J. Cent. South Univ. (2012) 19: 393-401

DOI: 10.1007/s11771-012-1017-2

Reservation-based dynamic admission control scheme for wideband code division multiple access systems

A. Y. Al-nahari, S. A. El-Dolil, M. I. Dessouky, F. E. Abd El-Samie

Department of Electronics and Electrical Communications, Faculty of Electronic Engineering, Menoufia University, Menouf 32952, Egypt

? Central South University Press and Springer-Verlag Berlin Heidelberg 2012

Abstract: Call admission control (CAC) and resource reservation (RR) for mobile communication are two important factors that guarantee system efficiency and quality of service (QoS) required for different services in a very scarce resource as the radio spectrum. A new scheme was proposed which extends the concepts of resource sharing and reservations for wideband code division multiple access (WCDMA) systems with a unique feature of soft capacity. Voice and data traffic were considered. The traffic is further classified into handoff and new requests. The reservation thresholds were dynamically adjusted according to the traffic pattern and mobility prediction in order to achieve the maximum channel utilization, while guaranteeing different QoS constraints. The performance of proposed scheme was evaluated using Markov models. New call blocking probability, handoff call dropping probability, and channel utilization were used as benchmarks for the proposed scheme.

Key words: admission control; resource reservation; wideband code division multiple access; resource utilization

1 Introduction

Wireless cellular networks are expected to provide multimedia services with different quality of service (QoS) requirements. A typical example is the universal mobile telecommunication system (UMTS) which is required to support a wide range of applications, each with its specific QoS [1]. Since the multimedia services have different traffic characteristics, their QoS requirements may differ in terms of bandwidth, delay and dropping probabilities [2]. The radio resource management unit is responsible for fair and efficient allocation of network resources among different users with different classes. Also from the user’s point of view, the handoff attempt failure is less desirable than the blocking of a new attempt. Therefore, due to the limited resources in wireless multimedia systems, the efficient call admission control (CAC) and resource reservation (RR) schemes are needed to maintain the desired QoS.

Resource allocation in wireless and mobile networks with multimedia services is more complex than that in traditional voice-oriented wireless cellular networks due to the different QoS requirements for different services. In general, resource allocation in wireless networks can be divided into two basic policies: complete sharing (CS) policy and complete partitioning (CP) policy [3-4]. The CS policy allows all users an equal access to the bandwidth available at all times. The CP policy, on the other hand, divides up the available bandwidth into several separate sub-pools according to the user type. The CS policy gives higher system utilization due to its statistical multiplexing gain, but it needs to adopt the tightest QoS bound among all traffic types. The CP scheme provides independent QoS bounds for different traffic types at the expense of a possible lower overall utilization. In order to provide a compromise between the two policies, hybrid schemes have been proposed in the literature to combine the strength of both policies [3, 5]. Since handoff calls are more important than new calls, these hybrid policies, also called reservation policies, are mainly introduced to minimize the handoff dropping probability.

The most popular way to handle handoff requests is the guard channel (GC) scheme, which reserves a fixed amount of resources for handoff traffic [6]. The total channels are divided into two groups: the shared channels, which are occupied by new calls and handoff calls, and the channels, which are reserved exclusively for handoff requests. However, such a static approach is unable to handle variable traffic loads due to mobility. It obviously lowers the spectrum utilization. Also, there have been a considerable amount of GCs based schemes supporting voice and data in the integrated mobile networks [7-8]. In Ref. [7], a dual-thresholds scheme was proposed: one to reserve bandwidth for handoff voice calls, and the other for the data traffic either new calls or handoff calls. This scheme was based on giving higher priority to handoff voice calls. The objective is to minimize the handoff dropping probability and increase the channel utilization. However, as in the GC schemes, some of the reserved channels for handoff requests may be unused; at the same time, the new calls are blocked. Therefore, a dynamic GC scheme is required.

Dynamic GC schemes were discussed in the literatures to improve the system utilization while providing QoS to guarantee higher priority calls [9-11]. However, these schemes were studied for frequency division multiple access/time division multiple access (FDMA/TDMA) systems, and not completely suitable for CDMA systems. In Ref. [12], a multi-mode dynamic guard bandwidth scheme was proposed. It extended the traditional dynamic GC scheme to CDMA systems, and allowed different adaptive thresholds for newly originating and handoff calls. The threshold value was adopted using the arrival rate estimation and blocking probability constraints. An extension of this work that took into account the fairness among different classes was presented in Ref. [13]. In Ref. [14], the well-known two resource-sharing algorithms known as CS and CP were studied for WCDMA system. It is concluded that the CS policy achieves the best system utilization at the expense of unfairness between different classes. The CP policy achieves the fairness, but it has the worst system utilization. From the above survey of different proposals, it is clear that several tradeoffs exist in the above mentioned schemes. An improvement of the resource utilization may be achieved at the expense of QoS for different classes. In Ref. [15], a multi-cell CAC scheme for WCDMA systems that improved the system throughput and the performance over the single-cell admission control was proposed.

In this work, a new scheme was proposed that extends the concepts of resource sharing and reservation for WCDMA systems with the unique feature of soft capacity. The proposed scheme combined both high system utilization and the required QoS. In other words, the proposed scheme provides a balance between CS and CP-based CAC. Voice and data traffic were considered. The traffic was further classified into handoff and new requests. The reservation thresholds were dynamically adjusted according to the traffic pattern and mobility prediction in order to achieve the maximum channel utilization, while guaranteeing different QoS constraints.

2 Dynamic reservation scheme

2.1 WCDMA capacity and load estimation

The capacity of the WCDMA system is limited by the total interference it can tolerate. Therefore, it is an interference-limited system. Assume that there are M traffic classes in the system. The maximum capacity is achieved, when the cumulative interference becomes so large that the energy per bit to noise density ratio for users of class i, (E_b/N₀)_i, cannot be fulfilled. As each class has different requirements, the bit energy to noise density ratio for class i is given by

                                                                                              (1)

where W is the chip rate; P_i is the received signal power from class i; v_i is the activity factor of class i; R_i is the bit rate of class i; I_tot is the total received wideband interference power including the thermal noise power at the base station. Solving for P_i, we obtain

                                                (2)

Let L_i be the load factor for each user of class i as [1]

                                            (3)

The total load factor η is defined as the sum of load factors from all active mobile stations in the system, and is given by

                                                                   (4)

where N_i is the number of active mobile stations of class i at the home cell. The other-cell interference must be taken into account when calculating the load factor. Other-cell interference is the interference caused by the mobile users of neighboring cell denoted as I_oth. Define l_i as the ratio of other-cell interference to P_i, and then   Eq. (3) becomes [16]

                                                     (5)

The QoS classification is a very important feature of future mobile networks and UMTS networks as well. The gain of these classifications is the ability to manage the available resources while accounting for the user QoS. Additionally, QoS classes can be defined regarding the bit rate used. For every traffic class, a different load factor can be calculated using the respective values of the bit rate R_i, the activity factor v_i and the required (E_b/N₀)_i. Table 1 gives an example of UMTS traffic classes and the respective parameters of E_b/N₀, activity factor, and load factor L_i. The parameters in the table are taken from Ref. [1].

Table 1 Examples of load factor calculations for different traffic types [1]

For simplicity in calculation, the definition of the greatest common divisor (GCD) presented in Ref. [16] was adopted, and denoted as ?L. In order to calculate ?L, we simply turn all decimals to integers and find the greatest common divisor of the integers. Therefore, the load factor of each class can be represented as

                                                                         (6)

where g is a positive integer. By this mapping, the concept of channel allocation in FDMA/TDMA systems can be easily extended to WCDMA by considering ?L as a basic logical channel.

2.2 System model

In the proposed model, two classes of traffic are considered: real-time (RT) such as conversational and streaming traffic, and non real-time (NRT) such as interactive and background traffic. Furthermore, traffic is classified as new and handoff according to the type of request. Therefore, we have four priority classes, which are listed in Table 2.

Table 2 Priority classes

Assume that each class of traffic has different requirements, i.e. the traffic classes have different multiples of the GCD described above. And the arrival processes of the new and handoff calls have rates     and  for handoff voice, handoff data, new voice, and new data, respectively. Let η_max, η₁, η₂ and η₃ be the loading limits for Class 1, 2, 3 and 4, respectively, as shown in Fig. 1. Without loss of generality, the voice call increases the load by the basic GCD (?L), and the data call increases it by (n·?L) with an arbitrary integer n. Only the uplink is considered in this scheme. It is assumed that whenever the uplink call has been assigned to a channel, the downlink connection is established.

Fig. 1 Proposed CAC scheme

From Fig. 1, this sharing scheme with the predefined thresholds can be easily extended to a CS scheme by letting η₁=η₂=η₃=η_max. However, in the proposed scheme, these thresholds are not static. They vary as explained in the following subsection according to the traffic conditions and mobility of the users. The CP scheme is not considered here because of its poor efficiency in terms of resource utilization [3, 14].

2.3 Reservation and admission strategy

The objective of proposed dynamic reservation channels (DRC) scheme is to satisfy a desired dropping probability for handoff calls, and at the same time, reduce the blocking probability of new calls as much as possible. The mobility of the calls in a cell was defined as the ratio of the handoff call arrival rate to the new call arrival rate [17-18]. The handoff prediction schemes can be classified into two main types: one predicts the handoff traffic according to the prediction of mobility, and the other calculates the handoff probability according to the call duration and the call residence time calculations. In Ref. [17], a heuristic formula defined the acceptance probability as the probability that the base station permits new calls to allocate the wireless resources reserved for the handoff calls. In other words, the acceptance probability defines the fraction of the reserved channels that can be occupied by new calls. The total resources assigned to one type of traffic are divided

into normal resources and reserved resources. The normal resources are accessed by the new calls with a probability of one. The reserved resources for handoff calls are accessed by new calls according to the acceptance probability. The acceptance probability, P_ac,k(j), is dynamically computed, which is defined as [17]

                              (7)

where η_max is the maximum load allowed for new calls of type k for a certain acceptance probability; k=1, 2 for voice and data traffic, respectively; j denotes the current state of the system; η_th,k is the predefined threshold for the new class under consideration and α_k=λ_h/λ_n, k = 1, 2 is the mobility parameter for voice and data traffic, respectively. Figure 2 shows the reservation thresholds which are included in Eq. (7). In this equation, the thresholds   are equivalent to η₁, η₂ and η₃, respectively, as shown in Fig. 2.

Fig. 2 Proposed reservation scheme

In order to verify the heuristic formula presented in Eq. (7), Fig. 3 shows an example of the acceptance probability of the voice traffic for different values of the mobility parameter. The threshold is 80% of the total load. As shown in Fig. 3, the acceptance probability is increased as the call mobility is decreased, and it is decreased as the call mobility is increased under the same offered load. When the mobility parameter equals one, the acceptance probability is reduced linearly. When the mobility of calls is smaller than one, the acceptance probability is slowly decreased in order to provide better chances for the new calls. When the mobility of calls is larger than one, the acceptance probability is decreased abruptly in order to save the resources for handoff calls.

Fig. 3 Acceptance probability versus cell load at different values of mobility parameter for voice traffic

The admission criteria of the proposed scheme can be summarized as follows (see Fig. 1):

1) When a handoff voice call arrives with η+?L≤ η_max, it will be accepted. Thus, a handoff voice call will only be dropped if there are no more free resources in the system.

2) When a handoff data call arrives with η+n?L≤η₁, it will be accepted, otherwise it will be dropped.

3) When a new voice call arrives with η+?L≤η₂, the call is accepted. Otherwise, the base station checks for the arrival rates of the new and handoff traffic. If the acceptance probability, P_ac,k(j), is higher than zero, the channels reserved for the higher class (handoff /data) are occupied by new voice traffic. Note that although the handoff data have higher priority than new voice calls, a new voice call can occupy partly the channels reserved for handoff data even with high mobility of data traffic, as shown in Fig. 2. This in turn improves the grade of service (GoS) for real time traffic.

4) When a new data call arrives with η+n?L≤η₃, the call is accepted. Otherwise, according to the mobility parameter of the data traffic (or the acceptance probability), the channels reserved for the new voice traffic can be shared with the new data calls.

3 Performance analysis

For performance measurements, the dropping probability of handoff calls, the blocking probability of new calls, a cost function comprising metrics and the resources utilization should be studied. The cost function gives the dropping of a handoff call a higher weight than blocking a new one. The system described before can be represented as a multi-transition truncated M/M/η_max/η_max loss model. Multi-transition means that from a given state η, transition could lead to states η+?L and η+n?L with different transition rates depending on the traffic class of the admitted connection either voice or data call, respectively. Truncated means that the arrival rates change when the current load reaches the state given by the guard loading limits η . Let us define      and  Figure 4 shows a representation of the above system using Markov chains.

The closed-form solutions for steady-state probabilities of the Markov chain diagram shown in  Fig. 4 are not easily obtainable because of the eigenvalue problem of the transition matrix. An approximation for the above model was used according to Ref. [13] for which closed-form solutions are easily obtainable. Let  denote the average loading required for data and voice calls:

                                                                 (8)

Using the average loading increment  the system can be represented by a birth-death process, as shown in Fig. 5. The load is normalized to a multiple of

 as follows:

                                                                                    (9)

Now, it is easy to obtain the local balance equations, then, evaluate the performance of the proposed system. Balance equations can be extracted from Fig. 5. At the equilibrium state, the probability of move to any state is equal to the probability of leaving that state. Defining P_j as the probability that the system is in state j, we get:

At state (0):

At state (1):  that is,

At state (2):  Substitute for P₁ and P₂ into the last equation, we get

Fig. 4 Multi-transition M/M/η_max/η_max loss Markov model

Fig. 5 Simplified birth-death process

Similarly, the total balance equations can be derived as

                             (10)

where

                                                         (11)

and P₀ can be calculated with the help of the following equation:

                                                                     (12)

The rationale behind the proposed scheme is that the new call is blocked, when the cutoff threshold for accepting new calls is reached and the acceptance probability does not permit further sharing of the reserved load for the higher priority class. Therefore, better resources utilization is gained compared to the fixed threshold scheme as will be shown in the numerical results.

After obtaining all the steady-state probabilities, the voice call blocking probability P_bv, the handoff voice dropping probability P_dv, the data call blocking probability P_bd and the data call dropping probability P_dd are given as follows:

                          (13)

                           (14)

                                                                (15)

                                                                       (16)

The resources utilization is defined as the ratio of occupied resources to the total system resources. Since the number of occupied resources is a random variable, which depends on the system state, the average number of occupied resources is used. The resources utilization is defined as

                         (17)

The GoS is a performance metric that is used to evaluate the performance of the proposed algorithm and is defined as

             (18)

where P_d,j is the handoff dropping probability; P_b,j is the new call-blocking probability; j=1, 2 stand for voice and data traffic, respectively. Parameter β (β=10) indicates the penalty weight for dropping a handoff call relative to blocking a new one [18-19].

4 Numerical results

A comparison study has been made among the proposed DCR scheme, the fixed channels reservation (FCR) scheme, which is similar to the GC scheme, and the full sharing (FS) or CS scheme. It should be noted that the FS scheme gives the minimal blocking probability of new calls and the FCR scheme gives the minimal dropping probability of handoff calls. The performances of these schemes are compared in a heavily loaded system. The objective of the proposed scheme is to merge the advantages of both FS and FCR schemes.

The GCD discussed above is taken as the basic resource unit. The load factor of each application is defined in terms of GCD. So, let ?L=0.01, then, we have 100 logical channels in the system. The percentage of each arrival rate out of the total offered traffic is    and  It is assumed an exponential distribution for the remaining time until the next birth and next death with an arrival rate λ and service rate μ. The average call holding time has been taken as 180 s [13]. The loading limit percentages used have been selected as 100%, 90%, 80% and 70% for classes 1, 2, 3, and 4, respectively. The mobility varies through the analysis by varying the parameter α. A high mobility is considered at α=1.5. Unless stated in the discussion, a low mobility is considered as α=0.6.

Figure 6 shows the variation of the voice call blocking probability with the offered traffic. It is clear that the FCR scheme has the highest blocking probability for high loads. For low loads, the FS scheme has the minimum blocking probability, because it allows the new voice traffic to be allocated to the entire resources. With the traffic increasing, the new voice has lower chances in capturing the resources. The DCR scheme allows both new data and voice traffic to partially capture the resources reserved for the higher classes. So, the same performance is obtained with the FS scheme at high loads. The proposed DCR scheme has almost the same performance as the FS scheme at high loads. Figure 7 shows the effect of the traffic load variation on the handoff dropping probability of the voice traffic in the high mobility case.

Fig. 6 Effect of load variation on voice blocking probability

Fig. 7 Effect of load variation on voice dropping probability

The case of voice traffic with high mobility has been studied, because a high handoff rate is expected. It can be seen that the proposed scheme gives the minimum dropping probability at high loads as the FCR scheme, whereas the FS scheme gives the worst performance. For low traffic loads, the FCR scheme has better performance in terms of the dropping probability, but this difference diminishes with high loads. Figure 8 shows the blocking probability of new data calls. The FS scheme has the optimal performance. The DRC scheme still outperforms the FCR scheme. Regarding the dropping probability of handoff data calls, with high mobility, the FCR scheme has the best performance, as shown in Fig. 9. In the case of the DCR scheme, the new voice traffic occupies the load margin reserved for handoff data calls even with low mobility of voice traffic. So, this scheme may be considered a special case of the scheme presented in Ref. [7], in which mixed data (new and handoff) have the same threshold. It can also be seen that at high loads the three schemes have the same performance.

Fig. 8 Effect of load variations on data blocking probability

Fig. 9 Effect of load variations on data dropping probability

The variation of the GoS for voice and data traffic with the total offered load is shown in Figs. 10 and 11, respectively. The DCR scheme has the best performance (see Fig. 10) because a higher threshold is reserved for Class 1 and Class 3 can share the resources reserved for Class 2 even with low mobility of voice traffic. The GoS for data traffic is shown in Fig. 11. Note that the FCR scheme has the best performance. However, for high mobility, almost the two schemes have the same performance, as shown in Fig. 12. Another important parameter is the resources utilization, which is depicted in Fig. 13. It is shown that the FS scheme is the most efficient in terms of resources utilization. The DCR scheme is more efficient than the FCR scheme for a wide range of traffic and mobility variations. For high mobility, the performance of the DCR scheme is still better than that of the FCR scheme, as shown in Fig. 14.

Fig. 10 GoS for voice traffic

Fig. 11 GoS for data traffic

Fig. 12 GoS for data traffic with high mobility

Fig. 13 Resources utilization

Fig. 14 Resources utilization with high mobility

5 Conclusions

1) A new DCR scheme with multi-thresholds for WCDMA systems is presented. It can be considered as an extension of the well-known GC scheme. The DCR scheme is proposed using the dynamic loading limits η₁, η₂ and η₃ to give priority to handoff data calls, new voice calls, and new data calls, respectively. Thresholds change according to the predictive mobility and traffic patterns.

2) Detailed results obtained from the analysis show that giving higher threshold for handoff voice reduces the voice dropping probability, and at the same time, a high channel utilization is achieved. As a result, DCR scheme is able to guarantee a high QoS for different applications. It is also able to utilize the network resources efficiently.

References

[1] HOLMA H, TOSKALA A. WCDMA for UMTS: Radio access for third generation mobile communications [M]. England: John Wiley and Sons, Ltd, 2004.

[2] AHMED M H. Call admission control in wireless networks: A comprehensive survey [J]. IEEE communications Surveys and Tutorials, 2005, 7(1): 50-69.

[3] EPSTIEN B, SCHWARTZ M. Reservation strategies for multimedia traffic in a wireless environment [C]// Proceedings of IEEE Vehicular Technology Conference. Chicago, USA, 1995: 165-169.

[4] LAI F S, MISIC J, CHANSON S T. Complete sharing versus partitioning: Quality of service management for wireless multimedia networks [C]// Proceeding of 7th International Conference on Computer Communication and Networks. Lafayette. LA, USA: 1998: 584-593

[5] TEKINAY S, JABBARI B. A measurement-based prioritization scheme for handovers in mobile Cellular Networks [J]. IEEE Journal Selected Areas in Communications, 1992, 10(8): 1343-1350.

[6] HONG D, RAPPAPORT S S. Traffic model and performance analysis for cellular mobile radio telephone systems with prioritized and non-prioritized handoff procedures [J]. IEEE Transaction Veh Technology, 1986, VT-35(3): 77-92.

[7] YIN L, LI B, ZHANG Z, LIN Y B. Performance analysis of a dual threshold reservation (DTR) scheme for voice/data integrated mobile wireless networks [C]// Proceedings of IEEE Wireless Communications and Networking Conference, 2002: 1100-1104.

[8] HUANG, LIN Y B, HO J M. Performance analysis for voice/data integration on a finite-buffer mobile system [J]. IEEE Transactions Vehicular Technology, 2000, 49(2): 367-378.

[9] RAMANATHAN P, SIVALINGAM K M, AGRAWAL P, KISHORE S. Dynamic resource allocation schemes during handoff for mobile multimedia wireless networks [J]. IEEE Journal Selected Areas in Communications, 1999, 17(4): 1270-1283.

[10] NASSER N, HASSANEIN H. Dynamic threshold-based call admission framework for prioritized multimedia traffic in wireless cellular networks [J]. IEEE GLOBECOM Conf, 2004, 2(1): 644-649.

[11] CHEN X, LI B, FAG Y. A dynamic multiple-threshold bandwidth reservation (DMTBR) scheme for QoS provisioning in multimedia wireless networks [J]. IEEE Transactions on Wireless Communication, 2005, 4(2): 583-592.

[12] YU O, SARIC E, LI A. Adaptive prioritized admission over CDMA [C]// Proceedings IEEE Wireless Communications Networking Conference. Chicago, USA, 2005, 1260-1265.

[13] YU O, SARIC E, LI A. Fairly adjusted multimode dynamic guard bandwidth admission control over CDMA systems [J]. IEEE Journal of Selected Areas in Communications, 2006, 24(3): 579-592.

[14] AL-QAHTANI S A, MAHMOUD A S. Uplink call admission schemes for 3G and beyond mobile wireless networks supporting heterogeneous traffic [C]// IEEE International Conference on Information and Communication Technology. Damascus, Syria, 2006: 2657-2662.

[15] AL-NAHARI A Y, EL-DOLIL S A, DESSOUKY M I, ABD EL-SAMIE F A. Power-based Multi-cell call admission call admission control scheme for wideband-CDMA systems [J]. Computers and Electrical Engineering, 2010, 36(5): 935-947.

[16] PASCHOS G S, POLTIS I D, KOTSOPOULOS S A. A quality of service negotiation-based admission control scheme for WCDMA mobile wireless multi-class services [J]. IEEE Transaction on Vehicular Technology, 2005, 54(5): 1875-1886.

[17] KIM Y C, LEE D E, KIM Y S, MUKHERJEE B. Dynamic channel reservation based on mobility in wireless ATM networks [J]. IEEE Communication Magazine, 1999, 37(1): 47-51.

[18] SALAMAH M, LABABIDI H. Dynamically adaptive channel reservation scheme for cellular networks [J]. Computer Networks, 2005, 49: 787-796.

[19] ALQAHTANI S A, MAHMOUD A S. Dynamic radio resource allocation for 3G and beyond mobile wireless networks [J]. Computer Communications, 2006, 30(1): 41-51.

(Edited by DENG Lü-xiang)

Received date: 2010-12-18; Accepted date: 2011-05-27

Corresponding author: A. Y. Al-nahari; Tel: +20-010-8130973; E-mail: a_alnahary@yahoo.com