Implementation of call admission control scheme in next generation mobile communication networks using particle swarm optimization and fuzzy logic systems

(1)

Implementation of call admission control scheme in next

generation mobile communication networks using particle

swarm optimization and fuzzy logic systems

Chenn-Jung Huang

a,*

, Yi-Ta Chuang

b

, Dian-Xiu Yang

a

Department of Computer and Information Science, College of Science, National Hualien University of Education, 123 Huahsi Road, Hualien, Taiwan 97043, Taiwan

b

Institute of Computer Science and Engineering, College of Computer Science, National Chiao Tung University, 1001 Ta Hsueh Road, Hsinchu, Taiwan 300, Taiwan

Abstract

In the present and next generation wireless networks, cellular system remains the major method of telecommunication infrastructure. Since the characteristic of the resource constraint, call admission control is required to address the limited resource problem in wireless network. The call dropping probability and call blocking probability are the major performance metrics for quality of service (QoS) in wireless network. Many call admission control mechanisms have been proposed in the literature to decrease connection dropping prob-ability for handoffs and new call blocking probprob-ability in cellular communications. In this paper, we proposed an adaptive call admission control and bandwidth reservation scheme using fuzzy logic control concept to reduce the forced termination probability of multimedia handoffs. Meanwhile, we adopt particle swarm optimization (PSO) technique to adjust the parameters of the membership functions in the proposed fuzzy logic systems. The simulation results show that the proposed scheme can achieve satisfactory performance when per-formance metrics are measured in terms of the forced termination probability for the handoffs, the call blocking probability for the new connections and bandwidth utilization.

Keywords: Call admission control (CAC); Bandwidth reservation; Wireless networks; Multimedia; Quality of service; Fuzzy logic; Particle swarm optimization

1. Introduction

With the increasing demand for the provision of multimedia applications, such as video on demand (VoD), videoconference, and many WWW-based applications, a great deal of attention is being paid to resource allocation for providing seamless multimedia access in the next gener-ation mobile communicgener-ation networks (Zahariadis, 2003).

There are two important quality of service (QoS) param-eters considered in wireless networks, namely the handoﬀ call dropping probability (CDP) and new call blocking

prob-ability (CBP). The CDP denotes the likelihood that an ongo-ing call is forced to terminate durongo-ing a handoff process when the allocated resources in the new cell are degenerated to an unacceptable level, while the CBP represents the possibility that a new connection request is denied admission into the cellular networks. Accordingly, one of the most important QoS issues in providing multimedia traffic in wireless net-works is to reduce handoff drops caused by lack of available bandwidth in the new cell while maintaining high bandwidth utilization and low new call blocking rate.

Call admission control (CAC), is a mechanism used to administer the quality of service in the wireless network sys-tem (Rao, Comaniciu, Lakshman, & Poor, 2004). Since the resource in a wireless network system is limited, the resource management, especially in multimedia application,

*

Corresponding author. Tel.: +886 38238091.

E-mail address:[email protected](C.-J. Huang).

www.elsevier.com/locate/eswa Expert Systems with Applications 35 (2008) 1246–1251

Expert Systems with Applications

(2)

has been a critical problem for years. The CAC mechanism is thus important to manage the constraint resource in wire-less communication system. The other method to guarantee the above QoS metrics is resource reservation mechanism and a variety of resource reservation algorithms have been proposed for traditional cellular networks in recent years (Ei-Kadi, Olariu, & Abdel-Wahab, 2002; Oliveira, Kim, & Suda, 1998; Wu, Yeung, & Hu, 2000).

A well-known machine learning technique, fuzzy logic, is employed in this work to compute the expected amount of bandwidth used in the neighboring cells for the handoff calls so that the CDP handoff calls can be effectively reduced. Meanwhile, the fuzzy logic system is also embed-ded in the call admission controller to block the new calls if a possible congestion situation occurs at the cell that the MH resides. Another machine learning technique, Particle swarm optimization (PSO) technique, is used to adjust the parameters of the membership functions employed in the fuzzy logic systems in order to deal with the volatile char-acteristics of the cellular systems. The motivation of using fuzzy logic and particle swarm optimization in this work is that the computation of the two techniques is fast and sim-ple, and they have been successfully applied in many areas (Hirota, 1993; Ren & Ramamurthy, 2000).

The remainder of the paper is organized as follows. The primitive call admission control scheme is introduced in Section2. The adaptive call admission control scheme with fuzzy network traﬃc load estimator is given in Section 3. Then Section 4 states the bandwidth reservation scheme using fuzzy logic control concept. Section5 is the simula-tion results. Conclusions are given in Secsimula-tion6.

2. Primitive call admission control mechanism

In the wireless mobile communication system, the incom-ing client is admitted into the system by the grantincom-ing of the call admission controller. Notably, the available channels are divided into two categories, free channels and reserved channel. The free channels are allocated to the MHs through the resource controller, whereas the size of reserved channels is adjusted by the bandwidth reservation control-ler. Besides, the traffic in cellular networks is usually catego-rized into the following two classes in the literature (Oliveira et al., 1998). Class I traffic denotes real-time multimedia traffic, such as interactive audio and video, while Class II is non-real-time data traffic, such as images and text. Since the Class II traffic can torrent unstable bandwidth, portion of the bandwidth allocated to Class II traffic is ‘‘borrowed’’ by new call arrivals and handoff calls that belong to Class I traffic if necessary. When a new call arrival requests a new connection, the CAC algorithm firstly checks whether the desired bandwidth, bdesire, requested the MH is satisfied

by the available free channels, Bf, plus borrowable channels,

Cb. If the new arriving call is accepted and it belongs to

Class I traﬃc, the embedded resource reservation algorithm will also compute the amount of reserved bandwidth at the appropriate neighboring cells.

Since it is inappropriate to block a handoﬀ call instead of a new call, we further limit the number of new call arriv-als to access the system resource by using an adaptive call admission control mechanism in order to reduce the hand-oﬀ call dropping probability. Whether a new arriving call should be admitted in our algorithm is determined by the following three factors:

• Whether the handoﬀ call dropping probability, Phf, is

too high.

• When a cell is hotter, there will be more chances that the MHs at the neighboring cells will handoﬀ to this cell. The hotness of a cell is then deﬁned by:

hotness¼DnumMH

Dt ; ð1Þ

where numMHis the number of the incoming MHs and t

denotes time.

• The amount of the free bandwidth available in current cell.

Based on the above consideration, the network traﬃc load of a cell is estimated by:

load¼ C Phf hotness

1 Bf

; ð2Þ

where C is a constant, Phf denotes the handoﬀ call

drop-ping probability of the cell, hotness represents the hotness of the cell, and Bf stands for the unused reserved

band-width in the current cell. The parameters used in Eq. (2)

can be obtained from the base station through simple com-putation. The computation result of Eq. (2) is then com-pared with a predeﬁned threshold to determine whether the traﬃc is overloaded.

While there is a Class I handoﬀ arriving, the CAC algorithm checks if the required bandwidth, bdesire, is

smaller or equal to the accumulation of the free channels, Bf, borrowable channels, Cb, and channels reserved by

the neighboring cells, Br. It the handoﬀ is admitted, the

resource reservation algorithm will be activated to compute the estimated bandwidth at the appropriate neighboring cells when another handoff for the MH occurs in the near future. If the handoff call belongs to Class II traffic, the CAC algorithm allocates the available bandwidth to the incoming handoff if there is any free channel available.

3. Adaptive call admission control mechanism using PSO-tuned fuzzy logic system

3.1. Fuzzy logic call admission controller

The fuzzy logic techniques have been used to solve sev-eral connection admission control and channel assignment problems eﬃciently in ATM and wireless networks in the

(3)

literature (Hirota, 1993). We thus replace the naive compu-tation of Eq. (2) by fuzzy logic systems to determine whether the traﬃc at the current cell is overloaded. Mean-while, we adopt particle swarm optimization technique to adjust the parameters of the membership functions in the proposed fuzzy logic systems.

The fuzzy network traﬃc load estimator proposed in our call admission control scheme is composed of the fol-lowing components as described:

• Fuzzifier: The fuzzifier performs the fuzzification func-tion that converts three types of input data from the net-work traffic load estimated scheme into suitable linguistic values which are needed in the inference engine.

• Fuzzy rule base: The fuzzy rule base is composed of a set of linguistic control rules and the attendant control goals.

• Inference engine: The inference engine simulates human decision-making based on the fuzzy control rules and the related input linguistic parameters.

• Defuzzifier: The defuzzifier acquires the aggregated lin-guistic values from the inferred fuzzy control action and generates a non-fuzzy control output, which repre-sents the estimated network traffic load of the cell. Notably, the input to the fuzzifier Phf represents the

handoﬀ call dropping probability of the cell. The input hot-ness denotes the hothot-ness of the cell, and Bfthe

normaliza-tion of the unused bandwidth in the cell.

The membership functions for the handoﬀ call dropping probability of the cell, Phf, where the linguistic variables

‘‘low’’, ‘‘intermediate’’ and ‘‘high’’ give the dropping degree measure of the handoﬀ call dropping probability of the cell. The membership function for the hotness of the cell, hotness, is mapped into three linguistic term sets, ‘‘cold’’, ‘‘medium’’ and ‘‘hot’’ and the membership func-tion for the normalizafunc-tion of the unused bandwidth of the cell, Bf, is mapped into three linguistic term sets,

‘‘low’’, ‘‘medium’’ and ‘‘high’’ represent the measure of the unused bandwidth. The output parameter of the infer-ence engine, Load, is defined as the estimated network traf-fic load of our network traffic load estimator. The fuzzy linguistic variables for the output are ‘‘low’’, ‘‘medium’’ and ‘‘heavy’’, which are represented the measure of the net-work traffic load.

Fig. 1 illustrates the reasoning procedure. The rule in

Fig. 1is:

IF the handoﬀ call dropping probability of the cell is intermediate, AND the level of the unused bandwidth of the cell is ‘‘low’’, AND the hotness of the cell is ‘‘low’’.

THEN the estimated network traﬃc load of the cell is ‘‘medium’’. Note that the inference rule could be estab-lished by network operators.

The non-fuzzy output of the defuzziﬁer can then be expressed as the weighted average of each rule’s output after the Mamdani defuzziﬁcation method is applied

loadBW¼ R llAðlÞldl R llAðlÞdl ; ð3Þ

where lA(l) denotes the aggregated output constructed by

the connective ‘‘AND’’ as shown in the above example is satisﬁed.

3.2. Particle swarm optimization

Particle swarm optimization (PSO) is a computational intelligence approach to optimization that is based in the behavior of swarming or flocking animals, such as birds or fish. In the PSO, every individual moves from a given point to a new one which is a weighted combination of the individual’s best position ever found, and of the group’s best position. The PSO algorithm itself is simple and involves adjusting a few parameters. With little modifica-tion, it can be applied to a wide range of applications. Because of this, PSO has received growing interest from researchers in various fields.

In this work, we allow each vehicle to execute its individ-ual PSO algorithm. The PSO is used to adjust the two parameters, mean and variance, in the membership func-tions of the inputs for the fuzzy logic systems.

A summary of the PSO algorithm is given below: Initialize the swarm of the particles such that the posi-tion ~xijðt ¼ 0Þ of each particle is random within the

hyper-space, where j = 1–6, denote the values of the two parameters, mean and variance in diﬀerent membership functions, respectively.

Compare the ﬁtness function of each particle, Fð~xijðtÞÞ,

which is the packet delivery ratio of each individual during current time period, to its best performance thus far, pbestij;

then

ðiÞ if Fð~xijðtÞÞ < pbestijpbestij ¼ F ð~xijðtÞÞ; ð4Þ

ðiiÞ ~xpbestij ¼ ~xijðtÞ: ð5Þ

Phf hot BW Load

μ intermediate μ cold μ high μ medium

Ruler Min

wr

(4)

Compare Fð~xijðtÞÞ to the global best particle, gbestj if

Fð~xijðtÞÞ < gbestj then

ðiÞ gbestj¼ F ð~xijðtÞÞ; ð6Þ

ðiiÞ ~xgbestj¼ ~xijðtÞ: ð7Þ

Revise the velocity for each particle:

~vijðtÞ ¼ ~vijðt 1Þ þ c1 r1 ð~xpbestijðtÞ ~xijðtÞÞ þ c2 r2

ð~xgbestjðtÞ ~xijðtÞÞ; ð8Þ

where r1and r2are random numbers between 0 and 1, and

c1and c2are positive acceleration constants, which should

satisfy c1+ c264 as reported inKennedy (1998).

Move each particle to a new position:

ðiÞ ~xijðtÞ ¼ ~xijðt 1Þ þ~vijðtÞ; ð9Þ

ðiiÞ t¼ t þ 1: ð10Þ

Repeat steps(2) through(6) until convergence.

4. Bandwidth reservation mechanism using PSO-tuned Fuzzy logic system

The representative bandwidth reservation schemes in the literature (Ei-Kadi et al., 2002; Oliveira et al., 1998; Wu et al., 2000) anticipate that a Class I connection request will make a handoff into one of its neighboring cells in the future and thus try to reserve some bandwidth in surround-ing cells before the connection request is admitted. Although the above-mentioned bandwidth reservation schemes can effectively reduce the CDP in traditional mac-rocell wireless networks, however, it will introduce too many overheads among the BSs and waste much band-width in microcell cellular wireless systems due to exces-sively frequent reservation process. In this section, we adopt fuzzy logic controller concept to estimate the reserv-ing bandwidth among the neighborreserv-ing cells in an effective manner, effectively decreasing the CDP for the multimedia application handoffs, while keeping bandwidth utilization at a reasonable level.

The amount of the reserved bandwidth is determined by the following three factors:

• The speed and the direction that the MH will move to a neighboring cell. The probability of moving to a neigh-boring cell will be higher if the MH move toward it in a fast speed.

• The current reserved bandwidth for the neighboring cells. The probability of moving to a neighboring cell is proportional to the bandwidth that the neighboring cell reserves.

• The distance of the MH’s position to the neighboring cells. There are more chances that the MH will move to the neighboring cell that the MH is closer to. Based upon the above considerations, the bandwidth reserved in cell B for the MH located at cell A as illustrated inFig. 2, can be derived as follows:

BRB¼ C SB BWMH

1 DB

; ð11Þ

where C is a constant, BWMHdenotes the minimum

band-width requested by a MH at cell A, SBrepresents the speed

that the MH moves to cell B, and DB stands for the

dis-tance of the MH’s position to cell B. Notably, The param-eters BWMH and SB can be obtained easily by a simple

computation in the base station, and DB can be acquired

based on the location management implemented in the wireless system (Hoare, 1962, Saha, Mukherjee, Misra, Chakraborty, & Subhash, 2004).

The architecture of the proposed fuzzy bandwidth-reserving estimator is quite similar to the fuzzy network traﬃc load estimator presented in Section 3. Notably, the input to the fuzziﬁer d represents the normalization of the distance of the MH’s position to the neighboring cells. The input Pm denotes the normalization of the current

reserved bandwidth for the six neighboring cells, and S the normalization of speed that the MH moves toward the neighboring cell. The output of defuzziﬁer RBWstands

for reserved weighted factor:

RBW¼ P27 i¼1BWi wi P27 i¼1wi ; ð12Þ

The estimated bandwidth reserved in the target cell, such as cell B as shown inFig. 2, can be derived as follows: BRB¼ RBW BWMH; ð13Þ

where RBWstands for the non-fuzzy output, and BWMHis

the minimum bandwidth requested by the MH.

5. Simulation results

To evaluate the proposed call admission control mecha-nism, a series of simulations are conducted to compare the proposed fuzzy call admission mechanism plus fuzzy

MH MH

(5)

bandwidth reservation mechanism scheme (ACAC-FLR), with the primitive call admission control scheme as given by Eq. (2), plus ﬁxed reservation scheme (PCAC-FR), and the scheme without bandwidth reservation (PCAC-NR). Meanwhile, the rate-based borrowing scheme incor-porated with the primitive call admission control scheme (PCAC-RBB) (Ei-Kadi et al., 2002) is also compared with the proposed work because it was reported in Ren and Ramamurthy (2000) that the RBB scheme achieves better performance than other representative bandwidth alloca-tion and reservaalloca-tion schemes in the literature, such as the scheme presented inOliveira et al. (1998).

There are 36 cells included in the simulation environ-ment as shown in Fig. 3. A total of 30 Mbps bandwidth is allocated in each cell. Both classes of the connections are listed inTable 1. Each MH is given a speed character-istic, which decides the time spent in a cell, in order to sim-ulate handoﬀs. The MH moves randomly to a target cell with a random speed then stays in that cell with a speciﬁc time according to its speed and chooses next target ran-domly with a random speed.

Fig. 4shows the comparison of call dropping probabil-ity (CDP) for multimedia handoﬀs (Class I), and Fig. 5

illustrates the CDP for combined Classes I and II traffic. We can see fromFig. 4that the CDP for multimedia hand-offs is the lowest for the ACAC-FLR. Besides, the CDP for combined Classes I and II traffic in ACAC-FLR schemes is still lower than the other three schemes as shown inFig. 5. The PCAC-NR has the worst performance as expected

since it does not reserve bandwidth for the handoffs at all. As for the PCAC-RBB scheme, although it uses band-width borrowing technique to lower down the CDP for handoffs, its fixed bandwidth reservation mechanism is still inferior to the approach of adaptive bandwidth reservation based on the dynamic change of mobile node conditions as taken in this work.

Fig. 6illustrates the CBP for combined Classes I and II traffic. The call blocking probability for the new connec-tions in the ACACp-FLR and the PCAC-RBB schemes is apparently improved by means of the channel borrowing technique. Meanwhile, the effectiveness of fuzzy admission control and bandwidth reservation contributes to the better performance achieved in the proposed ACAC-FLR scheme as illustrated in Fig. 6. The PCAC-FLR scheme has the highest CBP for new connections because it reserves fixed bandwidth for multimedia handoff connections and results in lessened bandwidth available for new connections.

The bandwidth utilization of various mechanisms is given in Fig. 7. Note that the bandwidth utilization is deﬁned as:

P

for each cellUsed bandwidth of each cell

P

for each cellMaximum bandwidth of each cell

: ð14Þ 35 30 31 32 33 34 35 29 24 25 26 27 28 29 23 18 19 20 21 22 23 17 12 13 14 15 16 17 11 6 7 8 9 10 11 5 0 1 2 3 4 5 35 30 31 32 33 34 35

Fig. 3. Cellular topology with 36 cells in the simulations.

Table 1

Multimedia Traﬃc Used in the simulations Traﬃc class Bandwidth requirement Average call duration Example

Class I 30 Kbps 3 min Voice service

Class I 256 Kbps 5 min Video-phone

Class I 1–4 Mbps 10 min Video service Class II 5–20 Kbps 30 s E-mail, paging Class II 64–512 Kbps 3 min Remote login and data

on demand Class II 1–5 Mbps 2 min Ftp 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.01 0.02 0.05 0.1 0 .2 0.5 1

Call arrival rate (call/sec)

CDP for class I traffic

PCAC-NR PCAC-FR PCAC-RBB ACAC-FLR

Fig. 4. Call dropping probability for Class I traﬃc in the four schemes.

0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.01 0.02 0.05 0.1 0.2 0.5 1

Overal CDP

Fig. 5. Call dropping probability for combined Classes I and II handoﬀs in the four schemes.

(6)

The ACAC-FLR scheme also achieves better performance than the other three schemes in bandwidth utilization due to the efficient usage of adaptive admission control and bandwidth reservation mechanisms. The PCAC-RBB scheme uss bandwidth borrowing technique to achieve higher bandwidth utilization than the PCAC-NR and PCAC-FR schemes. Bandwidth utilization is the poorest in the PCAC-FR scheme since it always reserves fixed bandwidth in each cell which is not necessarily used by the handoffs.

6. Conclusion and future work

In this paper, the adaptive call admission control band-width reservation schemes using fuzzy logic controller con-cept is proposed to reduce forced termination of multimedia handoﬀs in the mobile wireless communication systems. Fuzzy logic controller concept is employed to determine whether a new arriving call should be admitted

and to compute the amount of reserved bandwidth for the handoffs in the expected target cells. Meanwhile, parti-cle swarm optimization technique is adopted to dynami-cally tune the parameters in the membership function of the fuzzy systems due to the volatile characteristics of mobile wireless networks. The simulation results show that the proposed scheme, the ACAC-FLR, performs better then the fixed reservation (PCAC-FR) scheme, the scheme without reservation (PCAC-NR), and the rate-based bor-rowing scheme (PCAC-RBB) when call blocking probabil-ity for new connections, call dropping probabilprobabil-ity for the handoffs, and bandwidth utilization are compared. Subse-quent research will investigate the feasibility of applying other intelligent tools such as neuro-fuzzy and genetic algo-rithms into the proposed scheme to improve the accuracy of the motion prediction for the MH.

Acknowledgements

The authors would like to thank the National Science Council of the Republic of China, Taiwan for ﬁnancially supporting this research under Contract No. NSC 95-2221-E-026-001.

References

Ei-Kadi, M., Olariu, S., & Abdel-Wahab, H. (2002). Rate-based borrow-ing scheme for QoS provisionborrow-ing in multimedia wireless networks. IEEE Transactions of Parallel and Distributed Systems, 13(2), 156–166. Hirota, K. (1993). Industrial applications of fuzzy technology.

Springer-Verlag.

Hoare, C. A. R. (1962). Quicksort. Computer Journal, 5(4), 10–15. Kennedy J. (1998). The behavior of particles. In 7th international

conference on evolutionary programming, pp. 581–589.

Oliveira, C., Kim, J. B., & Suda, T. (1998). An adaptive bandwidth reservation scheme for high-speed multimedia wireless networks. IEEE Journal of Selected Areas in Communication, 16(6), 858–874. Rao, R. M., Comaniciu, C., Lakshman, T. V., & Poor, H. V. (2004). Call

admission control in wireless multimedia networks. IEEE Signal Processing Magazine, 21(5), 51–58.

Ren, Q., & Ramamurthy, G. (2000). A real-time dynamic connection admission controller based on traﬃc modeling, measurement, and fuzzy logic control. IEEE Journal of Selected Areas in Communication, 18(2), 184–196.

Saha, D., Mukherjee, A., Misra, I. S., Chakraborty, M., & Subhash, N. (2004). Mobility support in IP: a survey of related protocols. IEEE Network, 18(6), 34–40.

Wu, X., Yeung, K. L., & Hu, J. (2000). Eﬃcient channel borrowing strategy for real-time services in multimedia wireless networks. IEEE Transactions of Vehicular Technology, 49(4), 1273–1284.

Zahariadis, T. (2003). Trends in the path to 4G. Communications Engineer, 1(1), 12–15. 0 0.1 0.2 0.3 0.4 0.5 0.6 0.01 0.02 0.05 0.1 0.2 0.5 1

Overall CBP

Fig. 6. Call blocking probability for combined Classes I and II traﬃcs in the four schemes.

0 0.2 0.4 0.6 0.8 1 1.2 0.01 0.02 0.05 0.1 0.2 0.5 1

Bandwidth utilization