An Admission Control Strategy for Differentiated
Services in IEEE 802.11
Yu-Liang Kuo
∗, Chi-Hung Lu
∗, Eric Hsiao-Kuang Wu
†, and Gen-Huey Chen
∗∗ Dept. of Computer Science and Information Engineering, National Taiwan University, Taipei, Taiwan, R.O.C. †Dept. of Computer Science and Information Engineering, National Central University, Chung-Li, Taiwan, R.O.C.
Abstract— With the provisioning of high-speed wireless LAN
(WLAN) environments, traffic classes (e.g., VoIP or video-conference) with different QoS requirements will be introduced in future WLANs. The IEEE 802.11e draft is currently standardiz-ing a distributed access approach, called the enhanced distributed coordination function (EDCF), to support service differentiation in the MAC layer. However, since each mobile station transmits data packets egotistically in a distributed environment, the QoS requirement of each traffic class may not be guaranteed. In this paper, we develop an admission control strategy to guarantee the QoS requirement of each traffic class. In order to provide a criterion for admission decision, we introduce an analytical model for EDCF to evaluate the expected bandwidth and the expected packet delay of each traffic class. The admission control strategy uses the performance measures derived from the analytical model to decide if a new traffic stream is permitted into the system. We validate the accuracy of the analytical model by using the ns-2 simulator. Some performance evaluations are also demonstrated to illustrate the effect of the proposed admission control strategy.
I. INTRODUCTION
Recent developments in the IEEE 802.11 standardizations [2] have been successful to offer high-speed data services. Hence, traffic classes (e.g., VoIP or video-conference) with different QoS requirements will be provided in future WLANs. Since these traffic classes require distinct specific features, such as delay-sensitivity or required bandwidth, it is desired to provide a service differentiation mechanism in the IEEE 802.11 standard. Providing service differentiation in a wireless environment requires the MAC protocol to support some degree of separation between different traffic classes. Recently, the IEEE 802.11e draft is specifying a distributed access approach, called EDCF, to support service differentiation in the MAC layer [4]. It ensures that the packets sent by each mobile station can be differentiated by assigning different access parameters.
However, supporting service differentiation in the MAC protocol does not guarantee that the QoS requirement of each traffic class will be fully satisfied. Since each mobile station may transmit packets egotistically in a distributed environ-ment, radio performance might be led to an unacceptable level. An admission control strategy might alleviate the effect of egotistic transmissions. It could ensure that the acceptance of a new traffic stream will not cause the QoS of any ongoing sessions below an unacceptable level.
In this paper, we develop an admission control strategy to guarantee the QoS requirement of each traffic class. The service differentiation mechanism is based on EDCF and
the environment we consider is a single cell coordinated by an access point (AP). In the proposed admission control strategy, mobile stations make use of some MAC management messages specified in the IEEE 802.11e draft to transmit load conditions to the corresponding AP. The AP executes the admission control algorithm to estimate the performance of resource usage and decides if a new traffic stream is permitted into the system. In order to provide a criterion for admission decision, we introduce an analytical model to evaluate the performance of resource usage in EDCF. This model can evaluate the expected bandwidth and the expected packet delay for each traffic class. We validate the accuracy of the model and evaluate the performance by using the ns-2 simulator [1]. Since the model is simple in calculating the desired results, the admission decision can be made in real time. The performance evaluations for EDCF have appeared in the recent literatures [8][9][10], but the results are via simulations, and no theoretical analysis is given. The model we derived also provides a theoretical analysis for EDCF.
The paper is organized as follows. Section II presents the distributed coordination function (DCF) in IEEE 802.11 and the EDCF that is currently specified in the IEEE 802.11e draft. Section III presents the analytical model for the service differentiation mechanism. The validation of our analytical model is also presented in Section III. We present the proposed admission control strategy and simulation results in Section IV. The conclusions are drawn in Section V.
II. DCFANDEDCF A. Distributed Coordination Function
The fundamental access method in IEEE 802.11 is DCF, which is based on carrier sense multiple access with collision avoidance (CSMA/CA) protocol. A mobile station that intends to transmit a packet waits until the channel is sensed idle for a time period equal to the DCF interframe space (DIFS). If the channel is sensed idle for a duration of DIFS, then it can immediately transmit a packet. Otherwise, the mobile station will generate a backoff time counter. A discrete time counter is used and the time following an idle DIFS is slotted. A mobile station is allowed to transmit only at the beginning of a time slot.
When a mobile station senses the channel busy during the duration of DIFS, the backoff time counter is randomly selected from the range (0, CW), where CW is called the contention window. At the first transmitting attempt, CW is
TABLE I
THEFOURDEFAULTACSSPECIFIED IN THEIEEE 802.11E
AC AIFSD (AIFS) CWmin CWmax(m) Designation
AC0 34µs (2) 15 1023 (6) Best Effort
AC1 25µs (1) 15 1023 (6) Video Probe
AC2 25µs (1) 7 15 (1) Video
AC3 25µs (1) 3 7 (1) Voice
assigned the value CWmin, which is called the minimum con-tention window. In the consecutive unsuccessful transmissions (due to collisions), the value of CW is increasing up to the maximum value CWmax=2mCWmin, where m is called the maximum backoff stage.
The backoff time counter is decreased as the channel is sensed idle and suspended as the channel is sensed busy. After the suspension, the counter is reactivated as the channel is again sensed idle for a duration of DIFS. The mobile station will transmit a packet when the counter reaches zero. B. Enhanced Distributed Coordinated Function
The goal of EDCF is to provide a distributed access mech-anism to support service differentiation. EDCF introduces the concept of access categories (ACs), which are variants of the DCF access mechanism. The IEEE 802.11e draft currently specifies four default ACs. The four default ACs are listed in Table I in which the IEEE 802.11a physical layer [3] is adopted. Different ACs use different values of arbitration inter-frame space duration (AIFSD), CWmin, and CWmax. Traffic classes with smaller values of CWmin and CWmax yield higher priorities. Furthermore, different interframe spaces can be used by different traffic classes. DCF Interframe space (DIFS) is substituted for the AIFSD. AIFSD is at least a duration of short interframe space (SIFS) plus a slot time and can be enlarged individually by different traffic classes. Let the length of a slot time be denoted byδ. AIFSD can be computed as the following:
AIFSD = SIFS + AIFS × aSlotTime,
where AIFS is a positive integer which is equal to or greater than 1. Hence, AIFSD is determined by AIFS and traffic classes with smaller values of AIFS yield higher priorities. These three parameters are gathered and called a QoS param-eter set.
In EDCF, data packets are delivered through multiple back-off instances within one mobile station. A single mobile station may implement up to 4 transmission queues and each trans-mission queue uses a specific AC for contending the channel access, as illustrated in Fig. 1. Each queue within the mobile station contends for the channel access and independently starts its backoff procedure depending on its associated AC. If the backoff time counters of two or more parallel queues within a single mobile station reach zero at the same time, a internal scheduler will resolve the internal collision. The
Mobile station
Internal Scheduler
Wireless Channel
AC 0
Mapping to Access Category
Transmission queues Per-queue channel access functions AC 3 AC 2 AC 1
Fig. 1. A single mobile station can implement up to 4 transmission queues. Each queue is mapping to a particular access category.
scheduler grants the channel access to the queue in terms of its particular scheduling algorithm.
III. ANALYTICALMODEL ANDMODELVALIDATION
The environment we consider is a single wireless cell coordinated by an AP. In a single cell environment, each mobile station which intends to transmit a packet needs to forward its packet to the corresponding AP. The transmitted packets should be forwarded to the AP even if the packets are destined to the mobile station in the same cell. We assume all mobile stations and the corresponding AP can communicate with each other without obstacle and there is no hidden ter-minal problem. The access mechanism we consider is a four-way handshaking protocol by using the RTS/CTS/DATA/ACK dialogue.
A. Analytical Model
Without loss of generality, we assume there are K traffic classes with distinct QoS requirements. Specifically, there are
n = (n0, n1, . . . , nK−1) mobile stations, where nk (0 ≤
k ≤ K − 1) is the number of mobile stations which generate traffic classk packets. For convenience, let the traffic class k packet be denoted by class-k packet and the class-k station is referred to as the mobile station that generates class-k packets to transmit. Traffic class k stations use ACk to access the wireless channel. We assume that each class-k packet has constant length (number of bits)Lk and requires Lk
M seconds for data transmission time, where M is the average channel bit rate. We also assume that each mobile station always has a packet ready to transmit. In other words, we consider the saturation condition [6]. The maximum propagation delay for all packets is assumed to be a constant length ofτ seconds.
At each transmission attempt, we assume that each class-k packet has common probability pk of involving in collision when it is being transmitted, where pk is independent of retransmission history. The assumption of pk is originated from Bianchi’s model [6] but we extend the assumption to allowK traffic classes. Suppose that a class-k station that has involved in l times of collision will select the backoff time
counter Bkl (l ≥ 0) before (re)transmission. Let Jk denote
the number of collisions that a class-k station has involved. As a class-k station intends to transmit a packet, the expected backoff time counterE[Bk] can be computed by conditioning on Jk E[Bk] = ∞ l=0 E[Bkl]P {Jk= l}. (1) The distribution ofJk is P {Jk= l} = pl k ifl ≥ 1, 1 −∞i=1pi k ifl = 0. (2) Let Wk, AIFSk, and mk be the minimum contention window, AIFS, and maximum backoff stage for each class-k station. Since Bkl is selected from the current contention
window (the current contention window is obtained according to the binary exponential backoff procedure) in a uniformed way, its probability mass function is
P {Bkl = i} =
1
2βk(l)(Wk+ 1) + AIFSk, (3)
i = 0, 1, · · · , 2βk(l)(Wk+ 1) + AIFSk− 1,
where β(·) is defined as the following: βk(i) =
i if 0 ≤ i ≤ mk− 1, mk if i ≥ mk.
The expectation ofBkl can be easily derived from (3)
E[Bkl] =
2βk(l)(Wk+ 1) + AIFSk− 1
2 . (4)
According to (2) and (4), (1) can be expressed as E[Bk] = (1 − ∞ i=1 pi k)Wk+ AIFS2 k + ∞ i=1 pi k2 βk(i)(Wk+ 1) + AIFSk− 1 2 . (5)
At a given time slot, the probability that a class-k station will transmit is [7]
qk =E[Bk] + 11 . (6)
According to (6), pk can be computed as pk= 1 − (1 − qk)nk−1 K−1 j=0,j=k (1 − qj)nj . (7) By observing the transmission behavior in the wireless channel, a pattern of periodical cycles can be found. Each cycle, named as transmission cycle, consists of some idle periods, some unsuccessful periods (due to collision or error), and a successful period, as depicted in Fig. 2. When each cycle ends in a successful period, a consecutive cycle will restart with respect to the ordered sequences (idle periods, unsuccessful periods, a successful period). Let E[I], E[C], and E[S] be the sum of the expected lengths of all idle
Collision Collision ... Success
AIFSD AIFSD Transmission Cycle Idle period due to backoff Idle period due to backoff Idle period due to backoff
Fig. 2. A transmission cycle.
periods, all unsuccessful periods, and the expected length of a successful period, respectively. In addition, letE[Lk] be the expected proportional number of bits a class-k packet will be transmitted in a successful period. As a result, using renewal theory [11], the expected bandwidth for class-k stations is given as
ρk =E[I] + E[C] + E[S]E[Lk] . (8) The probability that a class-k packet will be transmitted in the successful period of a transmission cycle is
γk = P{Transmitting station = 1 and is class-k station| Transmitting station≥ 1}. (9) Hence (9) can be computed as
γk= nkqk(1 − qk)
nk−1 K−1
i=0,i=k(1 − qi)ni
1 − K−1i=0 (1 − qi)ni . (10)
Since only one particular class-k packet can be transmitted in the successful period of a transmission cycle, we compute E[Lk] according to the normalized probability γk. Thus
E[Lk] =γ γk
0+ γ1+ · · · + γK−1Lk. (11)
Also, the expected proportional AIFS a class-k packet required in the successful period of a transmission cycle is given as
E[AIFSk] =γ γk
0+ γ1+ · · · + γK−1AIFSk. (12)
The expected number of bits transmitted in a successful period is computed asE[L] =K−1k=0 E[Lk] and the expected AIFS required in a successful period is computed as E[A] = K−1
k=0 E[AIFSk].
Let TSIF S, TACK, TRT S, TCT S, and H be the duration required for a SIFS, transmitting a ACK, transmitting a RTS, transmitting a CTS, and transmitting a PHY/MAC header, respectively. The expected length of a successful period in a transmission cycle is approximately expressed as
E[S] = TRT S+TCT S+TACK+E[L]M +4τ+4TSIF S+δE[A]+H. (13) Given n= (n0, n1, . . . , nK−1), we let Nc be the random variable representing the number of colliding periods needed in a transmission cycle. The distribution of Nc is geometric with parameter (1 − pc), where pc is the collision probability viewed by the channel and can be computed as
P {Transmitting station ≥ 2|Transmitting station ≥ 1} = 1 − P {Transmitting station = 0} − P {Transmitting station = 1} P {Transmitting station ≥ 1} .
From (14), pc can be expressed as pc =1 − K−1 i=0 (1 − qi)ni− X 1 − K−1i=0 (1 − qi)ni , (14) where X =K−1i=0 niqi(1 − qi)ni−1 K−1 j=0,j=i(1 − qj)nj. As a result, we have E[Nc] = 1 − pcpc . (15) The sum of the expected lengths of all unsuccessful periods in a transmission cycle is
E[C] = E[Nc](τ + TSIF S+ δE[A] + TRT S+ H). (16) This follows since the length of an unsuccessful period isτ + TSIF S + δE[A] + TRT S + H and the expected number of unsuccessful periods isE[Nc].
The number of time slots required for an idle period in a transmission cycle can be viewed as a geometric random variable with parameter (1 − Ki=1(1 − qi)ni). Hence, the
expected length of an idle period can be computed as δ∞ j=0 j(1 −K−1 i=0 (1 − qi)ni)( K−1 i=0 (1 − qi)ni)j = δ K−1i=0 (1 − qi)ni 1 − K−1i=0 (1 − qi)ni . (17)
The lengths of all idle periods are assumed to be identical and independent distribution and there areE[Nc] + 1 idle periods in a transmission cycle. The sum of the expected lengths of all idle periods in a transmission cycle is
E[I] = (E[Nc] + 1) δ K−1i=0 (1 − qi)ni 1 − K−1i=0 (1 − qi)ni . (18) In this paper, the expected packet delay for a class-k packet is defined as the time between the arrival of a class-k pacclass-ket and its successful transmission to the receiver. The expected packet delay is composed of the access delay and the transmission delay. The access delay is the time interval from the packet arriving at the station to the packet being ready for acquiring the channel access. The transmission delay is the time interval from the packet being ready for acquiring the channel access to the packet being successfully received by the receiver. We assume that all stations have infinite queue length. Hence we can compute the expect packet delay for a class-k packet by using G/G/1 model [5]. The access and transmission delays of a class-k packet can be viewed as waiting time in queue and service time respectively in a queueing system. According to G/G/1 model, the average waiting time in queue (access delay) for a class-k packet satisfies
E[Qk] ≤ λk(σ
2
ak+ σ2bk)
2(1 − λkE[Tk]), (19) where
• λk - average packet inter-arrival time for class-k packets • E[Tk] - average transmission delay for a class-k packet,
TABLE II
THE VALUES OF PARAMETERS USED IN MODEL VALIDATION
FTP packet payload size (AC0) 1500 bytes
Video packet payload size (AC2) 1464 bytes
Voice packet payload size (AC3) 96 bytes
average packet inter-arrival timeλ0for FTP 0.012 pkt/sec average packet inter-arrival timeλ2for video 0.01 pkt/sec average packet inter-arrival timeλ3 for voice 0.02 pkt/sec
PHY header 6 bytes
MAC header 34 bytes
RTS 20 bytes
CTS 14 bytes
ACK 14 bytes
Propagation delay 1µs
Average channel bit rate 1 Mbps
SIFS 16µs
Slot time 9µs
channel bit error rate 10−5
• σ2ak - variance of packet inter-arrival times for class-k packets,
• σ2bk - variance of transmission delays for class-k packets. Assuming that the channel bit error rate in the wire-less medium is pb. The successful transmission proba-bility without error-prone for a class-k packet is (1 − pb)Lk. E[Tk] can be verbally represented as (The time
re-quired for retransmission)×(Average number of retransmis-sion)+(Successful data transmission time). As a result, we have E[Tk] = akE[Jk] + (1 − pb)Lkbk+ (1 − (1 − pb)Lk)E[Tk]
= (1 − pak
b)LkE[Jk] + bk. (20)
where ak = δE[Bk] + TRT S + τ + 2TSIF S + δAIFSk and bk = δE[Bk] + H + δAIFSk+ TRT S+ TCT S+LMk+ TACK+ 4TSIF S+4τ. According to (20) and (21), the expected packet delay for a class-k packet can be expressed as
E[Dk] = E[Qk] + E[Tk]. (21) B. Model Validation
We validate the analytical model by using thens-2 simula-tor. The values of parameters used in the analytical and sim-ulative models are summarized in Table II. In this validation, each class-k packet has constant packet payload size, as shown in Table II. We have three common applications (FTP, video, and voice applications) and each one is associated with default AC specified in the IEEE 802.11e draft. FTP, video, and voice applications are associated with AC0, AC2, AC3 respectively for acquiring channel access. The values of parameters are assigned depending on the IEEE 802.11a specification [3]. For simulation efficiency and simplicity, the average channel bit rate is assumed to be 1 Mbps.
In the model validation, we have three mobile stations. These three mobile stations perform FTP, video, and voice ap-plications, respectively. Fig. 3(a) shows the bandwidth results
0 5 10 15 0 100 200 300 400 500 600 700 800 900 1000 Time (sec) Bandwidth (Kbps) Voice (simulation) Video (simulation) FTP (simulation) Voice (analysis) Video (analysis) FTP (analysis)
(a) Expected bandwidth.
0 5 10 15 0 100 200 300 400 500 600 700 800 900 1000 Time (sec) Bandwidth (Kbps) Voice (simulation) Video (simulation) FTP (simulation) Voice (analysis) Video (analysis) FTP (analysis)
(b) Expected packet delay. Fig. 3. Analytical model vs. simulative model (one FTP, one video, and one voice applications).
0 20 40 60 80 100 120 140 160 180 200 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Time (s) Throughput
With admission control Without admission control
(a) Throughput for traffic class 2.
0 20 40 60 80 100 120 140 160 180 200 0 10 20 30 40 50 60 70 80 90 100 Time (s)
Expected packet delay (ms)
With admission control Without admission control
(b) Expected packet delay for traffic class 2. Fig. 4. Performance evaluations for the proposed admission control strategy.
of the simulative model, in which diamond line represents simulative bandwidth of video, square line represents simu-lative bandwidth of FTP, and circle line represents simusimu-lative bandwidth of voice, are close to the expected bandwidth results of the analytical model (solid lines).
The packet inter-arrival times of FTP, video, and voice ap-plications follow the exponential distribution, general distribu-tion, and deterministic distribudistribu-tion, respectively. The average packet inter-arrival times for these three applications are listed in Table II. Fig. 3(b) shows the delay results of the simulative and analytical models. The validation shows that the expected packet delays of the analytical model (solid lines) are also close to the one of simulative model.
IV. THEPROPOSEDADMISSIONCONTROLSTRATEGY AND
SIMULATIONRESULTS
A. The Proposed Admission Control Strategy
We assume that the QoS requirement for traffic class k consists of the following items:
• minimum data rate (MinDRk), • maximum data rate (MaxDRk),
• maximum tolerable packet delay (Delayk), and • maximum tolerate packet loss rate (LRk).
The average data rate can be computed as AvgDRk =
MaxDRk+MinDRk
2 . The objective of admission control is to
guar-antee the required bandwidths of all traffic class are satisfied. In addition, the expected packet delay for each traffic classk will not exceed Delayk.
The proposed admission control algorithm is implemented in APs. The AP should collect information of load conditions from mobile stations to estimate the radio performance. We make use of some MAC management messages specified in the IEEE 802.11e draft to transmit the required information. Each mobile station first transmits a ADDTS request message to the AP before transmitting a traffic stream (TS). A TS is a set of certain traffic class packets. The ADDTS request message contains the traffic class information of the TS and the MAC address of this mobile station.
Upon receipt of the request message, the AP estimates the radio performance by using the analytical model described in III-A. The proposed algorithm use the performance measures derived by the analytical model to see if the QoS requirements for all mobile stations can be guaranteed after the TS is permitted into the system. As ak-class station issues a ADDTS request message, the admission control algorithm will be triggered, as shown in Algorithm 1. If the QoS requirements are guaranteed for all mobile stations, then the AP responds a ADDTS response message with acceptance status to notify the mobile station of approving the TS into the system. Otherwise, the AP will respond a ADDTS response message with rejection status to notify the mobile station of rejecting the TS into the system. After the mobile station receives the ADDTS response message, it can transmit packets until there is no packet in the transmission queue of the mobile station. As the transmission queue of the mobile station gets empty, it will transmit a DELTS message to notify the AP of successful transmission of the TS. Upon receipt of the DELTS message, the AP responds a ACK to confirm the receipt. When there is another TS to be transmitted in the transmission queue, it should again issue a ADDTS request message for requesting admission. According to these messages (ADDTS request and DELTS message), the AP can obtain load conditions in the wireless medium and make precise estimation for the radio performance.
Algorithm 1 Admission Control Input:n0, n1, · · · , nK−1
nk ← nk+ 1
Estimating ρi andE[Di] for each traffic class i (0 ≤ i ≤ K − 1)
fori = 0 to K − 1 do
if Traffic class i is not for best effort traffic then if ρi≤ AvgDRi ×(1−LRi) orE[Di] ≥ Delayi then
Reject the ADDTS request nk ← nk− 1
Exit
end if end if end for
Accept the ADDTS request
B. Simulation Results
We made an experiment to demonstrate the effect of our pro-posed admission control strategy. The values of parameters in the simulation environment are the same as model validation, which is listed in Table II. We consider the scenario that the number of the video flow is increasing, while the numbers of FTP and voice flows stay constant.
In this simulation, the traffic load is increasing by adding a new video flows periodically every 5 seconds. The solid line depicted in Fig. 4(a) represents the throughput for video applications with admission control, while the circle line rep-resents the one without admission control. With the increment
in the number of video flows, the admission control strategy could keep the throughput stable. Similarly, with applying the admission control strategy, the expected packet delay is stable, which is independent of load conditions, as depicted in Fig. 4(b).
V. CONCLUSION
The contribution of this paper is as follows. First, we introduced an analytical model to evaluate the expected band-width and the expected packet delay of each traffic class for the service differentiation mechanism based on EDCF. This model provides a criterion for admission decision as well as a theoretical analysis for EDCF. We validated this model by using the ns-2 simulator and the validation showed that the estimation of the analytical model is accurate. Second, the proposed admission control strategy satisfies the required bandwidth of each traffic class and limits the packet delay of each traffic class to a predefined level which is independent of load conditions. Therefore, the diverse QoS requirements in each traffic class can be fulfilled. In addition, the admission decision can be made in real time due to the simplicity of the analytical model. In the further research, we will investigate the effect of the hidden terminal problem and the dynamic tuning of ACs to optimize the system performance under various network conditions.
ACKNOWLEDGMENT
This research was supported by Ministry of Education, ROC, with contract No: 91 A-H-FA07-1-4 (Learning Tech-nology) and the Communications Software Technology project of Institute for Information Industry and sponsored by MOEA ,R.O.C
REFERENCES
[1] ”The Network Simulator - ns-2,” http://www.isi.edu/nsnam/ns/. [2] Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY)
Specification, IEEE 802.11 WG, Aug. 1999.
[3] Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY)
Specifications: High-speed Physical Layer in the 5GHz Band, IEEE Std
802.11a-1999, Sep. 1999.
[4] Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY)
Specification: Medium Access Control (MAC) Enhancements for Quality of Service (QoS), IEEE Std. 802.11e/D4.0 , Nov. 2002.
[5] D. Bertsekas and R. Gallager, Data Networks, (Englewood Cliffs, NJ: Prentice Hall, 2nd ed., 1992.)
[6] G. Bianchi, ”Performance analysis of the IEEE 802.11 distributed coor-dination function,” IEEE Journal of Selected Areas on Communication vol.18, no.3, pp. 535-547, March 2000.
[7] F. Cali, M. Conti and E. Gregori, ”Dynamic Tuning of the IEEE 802.11 Protocol to Achieve a Theoretical Throughput Limit,” IEEE/ACM Trans.
on Networking, vol. 8, pp. 785-799, Dec. 2000.
[8] P. Garg, R. Doshi, R. Greene, M. Baker, M. Malek, and X. Cheng, ”Using IEEE 802.11e MAC for QoS over Wireless.” to appear in Proc.
IPCCC, Phoenix, Arizona, Apr. 2003.
[9] A. Lindgren, A. Almquist, O. Schelen, ”Evaluation of Quality of Service Schemes for IEEE 802.11 Wireless LANs,” in Proc. LCN, Tampa, Florida, Jul. 2001, pp. 348-351.
[10] S. Mangold, S. Choi, P. May, O. Klein, G. Hiertz, L. Stibor, ”IEEE 802.11e Wireless LAN for Quality of Service,” in Proc. European
Wireless, vol. 1, Florence, Italy, Feb. 2002, pp. 32-39.
[11] S. M. Ross, Introduction to Probability Models, (Academic Press, 8nd ed., 2003.)
