Admission control mechanism and performance analysis based on stochastic automata networks formalism

https://doi.org/10.1016/j.jpdc.2010.11.008Get rights and content

Abstract

IEEE 802.16 has been designed to support QoS (Quality of Service) in Wireless broadband Metropolitan Networks (WMAN), and specifically in the access networks. To achieve this, the IEEE 802.16.e amendment introduces the service differentiation by five service classes (UGS, ertPS, rtPS, nrtPS and BE). To maintain the QoS of active connections and to avoid any congestion in the network, an Admission Control (AC) is defined. AC aims to accept or reject a new connection according to the negotiated parameters and the availability of resources in the network. This mechanism is not standardized and let to the operators. We tackle this point, by proposing in this study a new Admission Control (AC) that improves the QoS of BE traffic by avoiding a strict bandwidth assignment of other traffics (rtPS and nrtPS) as is defined in major previous studies. The proposed mechanism is based on token bucket for rtPS, nrtPS and BE traffics in order to reduce lightly the number of accepted connections and to improve considerably the number of accepted BE connections. To evaluate the performance of the proposed scheme, we use PEPS (Performance Evaluation of Parallel Systems) which is a powerful tool based on models which can be described with Stochastic Automata Networks (SAN). PEPS can solve complex models with a large state space and with many synchronized events. Therefore, we show that this tool is suitable for wireless network, and specifically for the proposed scheme, toward numerical results we show as we expected that our proposed AC outperform the classical one by reducing lightly the performance of other service classes.

Research highlights

► Proposition of a new Admission Control (AC) for WiMAX networks based on token bucket. ► Performance evaluation are based on Stochastic Automata Networks (SAN). ► PEPS tool is used to analyse the proposed AC and it is shown by numerical results that the overall QoS of the network is improved.

Introduction

WLAN as IEEE 802.11 standard, is not adapted for wireless broadband networks because of the limited radio coverage (up to 250 m) and its random access MAC layer that cannot provide different QoS levels of each type of traffic, such as minimum throughput or bandwidth access. IEEE 802.16 is more adapted to provide QoS for multi-services traffics [6], [4], [5]. The aim of this standard is to increase the radio coverage (up to 70 Km) with a high capacity (up to 1 Gbps at a medium mobility and up to 100 Mbps at a high mobility that was defined in the last IEEE 802.16m standard) [14], [15]. The architecture consists of one base station (BS) that serves clients that are defined as subscriber stations (SSs). The role of the BS is to act as a central entity to exchange data between the SSs in a PMP (Point to Multipoint) mode. The uplink and downlink channels are multiplexed by TDD (Time Division Multiplexing) or FDD (Frequency Division Multiplexing) mode. The uplink is used by all connected SSs while the downlink is only used by the BS. To consider multi-services in this standard with different QoS levels, IEEE 802.16e amendment has been standardized through five service classes: UGS, ertPS, rtPS, nrtPS, and BE. Each service class has its own specificity. The specificities of each service class are as follows [3], [8], [11]:

  • (1)

    Unsolicited Grant Services (UGS) is a constant bit rate service class designed to support real-time traffic of fixed-size data packets that arrive at constant intervals. It is adapted for service class that have end-to-end constraints such as voice. For this type of traffic, the packets are issued at periodic intervals with a fixed data size grant. This mechanism reduce the overhead and reduce the delay transit in the SS.

  • (2)

    Extended Real-time Polling System (ertPS) is adapted for a sporadic traffic that need a real-time bandwidth allocation. This service class is designed for a service flows that generate variable-sized data packet on a periodic basis. It can be used for a Voice over IP (VoIP).

  • (3)

    Real-time Polling Services (rtPS) is dedicated for a sporadic real-time service flow that generate variable size data packets on a periodic interval. This service class is adapted for an MPEG (moving pictures experts group) video transmission. The BS provides periodic unicast (uplink) request opportunities, which meet the flow’s real-time needs and allow the SS to specify the size of the desired grant. It requires more request overhead than UGS, but support variable grant sizes that optimize the bandwidth utilization.

  • (4)

    Non-real-time Polling Services (nrtPS) is designed to support non-real-time service flows that require variable-size data packets for which a minimum data rate is required. It is adapted for FTP transmission. BS provides unicast polls on a regular basis, which guarantees that the service flow receives request opportunities even during network congestion.

  • (5)

    Best Effort (BE) is used to support data streams for which no minimum service guarantees are required and therefore may be handled on a best available basis. The SS may use contention request opportunities as well as unicast request opportunities when the BS sends any request. The BS does not have any unicast uplink request polling obligation for BE SSs. Therefore, a long period can run without transmitting any BE packets, typically when the network is in the congestion state.

The Table 1 describes the required parameters for each service class defined by the standard.

By providing different QoS levels for each service class, bandwidth allocation is under the control of the BS scheduler. The management is done on the uplink as the downlink is only managed by the BS. The uplink bandwidth management is done by the interrogation of the BS status queue. Four uplink channel allocation band is implemented, which are as follows [16]:

  • unsolicited bandwidth grants that consist of dedicated slots reserved for UGS traffic.

  • Piggyback bandwidth request.

  • polling or band allocation per interrogation system of SSs.

  • contention procedures.

Bandwidth assignment in the UL-MAP is done by the BS according to one of two following modes:
  • GPC (Grant per Connection): each connection is treated separately and bandwidth is allocated to each connection explicitly. SS then transmits in the order specified by the BS.

  • GPSS (Grant per SS): All connections from a single SS are treated as single unit and bandwidth is granted accordingly by the BS on a per SS basis. An additional scheduler in the SS (SS uplink scheduler) determines the service order among its connections in the granted slots.

To ensure that the applications data receive the appropriate QoS treatment, the standard provides scheduling services. These services represent the mechanisms of data processing supported by the schedulers for transmitting data over a connection. The standard defines two types of schedulers [10]:

  • the outbound scheduler that selects packets from the outbound queues to send (BS downlink scheduler or SS uplink scheduler);

  • BS uplink scheduler allocating bandwidth to the SSs in the uplink channel to allow them to send their packets. It is based on the QoS requirements of service flows and on bandwidth requests.

However, scheduler algorithms are not specified in the standard, only QoS parameters are defined. This part is let to the convenience of the vendors. Moreover, as WiMAX is connection oriented and in order to ensure QoS for active connections an Admission Control (AC) is implemented and it is not specified by the standard. The AC aims to accept or reject a new connection according to the negotiated QoS parameters (maximum sustained traffic rate, minimum reserved traffic rate, and maximum latency) without degrading the overall QoS. This mechanism is crucial and needs to be defined; this is the aim of this study.

The rest of this paper is organized as follows: after a presentation of related works in the literature in the next section, we present the proposed AC scheme in Section 3. In order to show the benefits of the proposed scheme, we evaluated the proposed AC by an analytical model using Stochastic Automata Networks (SAN). Therefore, we present briefly the SAN formalism in Section 4 and we give the associated SAN of our considered model in Section 5. We give some numerical results to show the benefit of our proposed scheme in Section 6. Finally, we summarize the main contributions of this study and we give indications on future developments of our future works in Section 7.

Section snippets

Related works

The AC [20] is a mechanism that decides whether a connection is or not accepted according to the QoS requirements. It is designed to provide the QoS needed by the new connection by keeping the QoS of all active connections. This mechanism is not defined in the standard and as an open issue, it is let to the operators to develop its own policy according to their business model.

Several AC schemes are proposed in the literature. In [11], the authors propose a new AC where they consider two broad

Admission control description

In this section, we describe the AC using a strict priority and the AC using our proposed model. The description is done using queueing networks.

We suppose that there is one finite buffer for each traffic class. BUGS (resp. Brt,Bnrt,BBe) represents the UGS queue (resp. rtPS, nrtPS, BE) where all new connections arrive.

We assume that the UGS traffic class arrival (resp. rtPS, nrtPS, BE) follows a Poisson process with rate λUGS (resp. λrt,λnrt,λBe). The duration of a connection request follows an

Stochastic Automata Networks

SAN have been introduced as an efficient method to represent complex systems with interacting components such as parallel systems or distributed systems [18]. This new method automatically provides an analytic derivation of Markov chain generator matrix using tensor algebra. The SAN seem to be more efficient than queueing networks or stochastic Petri nets to model systems with a large number of states and complex synchronizations. Queueing networks give a very compact representation of systems

CAC modelization by SAN

We present in this section how we modelize the AC scheme in WiMAX networks using SAN. We consider that we have different kinds of traffics service classes (UGS, rtPS, nrtPS, and BE). There is one finite buffer for each kind of traffic with finite capacity BUGS for real-time traffic UGS (resp. Brt,Bnrt,BBe). We assume that the arrival of traffic for each service class follows a Poisson process with rate λUGS (resp. λrtλnrt,λBe). The duration of a connection request follows an exponential

Numerical results

The performance measures are obtained using PEPS (Performance Evaluation of Parallel Systems). PEPS is a software which allows us to solve numerically very large Markov Chains. It uses an input interface based on SAN, a compact storage for the infinitesimal generator of the Markov Chain and tensor algebra to handle the basic vector matrix multiplication [17]. The compact formula of infinitesimal generator is given in Eq. (1). Three solution methods are implemented in PEPS (the power method,

Conclusion

With its large coverage, high throughput and the introduction of QoS, WiMAX standard is a promising technology considered by the operators, specially for access networks. The QoS is provided toward five services classes (UGS, ertPS, rtPS, nrtPS and BE) defined in the standard. Moreover, to keep a constant QoS level for all accepted connections, an Admission Control (AC) is necessary, but it is not defined by the standard and it is let to the convenience of the operators. Classical ACs proposed

Lynda Mokdad received the Ph.D. in computer science from the University of Versailles in 1997. She is full professor at university of Paris 12. Her main research interests are about performance evaluation techniques and applications in wired, mobile and wireless networks and in software technologies as Web services.

References (21)

  • B. Plateau et al.

    A methodology for solving complex Markov models of parallel systems

    Journal of Parallel and Distributed Computing

    (1991)
  • C.F. Ball et al.

    Performance analysis of temporary removal scheduling alied to mobile WiMAX scenarios in tight frequency reuse

    European Transactions on Telecommunications

    (2005)
  • C.F. Ball et al.

    Performance evaluation of IEEE 802.16 WiMAX with fixed and mobile subscribers in tight reuse

    European Transactions on Telecommunications

    (2006)
  • A. Belghith, L. Nuaymi, Design and implementation of a QoS-included WiMAX module for NS-2 simulator, in: ACM...
  • A. Boukerche

    Handbook of Algorithms and Protocols for Wireless and Mobile Networks

    (2005)
  • A. Boukerche

    Algorithms and Protocols for Wireless and Mobile Ad Hoc Networks

    (2008)
  • A. Boukerche et al.

    A secure distributed anonymous routing protocol for ad hoc wireless networks

    Computer Communications Journal

    (2004)
  • M. Castrucci, F. Delli Priscoli, C. Buccella, V. Puccio, I. Marchetti, Connection admission control in WiMAX networks,...
  • J. Chen, W. Jiao, Q. Guo, An integrated QoS control architecture for IEEE 802.16 broadband wireless access system, in:...
  • Mohamad El Masri, Slim Abdellatif, Guy Juanole, An uplink bandwidth management framework for IEEE 802.16 with QoS...
There are more references available in the full text version of this article.

Cited by (17)

  • Handover management framework for WiMAX Point-to-Multi-Point networks

    2016, Computers and Electrical Engineering
    Citation Excerpt :

    The CAC method proposed in [16] also uses the bandwidth borrowing scheme when enough bandwidth is not available to admit handover calls, but fails to address the fairness issue (see Section 3.2) while admitting the handover calls of various service classes. Similar to the above discussed methods, the CAC methods proposed in [12,17–20] did not address prioritized handling of the handover calls of various service classes and also their bandwidth borrowing schemes may lead to the starvation of the lower priority calls sometimes. In contrast, the CAC method proposed in this paper gives handover calls of all service classes a fair chance to enter the network according to their service class priorities.

  • Modeling Accordion Method Using SAN

    2018, IEEE International Conference on Communications
  • REFIACC scheme evaluation using analytical modeling

    2017, IEEE Wireless Communications and Networking Conference, WCNC
  • Admission control based on WRR in WiMAX networks

    2017, IEEE Wireless Communications and Networking Conference, WCNC
View all citing articles on Scopus

Lynda Mokdad received the Ph.D. in computer science from the University of Versailles in 1997. She is full professor at university of Paris 12. Her main research interests are about performance evaluation techniques and applications in wired, mobile and wireless networks and in software technologies as Web services.

Jalel Ben-Othman received his B.Sc. and M.Sc. degrees both in computer Science from the University of Pierre et Marie Curie, (Paris VI) France in 1992, and 1994 respectively. He received his Ph.D. degree from the University of Versailles, France, in 1998. He was an assistant professor at the University of Orsay (Paris 11) and University of Pierre et Marie Curie (Paris 6), in 1998 and 1999 respectively. He is now an associate professor at the University of Versailles since 2000.

Dr. Ben-Othman research interests are in the area of wireless ad hoc and sensor networks, Broadband Wireless Networks, multi-services bandwidth management in WLAN (IEEE 802.11), WMAN (IEEE 802.16), WWAN (LTE), security in wireless networks in general and wireless sensor and ad hoc networks in particular. His work appears in highly respected international journals and conferences, including IEEE ICC, Globecom, LCN, VTC, PIMRC, etc.

He has supervised and co-supervised several graduate students in these areas. His widely known for his work on wireless ad hoc and sensor network, in particular security. Since 2002, he has served as technical committee of more than 40 international IEEE/ACM conferences and workshops including ICC, Globecom, MSWIM, LCN, etc. He is a member of IEEE and ACM.

He is also serving as the local arrangement chair for the 13th IEEE International Symposium on Computer Communication (ISCC 09). He serves as a TPC Co-Chair of 9th international Workshop on Wireless local Networks (WLN09), as a publicity chair of 12th ACM International Conference on Modeling, Analysis and Simulation of Wireless and Mobile Systems (MSWIM 09), IEEE International Symposium on a World of Wireless Mobile and Multimedia Networks (WOWMOM 2010), 25th Biennial Symposium on Communications and he serves as a TPC co-chair of IEEE Globecom Wireless Communications Symposium (Globecom 2010).

And he is active member of IEEE CIS-TC, TC AHSN, WTC.

This paper was supported by grant PUMA from DGCIS.

View full text