Dynamic admission control and bandwidth reservation for IEEE 802.16e mobile WiMAX networks

R E S E A R C H

Open Access

Dynamic admission control and bandwidth

reservation for IEEE 802.16e mobile WiMAX

networks

Chiapin Wang

, Wan-Jhen Yan and Hao-Kai Lo

Abstract

The article presents a dynamic connection admission control (CAC) and bandwidth reservation (BR) scheme for IEEE 802.16e Broadband Wireless Access networks to simultaneously improve the utilization efficiency of network resources and guarantee QoS for admitted connections. The proposed CAC algorithm dynamically determines the admission criteria according to network loads and adopts an adaptive QoS strategy to improve the utilization efficiency of network resources. After new or handoff connections enter the networks based on current admission criteria, the proposed adaptive BR scheme adjusts the amount of reserved bandwidth for handoffs according to the arrival distributions of new and handoff connections in order to increase the admission opportunities of new connections and provide handoff QoS as well. We conduct simulations to compare the performance of our proposed CAC algorithm and BR scheme with that of other approaches. The results illustrate that our approach can effectively improve the network efficiency in terms of granting more connections by as large as about 22% in comparison with other schemes, and can also guarantee adaptive QoS for admitted new and handoff connections. Keywords: IEEE 802.16e WMAN, connection admission control, bandwidth reservation, resource allocation.

1. Introduction

Broadband wireless access networks have rapidly been growing in these years to support the increasing demands of wireless multimedia services, like streaming audio/video, Internet Protocol TV, and video conferen-cing. Mobile Worldwide Interoperability for Microwave Access (WiMAX), which has been standardized by IEEE 802.16e [1], is one of the most promising solutions to provide ubiquitous wireless access with high data rates, high mobility, and wide coverage. The IEEE 802.16e Media Access Control (MAC) layer provides differential Quality of service (QoS) for various classes of scheduling services, which are Unsolicited Grant Service (UGS), Extended Real-Time Polling Service (ertPS), Real-Time Polling Service (rtPS), Non-real-time Polling Service (nrtPS), and Best Effort (BE). Each scheduling class is associated with a set of QoS parameters for quantifying its bandwidth requirement, e.g., maximum/minimum data rates and maximum delays. The radio resources (i.

e., time slots and frequency spectrums) for different scheduling services are centrally controlled by the base station (BS). To provide QoS for data transmissions in WiMAX networks, BS generally applies a Connection Admission Control (CAC) scheme which determines whether a new connection should be established accord-ing to the available network resources. Essentially, the effectiveness of CAC schemes can be critical to both the performances of QoS for admitted connections and the utilization efficiency of network resources. However, the IEEE 802.16e standards do not specify how to imple-ment CAC mechanisms and remain that as open issues.

On the other hand, a bandwidth reservation (BR) mechanism is also important to the provisioning of QoS for some prioritized users like users in a handoff pro-cess. Handoff occurs when mobile station (MS) transfers its connection from the original serving BS with worse and worse link qualities to a neighboring BS with better qualities. In general, a handoff user will be prioritized over a new incoming user in order to provide better user-perceived satisfaction especially when it is with real-time applications which have specific QoS * Correspondence: chiapin@ntnu.edu.tw

Department of Applied Electronic Technology, National Taiwan Normal University, Taipei, Taiwan

© 2012 Wang et al; licensee Springer. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

requirements, e.g., throughput demands and delay/jitter constraints. Since the reserved bandwidth cannot be taken by a new coming user, the design of BR mechan-isms can significantly affect the performance of handoff QoS and also the utilization efficiency of network resources.

The CAC and BR problems have largely been investi-gated in previous study [2-20]. The authors of [2,3] pro-pose to adopt minimum bandwidth requirements as the admission criteria for all classes of scheduling services. The approach can provide more connections admitted into networks but may cause a relatively low QoS perfor-mance. The authors of [10] propose to divide the sche-duling services into two groups: one group consists of UGS, ertPS, and rtPS which adopt maximum bandwidth requirements for the admission criteria, while another group consists of nrtPS and BE which adopt minimum bandwidth requirements. The approach may over favor the higher-class services and cause a starvation of lower-class services. Instead of using fixed criteria for an admis-sion control as described above, the studies in [11,12] propose to dynamically determine the admission criteria by using a game-theoretic approach. However, it does not take the network load into consideration and may introduce great computational complexities.

With regard to the BR schemes, a fixed guard channel scheme [13] is proposed to reserve a certain amount of bandwidth for upcoming handoff connections to assure seamless handoff processes. When the total bandwidth utilization of existing users reaches the threshold, no more new connections can be admitted into the net-work. Nevertheless, when a fixed amount of bandwidth can never be used for new connections, a certain por-tion of network resources may be wasted. The study [10] proposes to dynamically adjust the quantity of reserved bandwidth based on the arrival and departure behavior of handoff connections to make the resource utilization more efficient. However, if handoff connec-tions occur infrequently, the quantity of reserved band-width for handoffs is almost fixed and this approach would be similar to the fixed guard channel scheme and cause a waste of network resources as well.

Both CAC schemes and BR mechanisms are important research issues in wireless networks due to scarce radio resources, dynamic channel qualities, and diverse user demands. However, to the best of authors’ knowledge, most efforts tackle one of the two problems individually while little work considers the joint design of the two mechanisms. We are thus motivated to present a joint design of CAC and BR mechanisms which aim at simul-taneously improving the utilization efficiency of network resources and guaranteeing QoS for admitted new con-nections and handoff concon-nections. The proposed CAC scheme dynamically determines the admission criteria

according to network loads and adopts an adaptive QoS strategy to increase the amount of admitted connections for the network efficiency. The key idea of our CAC scheme is based on the fact that most scheduling services are with adaptive QoS requirements, e.g., maximum and minimum rates. Therefore, the admission criteria can be determined according to the amount of available wireless resources for increasing the number of admitted connec-tions with adaptive QoS. For example, if the network capacity is adequate or sufficient, bandwidth require-ments for higher QoS might be adopted as the admission criteria. Alternatively, if the network load is quite heavy, the admission criteria may be degraded to meet lower QoS requirements. After the admission criteria are deter-mined, the proposed BR scheme dynamically adjusts the amount of reserved bandwidth for handoffs according to the arrival distributions of new/handoff connections to increase the connection admission opportunities and also guarantee the bandwidth requirements for handoff QoS. The basic idea of our adaptive BR scheme is a rational inference that generally the occurrences of new incoming connections may be much more frequent than that of handoff connections [21-24]. This observation originates from common BS deployment that the overlap areas of a given BS between its neighboring stations are parts of its coverage area. Since handoffs arise only when users cross through the overlap areas, it is a general situation to observe more new connections occurring than handoff connections. Thus, the optimal BR should take into account the arrival behavior of not only handoff connec-tions, but also new connections in order to avoid a waste of network resource as possible.

We conduct simulations of 802.16e transmission sce-narios to evaluate and compare the performances of the proposed CAC algorithm and BR scheme with that of other approaches. Simulations results illustrate that our approach can effectively improve the network efficiency in terms of increasing the number of granted connec-tions by as large as about 22% in comparison with other schemes, and also can guarantee adaptive QoS for admitted new and handoff connections. The remainder of this article is organized as follow. In Section 2, we briefly illustrate the QoS architecture and resource allo-cation mechanism of IEEE 802.16e networks. Section 3 presents the proposed CAC algorithm and BR scheme. In Section 4, we construct simulation scenarios to demonstrate the effectiveness of our approach. Section 5 draws our conclusions.

2. IEEE 802.16e QoS architecture and resource allocation mechanism

2.1 IEEE 802.16e QoS architecture

The IEEE 802.16e MAC layer provides QoS differentia-tion for various categories of scheduling services. The

IEEE802.16e uplink scheduling framework is shown in Figure 1. The scheduling of uplink packet transmissions is centrally controlled in the BS. The IEEE 802.16e stan-dards adopt a connection-oriented MAC protocol, i.e., each connection is associated with a connection ID. When a service flow generated at the application layer arrives at the MAC layer, the MS first sends a connec-tion establishment request to the BS. The admission control mechanism at BS then estimates whether the remaining bandwidth can support the QoS requirements

of new connections without violating existing users’

QoS. If the connection request is accepted, the BS replies with a connection response which indicates the connection IDs for each direction of this connection. After the process of connection establishment is fin-ished, the MS can issue a bandwidth request. The con-nection classifier then classifies the service data units into different scheduling classes according to their ser-vice flow identifier and connection identifier. The uplink bandwidth requests by users are performed on a per connection basis, whereas the BS grants bandwidth on a per subscriber station basis (GPSS). After the BS allo-cates a certain amount of bandwidth to each of the MSs, the packet scheduler at each MS will redistribute the bandwidth to the corresponding connection. By means of the connection-admission-control mechanism and request-grant bandwidth-allocation scheme, QoS for different scheduling classes can be guaranteed.

The IEEE 802.16e standard divides all service flows into five scheduling classes, each of which is associated with a set of QoS parameters for quantifying its band-width requirement. The five scheduling classes are described as follows.

(1) UGS: UGS is designed to support real-time service flows with fixed-size packets generated at periodic inter-vals (i.e., constant bit rate–CBR), such as T1 services and voice-over-Internet-Protocol (VoIP) applications without silence suppression. This service can grant a fixed amount of bandwidth for CBR real-time applica-tions without any requests.

(2) rtPS: rtPS is designed to support real-time service flows with variable-size packets generated at periodic intervals (i.e., variable bit rate–VBR), such as Motion Pictures Experts Group (MPEG) video. Based on a poll-ing mechanism to request bandwidth periodically, this service can guarantee QoS such as the minimum data rate and maximum latency for VBR real-time applications.

(3) ertPS: The characteristic of this service class is between UGS and rtPS. On detecting that the allocated bandwidth is either insufficient or excessive, ertPS can send a request to change the amount of allocated band-width like rtPS does. Otherwise, if the bandband-width demand remains unchanged, ertPS behaves as UGS. ertPS is designed to support VBR real-time data services such as VoIP applications with silence suppression.

(4) nrtPS: This service class is to support non-real-time VBR services which require minimum-data-rate guarantees but can be tolerant to delay, such as File-Transfer-Protocol (FTP) applications.

(5) BE: The BE service is designed for best-effort applications which have no explicit QoS requirements, e.g., web services or e-mail.

The QoS parameters and the supporting application types associated with each of the IEEE 802.16e schedul-ing classes are shown in Table 1.

2.2. Bandwidth allocation mechanism

The IEEE 802.16e physical layer (PHY) adopts an Orthogonal Frequency Division Multiple Access (OFDMA) slot as the minimum possible resource. The IEEE 802.16e PHY supports Frequency Division Duplex (FDD) and Time Division Duplex (TDD) for bandwidth allocation mechanisms. In FDD mode, the uplink (UL) and downlink (DL) channels are located on split fre-quencies, with which a fixed duration frame is used for both UL and DL transmissions. In TDD mode, the UL and DL transmissions are arranged at different time per-iods using the same frequency. In this article, we focus on the TDD mode for the IEEE 802.16e bandwidth allo-cation mechanism.

In TDD mode, Time Division Multiplexing (TDM) is used for DL transmissions and Time Division Multiple Access (TDMA) is used for UL transmissions. As shown in Figure 2, a TDD frame has a fixed duration and con-tains one DL subframe and one UL subframe whose durations can adapt to the traffic loads of UL and DL transmissions. The DL subframe consists of a preamble, Frame Control Header (FCH), and a number of data bursts. The FCH specifies the profiles of the DL bursts that immediately follow it. The broadcast messages

including downlink map (DL-MAP), uplink map (UL-MAP), DL Channel Descriptor (DCD), UL Channel Descriptor (UCD), etc., are sent at the beginning of these DL bursts. The UL subframe contains a conten-tion interval for initial ranging and bandwidth request and UL PHY protocol data units (PDUs) from different MSs. The DL connections are scheduled by BS in a broadcast manner, while the UL connections apply a request-grant mechanism for bandwidth allocation in a shared manner. The UL bandwidth requests are per-formed on a per connection basis, whereas the BS grants bandwidth on a per subscriber station basis (GPSS). After the BS allocates a certain amount of bandwidth to each of the MSs, each MS will redistribute the band-width to the corresponding connection. The information about bandwidth allocations for DL and UL transmis-sions is broadcast to the MSs through DL-MAP and UL-MAP messages at the beginning of each frame. Therefore, each MS can receive from and transmit data to BS in the predefined OFDMA slots.

2.3. Packet scheduling mechanism

As shown in Section 2.1, the IEEE 802.16e standard defines five scheduling classes. However, it does not spe-cify the scheduling mechanism for the five classes and the design is left for researchers [25]. The design of a scheduling mechanism must take into account the speci-fic QoS constraints of different applications, e.g. the maximum allowable delay and minimum data rate [3]. A feasible solution is to decide on a service class first according to the characteristics of each class and next choose an appropriate user in the selected class [26]. In the second phase, the packet scheduling of different users among a given class may consider some perfor-mance metrics such as throughput and fairness, while the maximum rate scheduling (greedy algorithm) and Proportional Fairness (PF) scheduling can be applied, respectively. The maximum rate scheduling is effective to advance the overall system throughput as it allocates resources to users with relatively good channel qualities among them [27]. On the other hand, the PF scheduling can improve the fairness of channel utilization among users as it distributes resources among them with con-sideration of their previous records of utilization [28-30].

Throughput and fairness, however, are conflicting per-formance metrics [31]. To maximize system throughput, more resources should be allocated to the users in good channel conditions. This may cause most radio resources monopolized by a small number of users, leading to unfairness. On the contrary, if resources are allocated in a fair manner, resources may be allocated to the users with weak channel conditions. This can result in the degradation of system throughput. To escape

Table 1 IEEE802.16e QoS classes

QoS classes

Applications QoS

parameters UGS T1 services, VOIP without silence

suppression

Max Rate Min Rate Jitter

rtPS Video Streaming Max Rate

Min Rate Max Latency ertPS VOIP with silence suppression Max Rate

Min Rate Max Latency Jitter

nrtPS FTP Max Rate

Min Rate

from the “throughput-fairness” dilemma, we can con-sider “utility” for packet scheduling. Utilities are a per-formance metric which can fully represent the degree of user satisfaction for a given application [32,33]. Thus, it is a more appropriate metric for packet scheduling since even users with the same service class actually are with various demands for network resources due to their spe-cific characteristics. This fact implies that resources should be allocated to users according to the application performance metric of“satisfaction” rather than network

performance metrics such as“throughput” or “fairness”

[34].

In this article, we focus on the problems of CAC and BR for mobile WiMAX networks and do not discuss the design of packet scheduling. In the following section, we will propose a joint design of CAC and BR mechanisms as a solution of the addressed problems.

3. Proposed CAC algorithm and BR scheme

3.1. Dynamic CAC algorithm

We design a dynamic CAC algorithm which adjusts the admission criteria depending on the network loads and uses an adaptive QoS provisioning strategy in order to increase the efficiency of bandwidth utilization. The key idea of our dynamic CAC scheme is based on the fact that most scheduling services are with adaptive QoS requirements, e.g., maximum and minimum rates. Therefore, the admission criteria can be adjusted with the amount of available wireless resources to increase the number of admitted connections with adaptive QoS. For example, if the network capacity is adequate or suf-ficient, higher QoS requirements could be adopted as the admission criteria. Alternatively, if the network load

is rather heavy, the admission criteria may level down for lower QoS.

We can take into account several approaches to design the admission criteria according to network loads. A simple way, as shown in Figure 3a, is to adopt the maximum or minimum bandwidth requirement alternatively as the admission criterion depending on the network load nl with respect to a threshold, g. That is

bi= 

bi,max, nl< γ

bi,min, nl≥ γ , (1)

where bi is the admission criteria for connection i; bi,

max and bi, minare the maximum and minimum

band-width requirements corresponding to the highest and lowest QoS, respectively, for connection i. Furthermore, we can consider a linear adaptation approach which smoothly regulates the admission criteria bi in

propor-tion to the change of network load nl as shown in Fig-ure 3b. Accordingly, bican be expressed as

bi=αbi,max+ (1− α)bi,min, whereα =

⎧ ⎪ ⎪ ⎨ ⎪ ⎪ ⎩ 1 , nl< nlmin th nlmax th − nl nlmax th − nlminth , nlmin th ≤ nl < nl max th 0 , nl≥ nlmax th (2)

where nlmin_th and nlmax_th refers to the minimum and maximum threshold of network loads respectively. If nl is less than nlmin_th , the weighted factor a will be set as 1 and bi, maxwould be adopted as the admission criterion

to provide the best QoS. On the other hand, if nl is equal to or larger than nlmax_th , a will be set as 0 and bi,

min would be used to provide the lowest QoS. If nl is

Frame

DL-PHY PDU Contention-initial ranging Contention bandwidth request UL-PHY PDU from SS #1 UL-PHY PDU from SS #2 Preamble FCH DL burst #1 DLFP DL-MAP, UL-MAP, DCD, UCD MAC PDUs Preamble UL burst MAC PDUs MAC PDUs PAD

MAC Header MAC Payload CRC DL Subframe UL Subframe

DL burst #2

within the range of [nlmin_th , nlmax_th ), bi will be linearly

regulated in proportion to the change of nl to provide an adequate QoS. In general, nlmin_th and nlmax_th can be varied with the network bandwidth B and each type of

service. With a given B, nlmin_th and nlmax_th for a higher service class should be larger in order to provide differ-ential QoS among various service classes. A feasible approach to determine the values of nlmin_th and nlmax_th

a. Hard-decision function

b. Linear function

c. Quantized-step function

Figure 3 Several functions to adjust the admission criteria according to the network load. (a) Hard-decision function. (b) Linear function. (c) Quantized-step function.

with regard to each service class is to set different ratios of nlmin_th and nlmax_th to B appropriately. For example, in the simulations we consider the following setting of (nlmin_th /B, nlmax_th /B): (3/6, 5/6), (2/5, 4/5), (2/6, 4/6), (1/5, 3/5), and (1/6, 3/6), respectively, for UGS, rtPS, ertPS, nrtPS, and BE in order. Finally, as shown in Figure 3c,

this approach is to quantize the bandwidth within [bi,

min, bi, max] with N + 1 levels, each of which refers to a

feasible admission criteria. The adaptive criteria bican

be expressed as bi= ⎧ ⎨ ⎩ b0_i(bi,max), nl< nl0_th(nlmin_th ) bn i, nlnth−1≤ nl < nlnth, 1≤ n ≤ N − 1 bN_i (bi,min), nl≥ nlN_th−1(nlmax_th ) (3)

where nln_th refers to the nth threshold of network load (1 ≤ n ≤ N - 2). In particular, nl0_th and nlN_th−1 refer to the minimum and maximum thresholds of network loads, nlmin_th and nlmax_th , respectively. Here, bk_i (1≤ k ≤ N - 1) refers to the kth feasible admission criteria as the network load nl is within the range of [nlk_th−1, nlk_th). In particular, b0_i corresponds to bi, maxas nl <nl0th, and bNi

corresponds to bi, min as nl≥ nlN_th−1, respectively. The

value of N + 1 refers to the number of quantization levels. When N is 1 in particular, the quantized step function has two levels and is equal to the hard-decision function as shown in Figure 3a. In contrast, if N is infi-nite, the quantized step function has countless levels and is equal to the linear adaptation approach as shown in Figure 3b. Normally, the value of N can be deter-mined considering a reasonable number of adaptive cri-teria bi between bi, min and bi, max, (e.g., N = 5-15 in

general). When the value of N is determined, the unknown values of nln_th (1≤ n ≤ N - 2) and bk_i (1≤ k ≤

N - 1) can therefore be obtained by using a uniform

quantizer which equally partitions the region within (b0_i, bN_i ) and (nl0_th, nlN_th−1). In this article, we consider the linear adaptation approach as shown in Figure 3b for our CAC algorithm and use it for performance evalua-tions in the following section.

Denote the bandwidth allocated to existing new con-nections and handoff concon-nections as bn and bh,

respec-tively. A handoff connection will be accepted in the network as long as the bandwidth available can satisfy the bandwidth requirement. Thus, the condition to accept a handoff connection is

ho accepted = (bi,ho+ bn+ bh)≤ B, (4) where bi, hois the admission criterion of handoff

con-nection i determined in Equation (2). A new concon-nection

will be accepted when the following condition is satis-fied:

new accepted = ((bi,new+ bn)≤ thad)∩ ((bi,new+ bn+ bh)≤ thmax), (5)

where bi, newis the admission criterion of new

connec-tion i provided in Equaconnec-tion (2). The first term in the rhs of Equation (5) aims for increasing the admitting oppor-tunities for new connections when it only determines whether the resources allocated to new connections exceed the threshold. The last term in the rhs aims for ensuring a minimum BR for handoff (i.e. B - thmax). In

Section 3.3 we will have a more detailed illustration for the criteria considered in Equation (5).

3.2. Estimation of system capacity

The system capacity B may be dynamic as the IEEE 802.16e standards on PHY support multiple transmis-sion rates by using adaptive modulation and coding (AMC) schemes. The transmitter will determine one from various modulation and coding schemes (MCSs) available according to the channel conditions of packet delivery to provide reliable link qualities, large network coverage, and high data rates as possible. The modula-tion types supported in the IEEE 802.16e standards include Binary Phase-Shift Keying (BPSK), Quadrature Phase-Shift Keying (QPSK), 16-Quadrature Amplitude Modulation (16-QAM), and 64-QAM. With the Convo-lutional Turbo Code (CTC) and different code rates, the MCSs provided for WiMAX with 5 and 10 MHz chan-nels are summarized in Table 2[35]. In a DL transmis-sion for example, each MS informs its current perceived channel quality to the BS periodically, and then the BS will choose a specific MCS corresponding to this chan-nel condition. The transmission data rate with a given MCS can be evaluated as

RMCSi = (nData SC/TS)∗ bMCSi, (6)

where nData_SC is the number data sub-carriers; TSis

symbol period, and bMCSi is the amount of information

bits per symbol with respect to the ith MCS, MCSi. In

addition, adopting a Multiple-Input and Multiple-Out-put (MIMO) mechanism can further increase the trans-mission data rates to several-fold the original amount.

Here, our CAC scheme estimates the current system capacity B by consideration of the proportion of used MCSs [36,37]. Take an example for the capacity estima-tion as follows. Consider a 10-MHz channel spectrum with a 2 × 2 MIMO mechanism in the downlink trans-mission. Assume the proportion of used MCSs is QPSK 3/4 = 25%, 16-QAM 1/2 = 25%, and 64-QAM 5/6 = 50%. Thus, the estimated system bandwidth B in this case will be (9.5*2)*25% + (12.67*2)*25% + (31.68*2) *50% = 42.77 Mbps. In general, B in the downlink can

range from 2.12 to 63.36 Mbps widely with AMC in dif-ferent channel situations. Based on the number of sup-ported users with respect to each MCS and the proportion of adopted MCSs, our CAC scheme can esti-mate the current system capacity and network load and update the information periodically every frame period in the BS site.

3.3. Adaptive BR scheme

We propose an adaptive BR scheme which dynamically adjusts the amount of reserved bandwidth for handoffs according to the arrival distributions of both handoff and new connections. The objective of our scheme is to simultaneously increase the admission opportunities for new coming users and guarantee QoS for handoff users. The basic idea of our adaptive BR scheme is a rational inference that generally the occurrences of new incoming connections may be much more fre-quent than that of handoff connections [22-24]. Thus, the optimal BR should take into account the arrival behavior of not only handoff connections, but also new connections in order to avoid a waste of network resource as possible.

The scheme considers two thresholds for BRs, thmin

and thmax, which refer to the minimum and maximum

thresholds of reserved bandwidth, respectively. The

adaptive threshold of BR, thad, is initially set as (thmin+

thmax)/2, and will be dynamically adjusted within the

range [thmin, thmax] according to the arrival behavior of

new connections and handoff connections to control the reserved bandwidth for handoff connections. When the condition shown in Equation (4) is met and a handoff connection is accepted, the adaptive threshold thad will

be decreased with the amount of allocated resources for the handoff connection without going below thmin. That

is

thad= min (thad− bi,ho, thmin) (7)

Alternatively, when the condition shown in Equation (5) is met and a new connection is accepted, thadwill be

increased with the amount of allocated resources for the new connection without exceeding thmax. That is

thad= max (thad+ bi,new, thmax) (8)

To more clearly illustrate the characteristics of our proposed BR scheme, we conduct a simplified transmis-sion scenario to provide a preliminary performance comparison between our scheme and the fixed threshold (FT) and dynamic threshold (DT) schemes [10]. With the FT scheme, the threshold of reserved bandwidth, thfixed, is fixed; the new connection would be accepted

Table 2 The mobile WiMAX PHY data rates (Source: WiMAX Forum [35])

Parameter Downlink Uplink Downlink Uplink

System bandwidth 5 MHz 10 MHz FFT size 512 1024 Null sub-carriers 92 104 184 184 Pilot sub-carriers 60 136 120 280 Data sub-carriers 360 272 720 560 Symbol period 102.9μs Frame duration 5 ms OFDM symbols/frame 48

Data OFDM symbols 44

Mod Coding rate 5 MHz channel 10 MHz channel

Downlink rate (Mbps) Uplink rate (Mbps) Downlink rate (Mbps) Uplink rate (Mbps)

QPSK 1/2 CTC, 6x 0.53 0.38 1.06 0.78 1/2 CTC, 4x 0.79 0.57 1.58 1.18 1/2 CTC, 2x 1.58 1.14 3.17 2.35 1/2 CTC, 1x 3.17 2.28 6.34 4.70 3/4 CTC 4.75 3.43 9.50 7.06 16 QAM 1/2 CTC 6.34 4.57 12.67 9.41 3/4 CTC 9.50 6.85 19.01 14.11 64 QAM 1/2 CTC 9.50 6.85 19.01 14.11 2/3 CTC 12.67 9.14 25.34 18.82 3/4 CTC 14.26 10.28 28.51 21.17 5/6 CTC 15.84 11.42 31.68 23.52

only if (bi, new+ bn+ bh)≤ thfixed. With the DT scheme,

the threshold of reserved bandwidth, thdyn, will be

adjusted within the range [thmin, thmax] depending on

the arrival and departure of handoff connections as fol-lows: when a handoff connection is accepted, the thresh-old thdynwill be increased with the amount of resources

allocated to the handoff user; when an existing handoff connection terminates, thdynwill be decreased with the

amount of released resources. The new connection will be accepted only when (bi, new+ bn + bh)≤ thdyn. With

regard to handoff connections, all the three algorithms have the same admission strategy that a handoff connec-tion will be accepted as long as the amount of band-width available can meet its requirement, i.e. (bi, ho+ bn

+ bh)≤ B.

We exploit the following simplified scenario to com-pare the performances of the three BR schemes. Assume that the total amount of network resources B is 100 units. Consider that the network is empty in the begin-ning. Assume that 80 new connections, 5 handoff con-nections, and 5 new connections arrive sequentially. Assume that each of the arrival connections requests a unit of resources. For the FT scheme, thfixedis set as 80

units. In the DT scheme, thmin and thmaxare 80 and 90

units, respectively, and thdyn is set as 80 units initially.

In our scheme, thmin and thmax are 0 and 90 units,

respectively, and consequently the initial value of thadis

45 units.

Figure 4 shows the process of connection admissions with the three schemes, respectively. It is shown that with the FT scheme, the first 85 connections (80 new and 5 handoff connections) are accepted. But, the last five new connections are rejected due to a fixed reserva-tion manner that a certain amount of bandwidth can never be used by new connections. The DT scheme has the same performance as the FT scheme that the last five new connections are blocked. The reason is that the DT scheme adjusts the reserved bandwidth according to the arrival and departure behavior of handoff connec-tions only. Thus, when the occurrence of handoff con-nections is relatively infrequent as the scenario shown above, the DT scheme will be similar to the FT scheme and can cause a waste of network resources as well. With our scheme, the 90 connections totally can be granted into the network. Note that our scheme can still be effective in the situations when the number of new connections is nearly equal to or less than that of hand-off connections (this will be further examined in the fol-lowing section). It is shown that the proposed scheme outperforms the two schemes in terms of increasing the number of admitted connections.

Although the DT scheme and our proposed scheme both apply an adaptive manner for their BR strategies, there are two essential differences which lead to their

dissimilar performances. One difference is between their criteria for accepting new connections. For the DT scheme, the new connection will be accepted if the amount of its bandwidth requirement and the bandwidth allocated to existing connections would not exceed the threshold, i.e. (bi, new+ bn+ bh)≤ thdyn. In our scheme,

one criterion for accepting a new connection as shown in the first term of the rhs in Equation (5) only examines whether the amount of the requirement and the

band-width allocated to existing“new connections” does not

exceed the threshold, i.e. (bi, new+ bn)≤ thad. Thus, with

our scheme, the existing handoff connections will not les-sen the resource for an incoming new connection. That is, when the DT scheme and our proposed scheme are

with the same amount of reserved bandwidth (thdyn=

thad), our scheme can increase a resource space of bhfor

accepting more new connections. Meanwhile, our BR scheme can guarantee QoS for handoff connections because of another condition for accepting a new con-nection as shown in the last term of the rhs in Equation (5). That is, the total amount of required bandwidth for a new connection and allocated bandwidth for existing connections would not exceed the amount of guarded bandwidth for handoff, thmax, i.e. (bi, new + bn+ bh)≤

thmax. Thus, at last the B - thmaxbandwidth amount can

be reserved for handoff users.

Another difference is between their adjustment strate-gies for the threshold of BR, thdynand thad. With the DT

scheme, thdynis increased or decreased when a handoff

connection is established or terminated respectively, and will remain constant whether a new connection con-cludes or runs its course. When the occurrence of hand-off connections is relatively infrequent as the scenario shown above, the DT scheme will be similar to the FT scheme and can cause a waste of network resources too. In our scheme, thadis increased or decreased when a new

or handoff connection is admitted and runs its course, respectively, and will remain constant whether a new or handoff connection concludes. Note that a lower thresh-old of BR advances the acceptance rate of handoff users whereas a higher threshold increases the admission opportunities for new connections. In the sense, our scheme has potential to grant more new connections into networks especially when the occurrences of new con-nections are much more than that of handoff connec-tions. As aforementioned, our scheme would not sacrifice handoff QoS for favoring new connections when no less than the bandwidth amount of B - thmaxwill be reserved

for handoff users. Thus, the proposed adaptive BR scheme can simultaneously improve the network effi-ciency by granting more new connections and also guar-antee handoff QoS.

To summarize, Figure 5 presents the joint design of the proposed CAC algorithm and BR scheme, and

Figure 6 shows the corresponding pseudocodes. The block diagram in Figure 5 contains the following three steps.

(1) When a new connection or handoff connection arrives, it will inform the BS of its highest and lowest bandwidth requirements (i.e., bi, max and bi, min). The

proposed dynamic CAC scheme will adjust the admis-sion criterion using Equation (2) according to the cur-rently estimated system capacity and network load.

(2) When the admission criterion is determined, the proposed adaptive BR scheme will accept or reject this handoff or new connection depending on the criterion in Equation (4) or Equation (5), respectively.

(3) If a handoff connection is established, thadwill be

decreased with the amount of allocated resources as Equation (7) shows; if a new connection is granted, thad

will be increased with the amount of allocated resources as Equation (8) shows.

a. FT scheme

b. DT scheme

c. Proposed BR scheme

thad= thad+ bi,new

thad= thad- bi,ho

Adaptive Bandwidth Reservation (BR) Scheme Dynamic Connection Admission

Control (CAC) Algorithm

bi= α bi,max+ (1-α) bi,min Accepted or rejected? New or Handoff? Admission Criteria Network Loads (1) (1) (2)

Bandwidth Reservation Threshold

Accepted (3) (3) New Handoff QoS Requirements of Connection i (bi,max, bi,min)

Figure 5 The proposed dynamic CAC algorithm and adaptive BR scheme.

endforall endif endif then if else then if else else endif then if then if then if else elseif if do forall )) ( , min( ) ( ) ( // // )) ( max( ) ( ) ( ) ( ) ( ) 1 ( ) ( ) ( } , , , , { is ) of flow service and s connection pending ( epoch time At 2 / ) ( , max , max , , , , min max , , min , min , max , max , min max new i ad ad ad n new i h n new i ad n new i ho i ad ad h n ho i h n ho i i i i i i i i ad b th th th th b b new reject new accept th b b b and th b b connection new is it handover reject B in capacity more no b th th th th b b b handover accept B b b b i is class service and handover is type i class of b b BW less BS i class of b b b BW ordinary BS i class of b b BW more BS BE nrtPS rtPS ertPS UGS in c i c t th th th + ← <= + <= + + <= + − ← <= + + <= + + = − + = = + ← α α

The implementation of the proposed CAC and BR schemes in practice can involve the overheads as follows.

(1) The estimate of system capacity: The system capa-city can be evaluated at the BS with the specific PHY characteristics like channel spectrum, the amount of data sub-carriers, supported MCSs, used MIMO mechanisms, etc. The estimation of system capacity can be obtained in the initial phase of a network built-up.

(2) The estimate of network loads: In general, the net-work loads can be evaluated at the BS with the informa-tion of currently adopted MCSs and the number of supported users with respect to each MCS. This part may need the exchanges of some context information between BS and SSs periodically, e.g., currently channel condition and used modulation.

(3) The determination of admission criteria for incom-ing connections: When a connection arrives and requests for an admission, it will inform the BS of its specific QoS requirements, e.g., maximum and minimum data rates. Based on the estimated system capacity and net-work loads and QoS parameters, the BS will compute the admission criteria for the incoming connection with its specific QoS parameters.

(4) The adaptation of BR for handoff connections: If a connection is admitted in the network, the BS will therefore adapt the BR threshold depending on the type of connection, i.e., new or handoff.

4. Performance evaluations and results

In this section, we conduct simulations of 802.16e trans-mission scenarios to demonstrate the effectiveness of the proposed CAC algorithm and BR scheme. The simu-lator is constructed in C and followed the IEEE 802.16e standard closely [35,38]. The channel spectrum is 10 MHz. The MAC frame duration is 5 ms, which consists of 1024 OFDM subcarriers (840 data and pilot subcar-riers). One MAC frame includes 48 OFDM symbols, while the first symbol is used for a preamble. The ratio of the symbols of the uplink subframe to those of the downlink subframe is 18:29. In the uplink, three symbols are used for control signaling, and there are 44 OFDM symbols used for data transmissions in the uplink and downlink in total. The simulation set-up considers a 2 × 2 MIMO mechanism and the AMC schemes. We used the following distribution for MCS levels: QPSK 1/12 = 3.71%, QPSK 1/8 = 12.01%, QPSK 1/4 = 29.10%, QPSK 1/2 = 29.67%, QPSK 3/4 = 9.23%, 16-QAM 1/2 = 12.51%, 64-QAM 1/2 = 0.75%, and 64-QAM 3/4 = 3.02% [39,40]. The OFDMA PHY parameters and their values are listed in Table 2.

We provide different simulation scenarios to examine the proposed CAC algorithm and BR scheme individu-ally or jointly to clearly show the performance effects

with the two schemes. For each of the scenarios, we assume the connection arrivals and departures are with the Poisson process with a mean arrival rate l and a mean departure rate equal to one-tenth of the arrival rate. We assume that the BS is aware of the amount of connections with regard to each kind of MCS and the bandwidth requirement of each connection. The BS can therefore exploit this knowledge to estimate the current system capacity and also network loads. Generally, the system capacity and network loads can therefore be esti-mated for our scheme with the information of currently adopted MCSs and the number of supported users with respect to each MCS. The total simulation period is 1000 s while the results are provided with the average values over 20 times of simulations.

4.1. The connection blocking rates and achieved throughput with different CAC schemes

In this section, we examine the performance of our CAC algorithm individually by the comparisons with that of other CAC schemes. Here, we do not take into account the reserved bandwidth for handoff users. Thus, each of the incoming connections will be admitted into the network as long as the bandwidth available can satisfy the admission criteria determined by various CAC schemes. The simulation set-up for this scenario considers three types of traffic classes which are rtPS, ertPS, and nrtPS. To clearly examine the achieved QoS performances regarding the three types of service classes with different schemes, we separately evaluate the per-formance for each class with its specific QoS require-ment, i.e., maximum and minimum data rates as shown in Table 3.

We compare the performance of our CAC algorithm with that of the static maximum (Static-max), static minimum (Static-min), and bandwidth adaptation (Adapt) scheme considered in [11,12]. The Static-max and Static-min schemes adopt the highest and lowest QoS requirements, respectively, as the admission teria. The Adapt scheme considers the highest QoS cri-teria for a new coming user. If the resources available are insufficient to meet the bandwidth requirement of a new connection, the bandwidth allocated to existing

Table 3 The maximum and minimum rates associated with different scheduling services

Scheduling services Maximum rate (kbps) Minimum rate (kbps)

UGS 100 100

rtPS 100 25

ertPS 75 25

nrtPS 60 20

users will be decreased to satisfy the QoS requirement of the new connection. In our scheme, we consider a linear adaptation approach as shown in Figure 3b which adjusts the admission criteria according to the variation of network loads. We use the following ratios of (nlmin_th /B, nlmax_th /B): (2/5, 4/5), (2/6, 4/6), and (1/5, 3/5), for rtPS, ertPS, and nrtPS, respectively, as we consider in Section 3.1. Then, the minimum and maximum thresholds of network loads (nlmin_th , nlmax_th ) in Equation (2) for rtPS, ertPS, and nrtPS can be derived, respec-tively, depending on the currently estimated system capacity B. The performances are indexed as the con-nection blocking rate and the achieved QoS in terms of per-flow throughput, i.e., the average data rates sup-ported per established connection.

Figure 7 presents the average connection blocking rates of the rtPS, ertPS, and nrtPS flows with different CAC schemes. It is shown that the blocking rates rise with the increase of connection arrival rates l for all schemes except the Adapt scheme. The blocking rate with the Adapt scheme is almost 0 steadily, which is the lowest among all schemes when the incoming connec-tions will always be established regardless of the amount of bandwidth available. The blocking rate with the

Static-max scheme is the highest among all due to its strictest admission criterion. With the proposed scheme, the blocking rate is between that with the Static-max and Static-min schemes since the admission criteria are dynamically adjusted according to network loads (i.e., connection arrival rates).

Figures 8, 9, and 10 show the achieved per-flow throughput of the rtPS, ertPS, and nrtPS service classes, respectively, with different CAC schemes. It is shown in these figures that the Static-max and Static-min schemes provide steady throughput corresponding to the highest-fidelity and minimum QoS, respectively. It is shown from Figures 7, 8, 9, and 10 that the Static-max and Sta-tic-min schemes cannot properly balance the tradeoff between network efficiency and user-perceived QoS due to using fixed admission criteria without the considera-tion of traffic loads. Consider the case of ertPS traffic as shown in Figure 9 for example. With the Adapt scheme, the per-flow throughput keeps at the highest level when the network load is relatively light (l is less than 0.2); but, it degrades rapidly as the traffic loads get heavier (l is larger than 0.2) because the resources allocated to existing users are taken away to satisfy new connections’ QoS requirements. As the traffic load get much heavier such that l is larger than about 0.5, the achieved

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

Connection arrival rate, λ

Co

nn

ec

tion

bl

oc

ki

ng

r

at

e

Proposed

Adapt

Static-max

Static-min

throughput can no more keep above the value for the lowest QoS (25 kbps). With the proposed scheme, the achieved throughput is varying from 38 to 66 kbps around with the change of network loads, and remains at about 38 kbps when l is larger than 0.3. It is shown that the per-flow throughput can keep higher than the minimum QoS level. In Figures 8 and 10, we have simi-lar observations regarding the rtPS and nrtPS traffic classes. The results demonstrate that the proposed dynamic CAC algorithm can efficiently improve utiliza-tion of network resources and also assure adaptive QoS above the minimum requirements for admitted connections.

4.2. The new connection blocking rate and handoff dropping rate with different BR schemes

In this section, we conduct simulation scenarios which consider new connections and handoff connections to examine the performance of the proposed BR scheme without CAC. The compared BR schemes for perfor-mance evaluations are the FT and DT schemes [10] described in Section 3.2. The simulation set-up in this scenario considers five classes of scheduling services, UGS, rtPS, ertPS, nrtPS, and BE, each of which is

associated with specific maximum rates and minimum rates as shown in Table 3. To simply and fairly compare the performances of different BR schemes, our approach here considers the same fixed admission criteria as that used by the FT and DT schemes: the maximum rate is adopted as the admission criteria for UGS, rtPS, and ertPS, and the minimum rate is adopted for nrtPS and BE [10]. Assume that the occurrences of the five sche-duling services are with uniform probabilities, with which the arrival rates of new connections and handoff connections are lnand lh, respectively. We refer to [10]

and adopt the following threshold setting. For the FT

scheme, thfixed/B is equal to 80%. In the DT scheme,

thmin/B and thmax/B are set as 80 and 90%, respectively;

thdyn/B is set as 80% initially. In our scheme, thmin/B

and thmax/B are equal to 0 and 90%, respectively, and

consequently the initial value of thad/B is 45%. We

examine the performance of different BR schemes with different ratios of handoff arrival rates to new-connec-tion arrival rates which can be 1:1 or 1:20 in particular. The performance metrics are indexed as the new con-nection blocking rates and handoff dropping rates.

Figure 11a, b shows the new-connection blocking rate and handoff dropping rate, respectively, with different

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

20

30

40

50

60

70

80

90

100

110

rtPS

Connection arrival rate, λ

P

er-f

low

th

roug

hput

(

K

bps

)

Proposed

Adapt

Static-max

Static-min

BR schemes as the ratio of handoff arrival rates lhto

new-connection arrival rates lnis 1:1 (the total

connec-tion arrival rate l = lh + ln). It is shown that the FT

scheme provides the lowest handoff dropping probability and highest blocking rate when it adopts an FT for the BR. The DT scheme and our proposed scheme have similar performances in terms of both blocking and dropping rates as they determine the BR in an adaptive sense. Generally, the total amount of admitted connec-tions with the three schemes is nearly the same. By comparison, our proposed scheme can admit a few more connections into networks than the FT and DT schemes as its increasing number of new connections is slightly larger than the decreasing amount of handoff connections. For example, when the arrival rate l is equal to 0.8 (ln= lh= 0.4) in particular, the total

num-ber of admitted new and handoff connections with FT, DT, and the proposed scheme is 219, 218, and 225, respectively. In general with various values of arrival rates l, our proposed scheme averagely can slightly increase the network efficiency by 0.75 and 1.47% in comparison with the FT and DT schemes, respectively.

Figure 12a, b shows the new-connection blocking rate and handoff dropping rate, respectively, when the ratio

of handoff arrival rates lhto new-connection arrival

rates ln is 1:20. The performance of the FT scheme is

similar to that shown in Figure 11a, b that it has the lowest dropping and highest blocking rate among all. With regard to the DT scheme and our proposed scheme, it is shown that their performances are quite dissimilar here, unlike what is shown in Figure 11a, b. We can observe that our proposed scheme provides a much lower blocking probability and a slightly higher handoff dropping rate than the FT and DT schemes (notice that the scales of the vertical axes in Figure 12a, b are different). Regarding this transmission scenario in which the occurrences of new connections are much more frequent than those of handoff connections, the proposed scheme can significantly outperform the two schemes in terms of the amount of admitted connec-tions. For example, when the arrival rate l is equal to 0.8 (ln= lh= 0.4) in particular, the blocking rate with

FT, DT, and the proposed scheme is about 0.54, 0.53, and 0.47, respectively, while the dropping rate with the three schemes is about 0, 0.007, and 0.009, respectively. Consequently, the total number of admitted new and handoff connections with FT, DT, and the proposed scheme is 387, 399, and 442, respectively. The DT

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

10

20

30

40

50

60

70

80

ertPS

Connection arrival rate, λ

P

er-f

low

th

roug

hput

(

K

bps

)

Proposed

Adapt

Static-max

Static-min

scheme grants 12 connections more than the FT scheme, providing a slight increase of network efficiency by 3.20% over the FT scheme. It is shown that the DT scheme provides a rather marginal improvement over the FT scheme when the occurrences of handoff con-nections are relatively sparse. In this situation, the DT

scheme may cause a waste of network resources as much as the FT scheme does. With the proposed scheme, the number of granted handoff connections is almost equal to that of the FT and DT schemes since a bandwidth guard is used in our scheme to ensure a minimum BR for handoff users. On the other hand, the

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

10

20

30

40

50

60

70

nrtPS

Connection arrival rate, λ

P

er-f

low

th

roug

hput

(

K

bps

)

Proposed

Adapt

Static-max

Static-min

Figure 10 The achieved throughput of the nrtPS traffic class with different CAC schemes.

(a) (b) 0.050 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

New-connection arrival rate, λn

N ew -c on ne ct io n bl oc ki ng r at e FT DT Proposed 0.050 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

Handoff-connection arrival rate, λh

H and of f-co nne ct ion d ro ppi ng r at e _FT DT Proposed

Figure 11 Aslh:lnis 1:1, (a) the blocking rates of new connections and (b) the dropping rates of handoff connections with respect to different BR schemes. In comparison with the FT and DT schemes, respectively, the proposed scheme slightly increases the network efficiency by 0.75 and 1.47%.

number of admitted new connections with the proposed scheme is much more than that with the FT and DT schemes. Totally, our scheme can grant 55 and 43 con-nections more than the FT and DT schemes, providing a significant improvement of network efficiency by as large as 14.15 and 10.61% over the FT and DT schemes, respectively. In general when the arrival rates l (l = lh

+ ln) varies from 0.4 to 0.9, our proposed scheme

aver-agely can increase the network efficiency by 13.09 and 7.93% in comparison with the FT and DT schemes, respectively. The simulations demonstrate that our adaptive BR scheme can simultaneously grant more new connections to improve utilization efficiency of network resources and also guarantee BRs for handoff users.

4.3. The new connection blocking rate and hadoff dropping rate with different CAC and BR schemes

In this section, we conduct simulation scenarios to examine the joint performance of our dynamic CAC algorithm and adaptive BR scheme. The simulation set-up for this scenario is similar to that described in Sec-tion 4.2, except that our scheme further adopts the pro-posed CAC algorithm to dynamically adjust the admission criteria for the five scheduling classes. We use the following ratios of (nlmin_th /B, nlmax_th /B): (3/6, 5/6), (2/5, 4/5), (2/6, 4/6), (1/5, 3/5) and (1/6, 3/6), for UGS, rtPS, ertPS, nrtPS, and BE, respectively, as we consider in Section 3.1. Then, the minimum and maximum thresholds of network loads (nlmin_th , nlmax_th ) in Equation (2) for UGS, rtPS, ertPS, nrtPS, and BE can be derived, respectively, depending on the currently estimated sys-tem capacity B. The FT and DT schemes use the same admission strategy aforementioned that the scheduling classes of UGS, rtPS, and ertPS adopt the maximum

rate for the admission criteria while the nrtPS and BE classes use the minimum rate for that.

Figure 13a, b presents the new-connection blocking and handoff dropping rate, respectively, when the ratio of handoff arrival rates lhto new-connection arrival rates ln

is 1:1. It is shown that the performance of blocking rates using our approach is much better than that with the FT and DT scheme, while the handoff dropping rate with our scheme is not far from that with the other two schemes. In particular, we compare the performances of our approach shown in Figure 11a with that shown in Figure 13a here (both in which the ratio of handoff arrivals to new-connec-tion arrivals is 1:1), and it is observed that the later pro-vides a great improvement over the former (as the blocking rate is decreased by 46.10% in average). The per-formance improvement is contributed by the joint use of the proposed CAC algorithm which dynamically deter-mines the admission criteria depending on network loads to increase the connection admission opportunities. As a result, our approach (dynamic CAC algorithm and adap-tive BR scheme) averagely can increase the network effi-ciency by as large as 19.71 and 19.95% over the FT and DT schemes, respectively, with various values of arrival rates.

Finally, Figure 14a, b presents the blocking and drop-ping rate, respectively, when the ratio of handoff arrival rates lhto new-connection arrival rates lnis 1:20.

Simi-larly, we can clearly observe the performance improve-ment with the joint design of our CAC algorithm by comparing the results shown in Figures 12a and 14a particularly (both in which the ratio of handoff arrivals to new-connection arrivals is 1:20). It is shown in Figure 14a, b that our approach can significantly improve the network utilization by totally admitting 22.45 and 17.32% connections more than the FT and DT schemes, respectively. From the results shown in Figures 13 and

(a) (b) 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

New-connection arrival rate, λn

N ew -c onne ct ion b loc ki ng ra te FT_DT Proposed 0.004 0.008 0.012 0.016 0.020 0.024 0.028 0.032 0.036 0.04 0.044 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08

Handoff-connection arrival rate, λh

H an do ff-c on nec ti on d ro pp in g rat e _FT DT Proposed

Figure 12 Aslh:lnis 1:20, (a) the blocking rates of new connections and (b) the dropping rates of handoff connections with respect to different BR schemes. When the arrival ratesl (l = lh+ln) varies from 0.4 to 0.9, the proposed scheme averagely improve the network utilization by 13.09 and 7.93% in comparison with the FT and DT schemes, respectively.

14, we can observe that our approach generally performs better in the latter where the occurrences of new con-nections are much more frequent than that of handoff connections. The performance improvement is attribu-ted to our adaptive BR scheme which increases the net-work efficiency by dynamically adjusting the amount of reserved bandwidth for handoff users according to the arrival distributions of both handoff and new connec-tions. The simulation results demonstrate the effective-ness of our dynamic CAC algorithm and adaptive BR scheme to increase the network efficiency in terms of granting more connections and also guarantee adaptive QoS for admitted new and handoff connections.

5. Conclusion

The mechanisms of CAC and BR are important research issues in wireless networks due to scarce radio

resources, dynamic channel qualities, and diverse user demands. The article presents a dynamic CAC algorithm and an adaptive BR scheme for IEEE 802.16e mobile WiMAX networks to simultaneously improve the utili-zation efficiency of network resources and guarantee QoS for admitted new connections and handoff connec-tions. Our CAC algorithm dynamically determines the admission criteria according to network loads to provide adaptive QoS for admitted users and improve the utili-zation efficiency of networks. When new connections and handoff connections enter the network based on the adaptive admission criteria, the proposed BR scheme further dynamically adjusts the amount of reserved bandwidth for handoffs according to the arrival distribu-tions of new/handoff connecdistribu-tions to increase the con-nection admission opportunities and also guarantee the bandwidth requirements for handoff QoS. Simulations

(a) (b) 0.050 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

New-connection arrival rate, λn

N ew -c onn ec tion b lo cki ng ra te FTDT Proposed 0.050 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

Handoff-connection arrival rate, λh

H an doff -c onn ec tion d rop pi ng ra te FT DT Proposed

Figure 13 Aslh:lnis 1:1, (a) the blocking rates of new connections and (b) the dropping rates of handoff connections with respect to different CAC and BR schemes. The proposed dynamic CAC algorithm plus adaptive BR scheme averagely increase the network efficiency by 19.71 and 19.95% over the FT and DT schemes, respectively.

(a) (b) 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0 0.1 0.2 0.3 0.4 0.5 0.6

New-connection arrival rate, λn

N ew -c onne ct ion b loc ki ng ra te FT_DT Proposed 0.004 0.008 0.012 0.016 0.020 0.024 0.028 0.032 0.036 0.04 0.044 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08

Handoff-connection arrival rate, λh

H andof f-conne ct ion dr oppi ng r at e _FT DT Proposed

Figure 14 Aslh:lnis 1:20, (a) the blocking rates of new connections and (b) the dropping rates of handoff connections with respect to different CAC and BR schemes. The proposed dynamic CAC algorithm plus adaptive BR scheme improve the network utilization by admitting 22.45 and 17.32% connections more than the FT and DT schemes, respectively.

results demonstrate that our approach can simulta-neously improve the efficiency of resource utilization by granting more connections in the network and provide

adaptive QoS for admitted new and handoff

connections.

Acknowledgements

This work was supported in part by Taiwan National Science Council under Grant 99-2221-E-003-005 and 100-2221-E-003-020.

Competing interests

The authors declare that they have no competing interests. Received: 28 April 2011 Accepted: 14 April 2012 Published: 14 April 2012

References

1. IEEE 802.16 WG, IEEE Standard for Local and Metropolitan Area Networks Part 16: Air Interface for Fixed and Mobile Broadband Wireless Access Systems. Amendment 2, IEEE (December 2005)

2. S-C Lo, Y-Y Hong, A novel QoS scheduling approach for IEEE 802.16 BWA systems, in Proceedings of the 11th IEEE International Conference on Communication Technology (ICCT), Hangzhou, China, pp. 46_{–49 (Nov 2008)} 3. A Sayenko, O Alanen, J Karhula, T Hamalainen, Ensuring the QoS

requirements in 802.16 scheduling, in Proceedings of the 9th ACM international symposium on Modeling analysis and simulation of wireless and mobile systems (MSWIM), Torremolinos, Malaga, Spain, pp. 108_–117 (Oct 2006)

4. B Rong, Y Qian, K Lu, H-H Chen, M Guizani, Call admission control optimization in WiMAX networks. IEEE Trans Veh Technol. 57(4), 2509–2522 (2008)

5. B Rong, Y Qian, H-H Chen, Adaptive power allocation and call admission control in multiservice WiMAX access network. IEEE Wirel Commun. 14(1), 14_{–19 (2007)}

6. WS Jeon, DG Jeong, Call admission control for mobile multimedia communications with traffic asymmetry between uplink and downlink. IEEE Trans Veh Technol. 50(1), 59–66 (2001). doi:10.1109/25.917873

7. D Niyato, E Hossain, Call admission control for QoS provisioning in 4G wireless networks: issues and approaches. IEEE Netw. 19(5), 5–11 (2005). doi:10.1109/MNET.2005.1509946

8. B Rong, Y Qian, K Lu, Integrated downlink resource management for multiservice WiMAX networks. IEEE Trans Mob Comput. 6(6), 621–632 (2007) 9. X Chen, B Li, Y Fang, A dynamic multiple-threshold bandwidth reservation

(DMTBR) scheme for QoS provisioning in multimedia wireless networks. IEEE Trans Wirel Commun. 4(2), 583_{–592 (2005)}

10. SB Chaudhry, RK Guha, Adaptive connection admission control and packet scheduling for QoS provisioning in mobile WiMAX, in Proceedings of IEEE International Conference on Signal Processing and Communication (ICSPC), Dubai, United Arab Emirates, pp. 1355–1358 (Nov 2007)

11. D Niyato, E Hossain, A game-theoretic approach to bandwidth allocation and admission control for polling services in IEEE 802.16 broadband wireless networks, in Proceedings of the 3rd international conference on Quality of service in heterogeneous wired/wireless networks (QShine’06), Waterloo, Ontario, Canada, (August 2006)

12. D Niyato, E Hossain, Radio resource management games in wireless networks: an approach to bandwidth allocation and admission control for polling service in IEEE 802.16. IEEE Wirel Commun. 14(1), 72–83 (2007) 13. L Wang, F Liu, Y Ji, N Ruang-chaijatupon, Admission control for

non-preprovisioned service flow in wireless metropolitan area networks, in Proceedings of the Fourth European Conference on Universal Multiservice Networks (ECUMN’07), Toulouse, France, pp. 243–249 (February 2007) 14. K Yu, X Wang, S Sun, L Zhang, X Wu, A statistical connection admission

control mechanism for multiservice IEEE 802.16 network, in Proceedings of IEEE Vehicular Technology Conference (VTC 2009-Spring), Barcelona, Spain, pp. 1–5 (April 2009)

15. H Wang, W Lei, DP Agrawal, Dynamic admission control and QoS for 802.16 wireless MAN, in Proceedings of the Wireless Telecommunications Symposium (WTS2005), Cal Poly Pomona, CA, USA, pp. 60–66 (April 2005)

16. TC Tsai, CH Jiang, CY Wang, CAC and packet scheduling using token bucket for IEEE 802.16 networks. J Commun. 1(2), 30–37 (2006) 17. D Niyato, E Hossain, Queue-aware uplink bandwidth allocation and rate

control for polling service in IEEE 802.16 broadband wireless access networks. IEEE Trans Mob Comput. 5(6), 668–679 (2006)

18. DS Shu_{’aibu, SK Syed-Yusof, N Fisal, Partition-based bandwidth} managements for mobile WiMAX IEEE802.16e. Int Rev Comput Softw (IReCoS). 5(4), 445_{–452 (2010)}

19. E-C Park, H Kim, J-Y Kim, H-S Kim, Dynamic bandwidth request-allocation algorithm for real-time services in IEEE 802.16 broadband wireless access networks, in Proceedings of IEEE INFOCOM 08, Phoenix, Arizona, USA, pp. 1526–1534 (August 2008)

20. E-C Park, Efficient uplink bandwidth request with delay regulation for real-time service in mobile WiMAX networks. IEEE Trans Mob Comput. 8(9), 1235_{–1249 (2009)}

21. Y Zhang, D Liu, An adaptive algorithm for call admission control in wireless networks, in Proceedings of the IEEE Global Communication Conference (GLOBECOM), San Antonio, TX, USA, pp. 3628–3632 (Nov 2001) 22. J Choi, T Kwon, Y Choi, M Naghshineh, Call admission control for

multimedia services in mobile cellular networks: a Markov decision approach, in Proceedings of IEEE ISCC, Antibes, France, pp. 594–599 (July 2000)

23. R Ramjee, R Nagarajan, D Towsley, On optimal call admission control in cellular networks, in Proceedings of IEEE INFOCOM_{‘96, San Francisco, CA,} USA, pp. 43–50 (March 1996)

24. Y Fang, Y Zhang, Call admission control schemes and performance analysis in wireless mobile networks. IEEE Trans Veh Technol. 51(2), 371–382 (2002). doi:10.1109/25.994812

25. S Nazari, H Beigy, A new distributed uplink packet scheduling algorithm in WiMAX networks, in Proceedings of IEEE ICFCC, Wuhan, China, pp. 232–236 (May 2010)

26. Eunyoung Lee, Hyung-Kun Park, Packet scheduling scheme for multiple services in mobile WiMAX system, in Proceedings of 2nd International Conference on Computer and Network Technology (ICCNT), Bangkok, Thailand, pp. 60_{–63 (April 2010)}

27. B Al-Manthari, N Nasser, H Hassanein, Packet scheduling in 3.5 G high speed downlink packet access networks: breadth and depth. IEEE Netw Mag. 21(1), 41–46 (2007)

28. M Nicolaou, A Doufexi, S Armour, Y Sun, Scheduling techniques for improving call capacity for VoIP traffic in MIMO-OFDMA networks, in Proceedings of IEEE Vehicular Technology Conference (VTC 2009-Fall), Anchorage, Alaska, USA, pp. 1_{–5 (Sept 2009)}

29. T Bonald, A score-based opportunistic scheduler for fading radio channels, in Proceedings of the European Wireless Conference(EWC_{‘04), Barcelona,} Spain, pp. 283–292 (Sept 2004)

30. J Holtzman, G Barriac, Introducing Delay Sensitivity into the Proportional Fair Algorithm for CDMA Downlink Scheduling, in Proceedings of IEEE Seventh International Symposium on Spread Spectrum Techniques and Applications(ISSSTA), Parsippany, NJ, USA, pp. 652–656 (Sept 2002) 31. W-H Kuo, W Liao, Utility-based resource allocation in wireless networks. IEEE

Trans Wirel Commun. 6(10), 3600–3606 (2007)

32. FP Kelly, Charging and rate control for elastic traffic. Eur Trans Telecommun. 8(1), 33_{–37 (1997). doi:10.1002/ett.4460080106}

33. FP Kelly, AK Maulloo, DHK Tan, The rate control for communication networks: shadow prices, proportional fairness and stability. J Oper Res Soc. 409, 237–252 (1998)

34. G Song, Y Li, Utility-based resource allocation and scheduling in OFDM-based wireless broadband networks. IEEE Commun Mag. 43(12), 127–134 (2005)

35. Forum™ WiMAX, Mobile WiMAX - Part I: A Technical Overview and Performance Evaluation. http://www.wimaxforum.org/technology/ downloads/Mobile_WiMAX_Part1_Overview_and_Performance.pdf (2006) 36. MR Ashayeri, H Taheri, Mobile WiMAX capacity estimation in various

conditions, in Proceedings of IEEE Iranian Conference on Electrical Engineering (ICEE), Isfahan, Iran, pp. 483–488 (May 2010)

37. A Belghith, L Nuaymi, WiMAX capacity estimations and simulation results, in Proceedings of IEEE Vehicular Technology Conference (VTC 2008-Spring), Marina Bay, Singapore, pp. 1741–1745 (May 2008)

38. IEEE 802.16e 2005, IEEE Standard for local and metropolitan area networks. Part 16: Air Interface for Fixed and Mobile Broadband Wireless Access Systems. (February 2006)

39. IEEE C802.16e-04/16, Coverage/capacity simulations for OFDMA PHY in ITU-T channel model including MRC, SITU-TC, AAS results. (November 2004) 40. H Lee, H-D Kim, D-H Cho, Smart resource allocation algorithm considering

voice activity for VoIP services in mobile-WiMAX system. IEEE Trans Wirel Commun. 8(9), 4688–4697 (2009)

doi:10.1186/1687-1499-2012-143

Cite this article as: Wang et al.: Dynamic admission control and bandwidth reservation for IEEE 802.16e mobile WiMAX networks. EURASIP Journal on Wireless Communications and Networking 2012 2012:143.

Submit your manuscript to a

journal and benefi t from:

7 Immediate publication on acceptance 7 Open access: articles freely available online 7 High visibility within the fi eld

7 Retaining the copyright to your article