priority based interface supporting 8 Priority classes according to q.

doc.: IEEE /398 Submission October 2000 Wim Diepstraten, LucentSlide 4 D-QoS objective Objective is to use a low complexity approach. Provide priority driven service differentiation between a number of different traffic priority classes. Provide mechanisms to limit the medium load in order to keep the delay of higher priority classes under control. –While avoiding starvation of low priority classes. The mechanism should be adaptive such that priority class differentiation mechanisms are only active when there is active traffic load in higher priority classes. –If there is only “Best Effort” traffic load, then the throughput performance should be similar to the legacy throughput. Should be able to provide priority differentiation also in the presence of legacy DCF devices, although effectiveness can be reduced. Can be applied in both ESS and IBSS.

doc.: IEEE /398 Submission October 2000 Wim Diepstraten, LucentSlide 5 Approach Up to 8 priority classes are distinguished, each having a separate Q. –Mapping to a limited subset of about 4 may be more practical. Objective is to control the total medium load such that the delay of high priority classes are acceptable. –There are no absolute limits, only relative. By using the medium load feedback to control the distinction between service classes. –By reducing the service rate of lower classes to assure better service of higher priority classes when load increases. –If this load feedback is indicating the load per priority class on the medium, then the differentiation model can be adaptive to scale up the lower priorities, when no traffic is present for any of the higher priority classes. So that performance of “Best Effort” only traffic remains as is. The same distinction in service within a station should work across the whole BSS.

doc.: IEEE /398 Submission October 2000 Wim Diepstraten, LucentSlide 6 Basic model for example Medium Access Control Multi- media Traffic Source System Interactive Stream Best Effort Drop Rate Control Service Rate Control System & Ntwrk Mant

doc.: IEEE /398 Submission October 2000 Wim Diepstraten, LucentSlide 7 Model discussion A limited number of Access Priority Q’s are defined, and 4 is considered a good limit. Two basic functions are provided: –“Service Rate Control” mechanism that is to assure service differentiation between priority classes by using different access priorities. –A “Drop Rate Control” mechanism that is to regulate the “Offered Load” based on medium capacity limits. Apart from this there can be a “Submission Control” function above the MAC.

doc.: IEEE /398 Submission October 2000 Wim Diepstraten, LucentSlide 8 D-QoS mechanism approaches The approach presented at the October 13 conference call has changed. –The previous (SR, CW based) submission rate mechanism did suffer from non-linear effects causing fairness issues, and poor differentiation control. –This is replaced by a superior approach suggested by Maarten Hoeben called Virtual-DCF (V-DCF). The handling of legacy DCF devices is addressed separately on slide 17. What is the definition of “Fairness”? –In the DCF “Fairness” is having statistically the same medium access opportunity to get a packet on the medium, independent of packet size or Q-depth.

doc.: IEEE /398 Submission October 2000 Wim Diepstraten, LucentSlide 9 Access mechanism A “Medium monitor” function does measure the “Load per Class” in terms of medium occupancy duration, over a period, of for instance a Beacon interval. A “Virtual-DCF” (V-DCF) mechanism is used for each priority class within a station, each using a separate CWmin (CWx, a per class parameter). –So in our example there are up to 4 DCF contenders within each station. The “Medium monitor” function does translate the “Load per Class” into a set of CWx parameters, which should be the same for all stations. –How this translation takes place does preferably not need to be standerdized. The “V-DCF” mechanism does generate a separate backoff count per “Access Priority” level. –Which individually decrement when the medium is not busy. A transmission starts when one of the counters decrements to zero. –A collision between local accesses are resolved locally. The higher priority frame will be send first, while the lower priority can be send after the post-backoff of the higher access priority frame Do we need to standardize this, or can it be a policy decision

doc.: IEEE /398 Submission October 2000 Wim Diepstraten, LucentSlide 10 CW Strategy CW use strategy: –Intent is to use a CWmin that is a good trade-off between collision probability, and delay increase due to excessive retries. –It is assumed that the nominal CWmin=31 as defined for the legacy systems is used for traffic with a number of contenders. Never use a CWmin lower then 31 for “Best Effort” Traffic. –It could be acceptable for low loads, but affects priority over legacy traffic. –A CWmin lower then 31 can be acceptable for higher priority classes that only have a few contenders. Note that the “Load Monitor” can also gather information how many contenders there are within one priority class. But the retry probability will increase when doing this. –Because of collision with lower priority stations that are in backoff, or due to direct contenders in the same priority class. –A low CWmin will increase the retry rate of the high priority traffic, which may result in more net delay. So a proper tradeoff needs to be made (simulations needed).

doc.: IEEE /398 Submission October 2000 Wim Diepstraten, LucentSlide 11 Example CW as function of load. Assumption: C0=low priority and “Best Effort” is C3 is highest priority. Load <= 50%: All classes used. Alternative If low C3 load. –C3:CW3=31CW3=15 –C2:CW2=63CW2=31 –C1:CW1=95CW1=47 –C0:CW0=127CW0=63 Load >50%: All classes used.Alternative if C3 is empty (no load monitored) –C3:CW3=31 CW3=15 –C2:CW2=63 CW2=31 –C1:CW1=127 CW1=63 –C0:CW0=255 CW0=127 Load > 50%: Class C3 and C1 empty –C3:CW3=15No load monitored at this level but good differentiation when traffic starts. –C2:CW2=31 –C1:CW1=47No load monitored at this level, but some differentiation when traffic starts –C0:CW0=63higher priority because C3 and C1 are empty. Example shows that as function of Load>50% more differentiation is achieved by further trottling down the lower priority traffic, because they are using higher CWmin values. –But if no load is present within a access priority class, then relative priority of C0 is increased. –The Red slanted case is used for further explanation in the next pages.

doc.: IEEE /398 Submission October 2000 Wim Diepstraten, LucentSlide 12 Differentiation example: >50% load: Alternative if C3 is empty (no load monitored) –C3: CW3=15No load in this access priority class –C2: CW2=31 –C1: CW1=63 –C0: CW0=127 Evaluating the red / slanted case –Note that if C3 is empty then higher priority CW is using better (smaller) CW settings so all the lower priorities are bumped up a level. –If in the mean time C3 traffic is starting, it still has access advantage because it then uses a CW=15 in that station as long as that Q is filled. This still provides class differentiation, while all CW’s of the lower classes are bumped up a class. A station that has only low priority class C0 traffic, will access the medium by using a CW=127 in the above >50% load example, compared to a CW=31 for the C2 class. –So compared to other stations on the medium that generates C2 traffic, the C4 traffic has a factor 4 lower access priority. Because its average backoff is a factor 4 larger.

doc.: IEEE /398 Submission October 2000 Wim Diepstraten, LucentSlide 13 Example cont’d >50% load: Alternative if C3 is empty (no load monitored) –C3: CW3=15No load in this access priority class –C2: CW2=31 average access after every 15 slot no medium busy. –C1: CW1=63 average access after every 31 slot no medium busy. –C0: CW0=127 average access after every 63 slot no medium busy. If we monitor that there is no load in the higher classes at all, then the parameters could be set as follows: –C1:CW3=7No active traffic in this class –C2:CW2=7No active traffic in this class –C3:CW1=15No active traffic in this class –C4:CW0=31Only Best effort traffic active. –All stations do now use CW=31, and send C0 traffic with same CW as legacy. –So will have identical throughput performance as legacy systems today. If in the above example high priority traffic is started at a station. –Then within that station the associated lower CW is used, providing relative priority –After the next monitoring period, the CWx can be adopted to increase CW in subsequent classes, to keep the collision probability at an acceptable level.

doc.: IEEE /398 Submission October 2000 Wim Diepstraten, LucentSlide 14 Retry mechanism changes The Retry mechanism should be such that high priority traffic does not suffer from lower priority traffic being exhaustively retried. –For stability reasons the CW need to increase (double?) at retry, as in the current DCF. Suggest that all pending backoff counters are doubled, and that the CW of the retrying frame is doubled also, as normal for a DCF. Goal of the doubling CW was that the station temporarily reduces its load on the medium, which is the main thing we want to achieve here. –So higher priority traffic does not have to wait until the lower priority Retry is completed, but its access priority is reduced compared to other medium contenders due to the CW increase, and the fact that pending backoffs are doubled. And CW will be changed to CWmin of that class after this frame is succesfully transmitted. Exhaustive retries are no longer happening in a station due to the inherent parallel nature of the V-DCF approach. –The first backoff counter reaching zero does determine which frame to send. –CWmax may need to be extended to different values per class.

doc.: IEEE /398 Submission October 2000 Wim Diepstraten, LucentSlide 15 V-DCF characteristics Because there are multiple traffic sources within the same station doing Virtual-DCF, it does have effect on the medium load. Example, when assuming a CW=31 is an equivalent load of 1. –Then for the case where CWx= 31, 63, 127, 255, with all Q’s active The equivalent load is 1+1/2+1/4+1/8=1.875 when C2 is empty then load is 1+1/4+1/8=1.375 when only C0 is filled then load is effectively 1/8=.125 –If the differentiation is less CWx= 31, 47, 63, 127 with all Q’s active The equivalent load is 1+2/3+1/2+1/4=2.41 –So a V-DCF station with multiple classes is more aggressively accessing the medium. So the actual load a station represent to the medium is depending on the CW differentiation, and the momentary Q situation. –Simulations must show us the effects of this.

doc.: IEEE /398 Submission October 2000 Wim Diepstraten, LucentSlide 16 Load monitoring Monitoring characteristics: –How adaptive does this need to be? To handle bursty traffic effectively we need a measurement granularity which is relative short. A Beacon interval granularity seems to be in the ballpark. –Given that for QoS the Beacon interval is expected to be in the ms range. Two approaches possible: –Each station does do its own monitoring, determining the load per class. This will have effect on its Power Management capability, if measurements are done continuously. Or only do the measurement just prior to sending traffic. –This may not be effective, and increases latency. Stations do also take into account the visible overlapping BSS load. –Only the AP is doing the measurements, and distribute the CWx per class in every Beacon (new element). No effect on Station Power Management. But AP probably does not see any overlapping BSS load. Each station uses the same Service Rate scheduling parameter (CWx), assuring fair access differentiation. Having the monitor function in the AP and distribute the CWx to the stations is considered the superior solution.

doc.: IEEE /398 Submission October 2000 Wim Diepstraten, LucentSlide 17 ESS approach Monitoring is default done in the AP. –The translation from “Load per Priority” to the CWx value per class, can be policy based, and does not need to be specified in the standard. The Policy management only needs to be done in the AP, and applies identically in the whole BSS. So that the whole BSS uses the same CWx values. –A list of CWx parameters are distributed to each station in the Beacon. Using a new element. –Only the rules how to use the CWx per class parameter in any station needs to be standerdized. –There can still be a “Monitor function” active in a station, which can further tune the CWx parameters to the locally monitored situation. The AP supplied parameters are considered the default. But only unidirectional CW changes (getting larger) are allowed from the AP supplied CWx values. This can help to reduce BSS overlap issues.

doc.: IEEE /398 Submission October 2000 Wim Diepstraten, LucentSlide 18 IBSS How do we support IBSS? –For IBSS we can use a similar approach. With the Beacon Master (the one winning the distributed Beacon contention) doing the monitoring. –And send a Beacon with the QoS parameter element in a next Beacon with Backoff=0, resulting in one measurement in two Beacon intervals. –But this requires that each station does need to implement the capability to do the Load Monitoring, and do the translation to the CWx parameters. Either in a standardized way, so that each station comes to a similar conclusion or in a proprietary way, so that a different “load translation” regime may occur per Beacon interval, due to the distributed Beacon approach. –This may not be too bad, getting to an average result. –Or use a default CW parameter table, which can be overwritten when a measurement function is present in any station. –Issue: Can we do without standardizing the Load to CW translation? Is it acceptable to make the measurement function voluntary? Using a fixed default when no measurement function is implemented?

doc.: IEEE /398 Submission October 2000 Wim Diepstraten, LucentSlide 19 AP access priority Access efficiency of an AP can be significantly improved, which is desirable because: –Most traffic is going via AP, while majority of traffic is down traffic. –Increasing Access Priority by using smaller CW for an AP translates into higher collision probability, so this can be counter productive. Better is to allow more packet transmission per single access for an AP. –This can be achieved by allowing an AP to “Burst” its traffic per access up to a certain limit. Generating either Directed or BC/MC frames with SIFS in between, like we do with fragmentation. Abort the burst when no Ack is received where expected, and continue the burst after another DCF access. AP’s can apply some hysteresis to adapt fast to load increase, and slow to load decrease. –The burst content should take into account the “Access Priority” controlled scheduling mechanism, such that the service rate ratio per Q are followed. –This is also compatible with Legacy devices. Should the Burst mechanism also be allowed for stations? –Perhaps we can allow a limited burst, up to a fixed max duration. The max duration can be the medium time associated with a max size (2304 Byte) frame at the highest Basic rate. Do we need to standerdize an algorithm? –At least need to agree on limits.

doc.: IEEE /398 Submission October 2000 Wim Diepstraten, LucentSlide 20 Legacy device handling Legacy devices are assumed to use the CWmin=31 and contend with the priority traffic in a way that breaks the differentiation model. –For those priority levels that do not use a lower then 31 CWmin. However all traffic from legacy devices will go through the AP. So an “Enhanced DCF” AP does put the down traffic in the lowest priority Q. –Management frame responses could be an exception. In most higher layer protocol environments this will automatically reduce the traffic in the up direction. –AP’s could implement an even lower priority Q for its legacy traffic, so that it bandwidth is extra reduced to balance this further. If we assume that the number of stations that generate high priority traffic is very limited, then a lower CWmin then 31 can be used, which improves the relative priority compared to legacy traffic.

doc.: IEEE /398 Submission October 2000 Wim Diepstraten, LucentSlide 21 Drop Rate Control Intent is to apply mechanisms that reduce the offered load from the application when average delay increases due to too high a medium load. What mechanisms can we apply to control the traffic load coming from the higher layers? –Best method is to trigger the congestion control mechanisms of the higher layers. –How acceptable is “Drop control” in different environments. Most higher layer protocols that use some kind of Ack mechanism are sensitive to dropped packets, and it is causing them to (temporarily) reduce their load. It is the main congestion control mechanism in TCP/IP. –However there are also protocols that are not sensitive for this. Like UDP, or RTP flows in some cases. Here you have to police the traffic stream itself by dropping frames (probably more aggressively). For this reason it is also needed that drop control is distributed over the stations, because doing drop control only in the AP only affects one side. Some applications have their own higher level approaches to deal with that. –Reacting on lost packets, or end-to-end delay measurements.

doc.: IEEE /398 Submission October 2000 Wim Diepstraten, LucentSlide 22 Drop Rate Control What “Drop Control” needs to be applied? –Does every Class have a specified Q-depth, that needs to be controlled? The Q-depth is felt to be a function of the burstyness of the stream, which causes a temporary high load, while the average Q depth should probably be low. –There probably needs to be a “Time to Live” driven drop control at the output of the Q. –What “Drop Control” mechanism can be applied at the input of the Q, when the Q is full? Or use a mechanism that reacts on a Q-average threshold. Could a “Limited retry” (different per class?) be an automatic drop control mechanism? –For instance Retry C3,C2 = 2, and Retry C1 = 3, and Retry C0=4 –Assuming that C0 does need less load control because the best effort traffic can handle much more delay. While the SR differentiation does trottle down this traffic significantly anyway, when the load increases. Should we standerdize this, or leave it to the implementations?

doc.: IEEE /398 Submission October 2000 Wim Diepstraten, LucentSlide 23 Mechanisms needed Priority in frame header –a new field compatible with also the PCF approach needs to be defined. D-QoS Element in the Beacon –Containing the CWx list per access priority level. –Which can also provide the “Total Load” info, which can be used for “Load Balancing” purposes. Rules to use CW and retry change –See elsewhere in this document. Sequence# generation and Duplicate detection rule changes. –To support the non-exhaustive retry approach in a station, it is needed to maintain a sequence# per SA.Class combination. So the sequencing rules need to change, such that individual S# is maintained per Class, and will increment. And AP’s and Station receivers need to maintain duplicate detection mechanism on a SA.Class basis, and expand its resources for that. –Issue: If we want to detect reception errors based on S# (for the cases where no Acks are used), then we may need to maintain a S# per SA.DA.class. Capability exchange mechanisms to determine service levels.

doc.: IEEE /398 Submission October 2000 Wim Diepstraten, LucentSlide 24 Mechanisms cont’d In order to improve the efficiency of the AP, the AP should be allowed to concatenate multiple frames in one Tx-Access Opportunity. –A (MIB?) limited burst of frames are allowed per access opportunity with SIFS in between. –This can be done in a similar way as specified for fragmentation. With the “Duration” field containing the proper values for the next exchange. So Ack contains the duration of the next data frame, and its associated Ack. Also a station can be allowed to send a burst, but limited to a max size (2304 Byte) duration equivalent size. –We probably need to limit this functionality only to a 2304 Byte duration equivalent for the highest Basic rate only. To prevent excessive jitter.

doc.: IEEE /398 Submission October 2000 Wim Diepstraten, LucentSlide 25 Other mechanisms Orthogonal to this, D-QoS can also make use of any other efficiency improvements that are going to be defined in the standard. –Like direct Station to Station traffic, and the Bridge Portal concept. –Apply different Ack policies (Ack, No-Ack). Delayed Ack is a complexity level that won’t be supported at level 1. –Data Aggregation (if any) (multiple MPDU’s in one PSDU to reduce PHY overhead). This should however be “Conformance level” dependent. –The lowest “Enhanced DCF” level should probably not support the use of new Ack policies or data aggregation. –Because that is an extra complexity step, which may only apply to the “Enhanced DCF” in QoS Level 2 or 3. It will depend on how generic this functionality is.

doc.: IEEE /398 Submission October 2000 Wim Diepstraten, LucentSlide 26 Behavior evaluation Relative priority differentiation is achieved between classes If a class is empty within the whole BSS, then the distributed CW are such that maximum throughput is achieved. –By using next level up priority. If only “Best Effort” is used, then the performance will be the same as today. –Because the CW value for the traffic used will be identical for the whole BSS, equal to legacy devices. –That means that there will be transitional issues when higher priority traffic is started, because during that Beacon interval there will not be a class differentiation. There can however still be a CW differentiation for the empty classes, so that when they start they do have a higher access priority. AP’s can apply some hysteresis to adapt fast to load increase, and slow to load decrease. The DCF will defer for some of the overlapping traffic, but will not see the majority of its interferers, which will then result in more retries, and so increased delay. –How acceptable is this for QoS?? –This will in most cases result in more dropped frames, so a reduced offered load. –Applying RTS/CTS to retry traffic may help in some cases. Simulations must show the difference in “Load measurement” method in either AP or Station.

doc.: IEEE /398 Submission October 2000 Wim Diepstraten, LucentSlide 27 Conclusion A D-QoS mechanism is proposed that provides for a minimum complexity enhancement of the DCF only, by the implementation of a Virtual-DCF scheme. The mechanism is dynamic such that if not all traffic classes are active, then the overall throughput performance of lower priority traffic is increased. Translation of the load per class to the CWx parameters do not need to be standerdized when the measurement function is implemented in the AP. This mechanism can also be applied to IBSS in a number of ways, that need further discussion. Interoperability, and priority control over legacy DCF is achieved in most cases. Further efficiency improvements are suggested to give an AP more access priority, through the use of a traffic burst mechanism.