05235663 Quan Trong

download 05235663 Quan Trong

of 4

Transcript of 05235663 Quan Trong

  • 8/13/2019 05235663 Quan Trong

    1/4

    Survey of admission control algorithms in IEEE 802.11e Wireless LANs

    Yanbing Liu

    Department of Computer Science and Technology

    Chongqing University of Posts and Telecommunications

    & Chongqing University Chongqing, China

    Email: [email protected]

    Man Meng

    Department of Computer Science and Technology

    Chongqing University of Posts and Telecommunications

    Chongqing, China

    Email: [email protected]

    AbstractThe provision of Quality of Service (QoS) inwireless network becomes very important. IEEE 802.11estandard was introduced to enhance the Media AccessControl (MAC) layer operations to provide the QoS expe-rienced by traffic flows. But only IEEE 802.11e can notguarantee network congestion avoidance and traffic QoSespecially to multimedia traffics. So, an efficient admission

    control scheme is necessary to transmit multimedia trafficin high efficiency and quality. At present many admissioncontrol schemes have been proposed, in this paper we overallsummarize the purpose, basic components, and classificationsof admission control. And then present a survey of admissioncontrol in IEEE 802.11e EDCA and HCCA, these admissioncontrol schemes are discussed according to different cate-gories: Measurement-Based, Model-Based, Hybrid schemesof EDCA and Physical-Rate and Contention-Window-Basedof HCCA in IEEE 802.11e WLANs. Finally we will comparethese schemes to get the advantages and disadvantages andpoint out the remaining challenges which are useful for us tostudy the admission control under heterogeneous networksand make admission control schemes accurate and simple.

    Keywords-IEEE 802.11e; admission control; QoS;

    I. INTRODUCTION

    With the rapid growth of wireless communication sys-

    tems, the number of wireless users has consequently

    increased. Therefore, wireless networks should be able

    to provide guaranteed QoS for different services. IEEE

    802.11e standard [1] was introduced to overcome the lack

    of QoS support for the legacy IEEE 802.11 WLANs.

    It introduces a new access method named Hybrid Co-

    ordination Function (HCF) which consists of two parts:

    Enhanced Distributed Channel Access (EDCA) which can

    only support relative QoS; and the HCF Controlled Chan-

    nel Access (HCCA), which can support absolute QoS, but

    involve complex polling and scheduling to allocate the

    resources to the competing stations. The former supports

    QoS by providing service differentiation through assigning

    different channel access parameters (CAPs) to different

    Access Categories (ACs). But only service differentiation

    can not satisfy the QoS requirement of each traffic class.

    The latter is not suitable for VBR traffics. To solve

    the above problems, proper admission control scheme is

    essential to support multiple types of traffics with various

    QoS requirements. Reference [2] has summarized some

    admission control schemes and compared their character-

    istics.In this paper, we provide a new survey of recent

    advances in admission control scheme of IEEE 802.11e

    WLANs. Our purpose is to study how the new QoS

    schemes and parameters provided in EDCA and HCCA

    can be well utilized to fulfill the requirements of admission

    control, thus will help us analyze the admission control

    under heterogeneous networks in future. We also compare

    these schemes to point out their advantages and disadvan-

    tages which are useful to make admission decision simple,accurate, and improve network performance.

    The remainder of this paper is organized as follows.

    Section II gives an overview of admission control. Section

    III and IV describe the survey of recent research works

    in admission control for both EDCA and HCCA. Section

    V presents the comparison of these admission control

    schemes. Finally, we draw the conclusion.

    II. AN OVERVIEW OF A DMISSION C ONTROLS CHEME

    A. Propose of Admission Control

    It is a key component of QoS-based resource man-

    agement schemes and mainly manages wireless resourcesin order to adapt to traffic variations. In other words,

    admitting a new flow must meet two conditions: one is

    that there are enough resources to meet the QoS of new

    traffics; the other is that the upcoming real time traffic

    into a particular service class does not degrade the QoS of

    the existing traffics, while simultaneously ensuring that the

    scarce bandwidth is utilized efficiently. But many purposes

    of admission control schemes vary in term of the design

    principles as indicated in [3]. Also the issue of fairness

    among different traffic classes should be taken into account

    in assigning admission control schemes.

    B. Basic Components of Admission Control

    Admission control extracts its decision based on the

    collaboration of three basic components. As shown in

    figure 1 [4], traffic descriptor, admission criteria, and net-

    work QoS state and flow information are the fundamental

    architectures of an admission control scheme. These three

    components are in cooperation with each other in order to

    achieve specific objectives.

    An admission control module obtains the traffic descrip-

    tor, and the QoS requirements of the flow as its inputs,

    and outputs its decision of either admitting the flow at

    the demanded QoS or denying it if that QoS is not met.

    A traffic descriptor is a set of parameters of the sourcethat describes the traffic characteristics. In order to obtain

    the admission control decision, the admission controller

    2009 ETP International Conference on Future Computer and Communication

    978-0-7695-3676-7/09 $26.00 2009 IEEE

    DOI 10.1109/FCC.2009.47

    230

    2009 ETP International Conference on Future Computer and Communication

    978-0-7695-3676-7/09 $26.00 2009 IEEE

    DOI 10.1109/FCC.2009.47

    230

  • 8/13/2019 05235663 Quan Trong

    2/4

    Traffic Descriptions and

    QoS Requirements

    Admission Criteria

    Admission Control

    Unit

    Network QoS State

    and Flow

    Information

    Admission control

    module

    Admission Decision

    Figure 1. Basic Components of Admission Control

    Admission Control

    Classifications

    Centralized or

    Distributed

    Approaches

    Model-based or

    Measurement-based or

    Hybrid

    Approaches

    Uniform Cell-based,

    Non-

    Uniform Cell-based or

    User-based

    Figure 2. Classifications of Admission Control

    performs the admission criteria module, which is a set of

    rules used by the admission control scheme to make the

    decision [5].

    C. Classifications of Admission Control

    The admission control schemes can be classified by

    several properties. Some of these properties are shown

    in figure 2. They are based on various criterions. Each

    criterion has its advantages and disadvantages. In next

    section, We will study the admission control schemes for

    IEEE 802.11e in detail.

    III. ADMISSION C ONTROLF OR EDCA

    Recently, several admission control schemes have been

    proposed for EDCA. We will classify these admission

    control schemes into three classifications: measurement-

    based, model-based, and a joint measurement-based and

    model-based admission control scheme.

    A. Measurement-Based Admission Control

    1) Threshold-Based Admission Control: Generally, in

    this scheme, each station needs to measure the traffic

    conditions and the network defines suitable upper or lower

    bound threshold values that indicate the current network

    load like [6]. It is centralized and Network Utilization

    Characteristic (NU C) of a flow will be used as the

    decision criterion in admission control. N U C of a flow

    is defined as the fraction of time per time unit needed to

    transmit the flow over the network. If the N U C of all

    the flows (NU C total) is below the set N U C threshold

    (NU C threshold), the flow can be admitted.

    Although this scheme is very easy to implement and

    can guarantee the QoS of high priority flows when the

    medium is heavily loaded, the throughput of the low

    priority flows with admission control performs worse than

    the case without admission control, that is the issue offairness is not considered. Another problem is difficult to

    set the NU C threshold value.

    2) Resource Sharing-Based Admission Control: The

    reservation and sharing of bandwidth is an important

    method to achieve QoS guarantees like [7]. In this scheme,

    bandwidth is reserved for a particular traffic AC and

    also be shared among them, such as 20% reserved to

    VoIP (AC0); 40% reserved to video (AC1); 30% sharedbetween AC0 and AC1; 10% exclusively reserved tocontentions between web and ftp traffic belonging toAC2andAC3respectively. Thus, utilization percentages of thechannel can be got: UAC0 0.5 and UAC1 0.7 andUAC0 + UAC1 0.9. UACi can also be expressed as theratio between the total throughput of the corresponding

    AC and the BSS maximum throughput Thpeq with a

    certain mean packet size during an observing interval.

    So, the detail criteria of admission control is as follow:

    IfUAC0 0.5, reject new flow. IfUAC1 0.7, reject newflow. IfUAC0 + UAC1 0.9, reject new flow.

    The advantages of this scheme are its simplicity and fair

    to all different traffics. However, one drawback is the staticpartition of bandwidth which could lead to an inefficient

    utilization of resources if significant load variations occur.

    B. Model-Based Admission Control

    The earliest model of WLANs is Markov Chain Model

    which is the foundation of the latter model, but it is

    proposed for IEEE 802.11 and it has some limitations.

    Researchers have built new models for IEEE 802.11e

    through improvement.

    1) G/G/1-Based Admission Control: In [8], an im-

    proved analytical model G/G/1 is firstly built to derive

    an average delay estimation, as well as channel utilization(cu) which are the decision criteria. In this scheme, the

    average data rate (Rmean), peak data rate (Rpeak) and

    average packet length (P Kl) are used to characterize the

    bandwidth requirement of a real-time traffic. Then, cu

    can be calculated corresponding to a flows bandwidth

    requirement as follows: cu= RPKl

    Tsuc, where R is the

    traffic rate. Thus, bandwidth requirement of a flow can be

    translated into (cumean, cupeak).

    The coordinator records the total channel utilization

    due to all admitted real-time flows into two parameters

    (cuA,mean, cuA,peak). We also estimate the average delay

    Di using the G/G/1 model. So in order to admit a

    new QoS flow, three requirements need to be satisfied:

    the first is cuA,mean+cui,mean < CUrt; the second is

    cuA,peak+cui,peak < CUmax; the last is the average delay

    Di less than the delay bound Di, i=2,3.

    The proposed scheme successfully guarantees strict QoS

    requirements of real-time traffics, while achieving high

    channel utilization. But the G/G/1 model delivers a rough

    upper bound for the average delay, which become looser

    as the number of flows increases. As a consequence, this

    scheme can only suggest a pessimistic limit on the number

    of admission users for small-size networks.

    2) Parameters-Based Admission Control: To improve

    the performance of network, only adjusting EDCA pa-rameters cant satisfy the QoS of traffics. So We should

    combine adjusting parameters with the admission control

    231231

  • 8/13/2019 05235663 Quan Trong

    3/4

    scheme like [9]. This scheme works jointly with a tuning

    algorithm which decides the most suitable configuration

    of the MAC parameters when a new flow arrives or

    leaves the system. An EDCA analytical model, which is

    presented in [10], is used to estimate the feasibility of

    the new state conditions (including the requesting flow)

    and allows to evaluate all EDCA parameters. The most

    suitable combination of the MAC parameters is from an

    heuristic real-time algorithm. The algorithm first computes

    the minimum aggregated bandwidth required by all flows.

    Using this value and the achievable maximum physical

    bandwidth, the data rate, the MAC parameters are adjusted

    based on a set of predefined thresholds.

    This scheme sufficiently considers all EDCA parameters

    and fairness between uplink and downlink. But the values

    of threshold are difficult to set and some assumptions exist

    which is inaccurate to algorithm.

    C. Measurement-Aided, Model-Based Admission Control

    To avoid the drawback of measurement-based and

    model-based schemes, a hybrid scheme has become the

    main method in admission control scheme.

    1) Threshold-Based Admission Control: This scheme

    like [11] is that a QAP monitors the channel continu-

    ously and measures the contention probability. When a

    new flow ofACi requests admission, the QAP estimates

    the equivalent number of competing entities of class i

    which is a method for converting a heterogeneous EDCA

    network into a homogeneous equivalent DCF network, and

    predicts the achievable bandwidth and one-hop delay of

    the new flow. This can be done using the non-saturation

    homogeneous model. Only if the bandwidth and delay

    requirements of new and admitted flows are all satisfied,

    the new flow can be admitted.

    This scheme proves to be very simple, and more practi-

    cally feasible without complex computation compared to

    the approach that relies on a direct study of the EDCA

    system. However, the analytical model of a non-saturated

    is for IEEE 802.11 DCF which is not accurate to admission

    control for EDCA.

    2) Channel Access Parameters (CAPs)-Based Admis-

    sion Control: This method [12] is similar to the above

    [10]. But in this scheme, the admission control parameters

    including the maximum tolerable collision rate for traffic

    i, the maximum retransmission limit, , and TXOP period,

    TXOPi are obtained through existing analytical model

    [13], then dynamically updates the traffic CAPs based

    on periodical measurement of current channel conditions.

    If the traffics maximum tolerable collision rate is less than

    the channel collision rate, the traffic required delay bound

    and dropping rate cannot be satisfied under the current

    channel condition, the traffic will be rejected at the station

    level. If the traffic can be admitted at the station level, thestation will further forward this admission request to the

    AP. The AP makes a decision according to (1) to accept

    or reject this admission request.

    TXOPi,k+1 Ri(n) +

    K

    k+1

    TXOPk Rk(n) SI (1)

    where k is the number of existing flows, R is the ratio

    of the total allocated time over the stated application ratesrequired time for successful MSDU transmission, the value

    ofRi(n) is used as an index of the current channel packet

    error rate, k+ 1 is used as an index for newly arrivingstream, and Kis the total number of new flows, S I is the

    service interval.

    Through dynamically adjusting channel access param-

    eters, it is possible to provide the guaranteed QoS for

    admitted real-time traffic while maintaining good channel

    utilization. However, If the chosen increment or decrement

    of the channel access parameters is too large, the systems

    will oscillate, if it is too small, the system will take a long

    time to reach the optimal status.

    IV. ADMISSION C ONTROLF OR HCCA

    Due to the centralized control of HCCA, it has higher

    complexity and inefficiency for normal data transmissions.

    Thus, there is not much research on the admission control

    issue in it. But HCCA is efficient in handling time-

    bounded multimedia traffic and can provide absolute QoS

    guarantee, so it is necessary to study it and its admission

    control scheme. A simple admission control has been

    developed as a reference in the 802.11e standard [1]. This

    reference scheme only works fine for CBR traffic other

    than VBR traffic.

    A. Physical-Rate and Contention-Window-Based Admis-

    sion Control

    This scheme [14] uses the current rate transmission

    of QSTAs instead of the minimum rate transmission

    according to their positions for calculating the load of

    network (L), such as L=36Mbps if the position p of

    the station from the QAP is 5 meters; L=11Mbps if

    5

  • 8/13/2019 05235663 Quan Trong

    4/4

    Table ICOMPARISON OF A DMISSIONC ONTROL M ETHODS

    Index

    AdmissionMethod Measurement-

    basedModel-based

    Hybrid-based

    HCCA

    Complexity Simple Complex ComplexMoreCom-plex

    Theoretical Founda-tion

    Worse Better Better Worse

    Network Utilization High Low Higher High

    Flexibility Good Poor Better Good

    Expansibility Good Good Good Worse

    at dynamically adjusting parameters based on channel

    condition but difficult to choose a proper threshold and

    lack of theoretical foundation; Model-based admission

    control schemes are based on some analytical models

    and it is possible to optimize the entire system, but are

    weak at dynamically adjusting the channel parameters to

    achieve optimal channel utilization; A joint measurement

    and model methods can avoid mentioned limitations. For

    HCCA there is not much research work on the admission

    control due to its centralized control. The detail compari-

    son is shown as table I. So, we should take into account to

    use new theoretical analysis methods such as cross-layer

    [15] [16] and game theoretical[17][18] for satisfying QoS

    of traffics maximally.

    V I. CONCLUSION

    In this paper we have surveyed various admission con-

    trol schemes for both EDCA and HCCA. Although manyadmission control schemes have been proposed for them,

    there are more or less limitations to support QoS especially

    multimedia traffic in WLANs. Many challenges remain

    for them. We should decrease the assumption to make the

    admission decision accurate. Especially, we need to know

    how to ensure end-to-end QoS, how to optimally map

    the QoS requirements between different network layers,

    how to map the QoS from application into the channel

    access parameters, and how to dynamically adjust the

    QoS on upper layers while underlying network condition

    changes. For HCCA it is not clear how to optimally

    select effective TXOPs to trade off between flow QoSand network utilization. So, cross-layer frame work and

    game theoretical analysis methods become more and more

    important field of In the future, we will study admission

    control schemes under heterogeneous wireless networks.

    ACKNOWLEDGMENT

    This work is supported by the Science Research, Foun-

    dation of MOE of China under Grant No.209101, and CQ

    CSTC.

    REFERENCES

    [1] IEEE std 802.11e, Wireless LAN Medium Access Control(MAC) and Physical Layer (PHY) specifications: MediumAccess Control (MAC) Quality of Service Enhancements,2005.

    [2] D. Gao and J. F. Cai, Admission Control in IEEE 802.11eWireless LANs, IEEE Network, 2005.

    [3] M. Ahmed, Call admission control in wireless networks:a comprehensive survey, IEEE Communications Surveys &Tutorials, 2005.

    [4] X. Chen, C. Wang, D. Xuan, and Z. Li, Survey on QoSmanagement of VoIP, ICCNMC. Shanghai, China, 2003.

    [5] T. Tugcu,Resource Management and Connection AdmissionControl in Wireless Networks, Network Theory and Appli-cations, 2006.

    [6] S. Nor, A. Mohd, and C.Cheow,AN ADMISSION CONTROLMETHOD FOR IEEE 802.11e, Network Theory and Appli-cations, 2006.

    [7] A. Andreadis, G. Benelli, and R. Zambon, An AdmissionControl Algorithm for QoS Provisioning in IEEE 802.11eEDCA, 3rd ISWPC. Santorini, Greece, 2008.

    [8] X. Chen, H. Zhai, and X. Tian, Supporting QoS in IEEE802.11e Wireless LANs, IEEE Transactions on WirelessLANs Communications, 2006.

    [9] C. Cano, B. Bellalta, Flow-Level Simulation of Call Ad-mission Control schemes in EDCA-based WLANs, the 8thCOST 290 Management Committee Meeting Universidad deMalaga, 2007.

    [10] B. Bellalta, M. Meo, and M. Oliver, VoIP Call AdmissionControl in WLANs in Presence of Elastic Traffic, IEEEJournal of Communications Software and Systems, 2007.

    [11] B. Bensaou, Z. Kong, and D. Tsang, A Measurement-Assisted, Model-Based Admission Control Algorithm for

    IEEE 802.11e, The International Symposium on ParallelArchitectures, Algorithms, and Networks. Sydney, Australia,2008.

    [12] Y. Liu, S. Pawar, C. Assi, and A. Agarwal, DynamicAdmission and Congestion Control for Real-time Traffic inIEEE 802.11e Wireless LANs, WiMob2006. Canada, 2006.

    [13] P. Chatzimisios, A. Boucouvalas, and V. Vitsas, IEEE802.11 Packet Delay-A Finite Retry Limit Analysis, GLOBE-COM2003. San Francisco, USA, 2003.

    [14] F. Didi, M. Feham, H. Labiod, and G. Pujolle, DynamicAdmission Control Algorithm for WLANs 802.11, 3rd In-ternational Conference on Information and CommunicationTechnologies: From Theory to Applications. Damascus,Syria, 2008.

    [15] I. Alocci, G. Gianbebe, and Y. Koucheyavy, Optimizationof IEEE 802.11e Access Class Parameters with Cross-layerApproach, 2nd ISWPC. Damascus, Syria, 2007.

    [16] W. Tao, S. Hua, A Novel Cross-Layer Flow ControlScheme In Ad Hoc Network And its Performance Evaluation,WiCom. Shanghai, China, 2007.

    [17] P. Nuggehalli, M. Sarkar, K. Kulkarni, and R. Rao, AGame-theoretic Analysis of QoS in Wireless MAC, IEEEINFOCOM 2008 proceedings. Arizona, USA, 2008.

    [18] L. Zhao, X. Zou, J. Zhang, H. Zhang, and R. Kwan,Game-

    theoretic Cross-layer Design in WLANs, IEEE InternationalWireless Communications and Mobile Computing Confer-ence. Crete Island, Greece, 2008.

    233233