8/13/2019 New CallAdmission Lte Femtocells

1/5

A New Call Admission Control Strategy for LTE Femtocell Networks

Jie Wang, Yangfan QiuNational Mobile Communication Research Lab (NCRL), Southeast University

Nanjing, 210096, China

Email: wangjie@seu.edu.cn

AbstractThis paper introduces a call admission control

(CAC) strategy for LTE femtocell networks supporting

multimedia services with different classes of traffic and

diverse bandwidth requirements. In this work, the multi-

service CAC strategy composes of two parts: subscriber

authentication and queuing admission control. Theoretical

analysis and simulation are performed on three key

principles. Comparing with the typical strategy, simulation

results indicate that better probabilistic connection-level

QoS parameters are achieved by using the proposed strategy.

Keywords- femtocell; HeNB; LTE; CAC; CSG

I. INTRODUCTIONFemtocell is a low-cost and low-power cellular home

base station that works on the licensed spectrum andconnects mobile users to the operators core network overthe Internet via broadband backhaul. With the help offemtocell technology, the next generation mobilecommunication systems can provide faster and morereliable indoor communication [1]. As a result, there is agrowing interest in deploying multimedia services inindoor mobile cellular networks. However, several aspectsrelated to the macro/femto-cell hierarchical structure arestill remaining to be adapted. Call admission control (CAC)is one of such areas in need of adaptation to accommodatemultimedia services.

Typical CAC schemes of multi-service are completesharing (CS), complete partitioning (CP) and threshold [2].In the CS scheme, calls of every class of services share the

bandwidth pool [3]. Whereas, in the CP scheme, thebandwidth for each class of services is exclusivelyreserved [4]. In the threshold scheme, a newly arriving callis blocked if the number of calls of corresponding serviceis greater than or equal to a predefined threshold. It is

proven that the threshold scheme is optimal in two-classnetwork [5]. However, the threshold scheme is not optimalin general multi-service networks.

There are three access modes which allow users toaccess femtocell: close access, open access, hybrid access.The close access mode only allows the closed subscribergroup (CSG) members entering the CSGs femtocell, theopen access mode enables all users to share the resource

provided by the femtocell. The hybrid access mode is acombination of both, it affords other UEs to access theCSGs femtocell. Normally, the hybrid access femto BS isowned by CSG member subscribers, there isdiscrimination for service between CSG members andnon-CSG members. Service quality for CSG membersshould not be affected by non-CSG members. Therefore it

has become necessary that CAC considering prioritycontrol for CSG members.

Many works have been conducted in the area of theCAC strategy in hierarchical macro/femto-cell networks,

but most of them focus on the policies for different accessmodes. In [6], the authors considered the differentadmission control policies for femtocell operation,

proposed an operating region where femto BS hasmotivation to operate in open access mode. Anoptimization CAC mechanism in hybrid access mode

based on the discrimination for service between CSG

members and non-CSG members was described in [7].Reference [8] studied the femtocell access control strategyfor UMTS and LTE was provided based on 3GPP Release8 specifications. In this paper, we consider multi-servicecall admission control for LTE femtocell networks.

The rest of the paper is organized as follows. Nextsection describes the proposed call admission controlstrategy. The simulation scenario and parameters areexplained in Section III. Numerical results are discussed inSection IV and the paper is concluded in Section V.

II. PROPOSED CALL ADMISSION CONTROL STRATEGYA bearer is an IP packet flow with a defined Quality of

Service (QoS). E-UTRAN Radio Access Bearer (E-RAB)

refers to the concatenation of an S1 bearer and thecorresponding radio bearer. LTE uses the concept of E-RAB to route the packets of an Evolved Packet Core (EPS)

bearer between a serving gateway (S-GW) and a userequipment (UE) [9]. The task of call admission control inLTE is to decide whether or not to establish new E-RAB.

With the three access modes we mentioned before, thefemto BS is a private station, therefore the CAC strategyfor femtocell should compose of two parts: subscriberauthentication and admission control. The flowchart in Fig.1 describes the proposed CAC process.

This work is supported by the China MIIT Municipal Science andTechnology Project (2011ZX03001-006-02) and the National High

Technology Research and Development Program of China (863 Program2012AA011401)

2nd International Conference on Advances in Computer Science and Engineering (CSE 2013)

2013. The authors - Published by Atlantis Press334

8/13/2019 New CallAdmission Lte Femtocells

2/5

Figure 1. Femtocell call admission control process

In the following section, we describe three keyprinciples of the proposed call admission control strategy.

A. Threshold base on subscriber authenticationGenerally, admission control algorithm estimates the

demand for physical resource block (PRB) demandiPRB of

servicei , when a bear establish request arrives. We

assume base station has totaliPRB amount of PRBs can be

allocated to service classi . With the occupied PRBsused

iPRB and the remainder PRBs availiPRB , there are

following relationships:avail total used

i i iPRB PRB PRB= (1)

1

ukused i

i mcs

u u

RPRB

TB=

= (2)

Here uiR andmcs

uTB represent the rate of user u and

the amount of transport block (TB) with currentmodulation and coding scheme (MCS).

The criterion for admission can be expressed as follow:

demand avail

i iPRB PRB (3)

is the threshold decided during subscriberauthentication which has different policies for CSGmembers with high priority. As a result, of CSG

membersCSG

and of non-CSG membersnon CSG

have

following relationship:

CSG non CSG > (4

B. Queuing priority base on multimedia serviceFor the sake of clarity, multimedia services belong to

multiple types will be referred to three classes:conversational class, streaming class, best effort class.Conversational class service represents real-time trafficsuch as VoIP, therefore we focus on its handoff calldropping probability and allocate guard channel (GC) forthe handoff calls. Streaming class service is another kindof real-time traffic such as online video, but thanks to thevideo buffer it is not as delay-sensitive as conversationalclass service. Best effort class service represents non-real-time traffic such as FTP, it is more sensitive to error rate,and has low requirement for delay.

Our queuing model is shown in Fig. 2. In addition tothe GC for the conversational class service, the resourceof system is divided into two parts. One of them is thespecial resource for best effort class service, the other oneis the public resource which the real-time traffic can

perform preemption in case of radio congestion.We take the GC as a dynamic proportion rather than

fixed amount of PRBs. Newly arriving conversationalcalls are blocked if the amount of available PRBs are lessthan the dynamic proportion of total PRBs, however theconversational handoff calls are blocked only when allPRBs are exhausted.

Figure 2. Queuing model for multimedia service

Note that a service is associated with a queue, and thequeue length needs to be monitored. Once the length ofqueue exceeds a certain size, the next coming calls should

be discarded. Moreover, there are corresponding timersfor each queue; the call waiting in the queue would bediscarded when a timeout occurs.

335

8/13/2019 New CallAdmission Lte Femtocells

3/5

C.Dynamic GC base on reinforcement learningGC makes handoff calls have higher priority, so that

the handoff calls can get expected QoS requirement.However, allocating GC will certainly raise new call

blocking probability. We model the update process of thedynamic GC as a Markov decision process (MDP), whichconsists of a set of states and actions aiming at finding a

policy that minimizes the corresponding cost for each state[10].

In particular, at time t the system state is defined as

{ }, ,ho newt con cons G L L= . { }1 2, , , , ,l i jG g g g g g i j <

8/13/2019 New CallAdmission Lte Femtocells

4/5

Log-normal shadowing Standard deviation = 4dB

Auto-correlation distance 3m

Threshold 1CSG

= , 0.8non CSG

=

Weights 0.8ho

= . 0.2n =

Figure 3. 5 5-grid model

IV. SIMULATION RESULTSIn first set of simulations, we compare the performanceof complete sharing scheme, complete partitioning scheme

and proposed queuing priority scheme. The performancecan be quantitatively expressed in terms of probabilisticconnection-level QoS parameters such as call blocking

probability. As expected, CS can maintain high levelquality of high priority service such as VoIP, at theexpense of decreasing performance of low priority service.However, CP can ensure the performance of each class ofservices, but the resource utilization is relatively low.

Numerical results show that the service of both highpriority traffic and low priority traffic is guaranteed by theproposed scheme, and moreover, the resource utilization ispromoted.

Due to the limited space we just show a part ofsimulation results. As shown in Fig. 4, the blocking

probability of VoIP increases with the call arrive rateincreases. However, the proposed queue priority scheme

provides significant reduction in blocking probabilitycompared to the complete partitioning scheme. Figure 5illustrates that the proposed scheme make the PRButilization of CBR as high as complete partitioning scheme.

Figure 6 shows the average cost as a function ofhandoff rate . The cost is calculated as Eq. (5) while

VoIP accounts for 50 percent of the overall traffic. It canbe observed that as increases the cost of rigid division-

based GC increases, however, proposed dynamic GCmaintain the cost at a low level.

V. CONCLUSIONIn this paper, a multi-service call admission control

strategy for LTE femtocell networks is proposed. Theproposed strategy considers multiple classes of serviceswith different QoS requirement. Three related componentscomprise the main building block of the strategy: (1)threshold base on subscriber authentication, (2) Queuing

priority base on multimedia service, (3) Dynamic guardchannel base on reinforcement learning. Simulation resultsdemonstrate that our proposed strategy can guarantee thecall blocking probability for each class of traffic, while theresource utilization is also promoted. It is also shown the

proposed dynamic guard channel scheme is a solution tothe contradiction between call handover dropping and

congestion of new calls.

Figure 4. Call blocking probability of VoIP

Figure 5. Resource utilization of CBR

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.50

0.05

0.1

0.15

0.2

0.25

new call arrive rate (calls/second)

callblocking

probability

complete sharing

complete partitioning

proposed queuing priority

0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.50.1

0.15

0.2

0.25

0.3

0.35

call arrive rate (calls/second)

resource

utilization

complete sharing

complete partitioning

proposed queuing priority

337

8/13/2019 New CallAdmission Lte Femtocells

5/5

Figure 6. Average cost as a funciton of hanoff rate

REFERENCES

[1] V. Chandrasekhar, J. G. Andrews, and A. Gatherer, Femtocellnetworks: a survey, IEEE Communications Magazeine, vol.46, no.9, pp. 59-67, Sept. 2008.

[2] Jihyuk Choi, Taekyoung Kwon, Yanghee Choi, Naghshineh M,Call admission control for multimedia services in mobile cellularnetworks: a Markov decision approach, ISCC 2000, July 2000,

pp.594-599.

[3] R.A.Guerin, Queuing blocking system with two arrival atreamsand guard channels, IEEE Transactions on Commnication Vol. 36,

No. 2, pp.153-163, Feb. 1988.

[4] S. A. AlQahtani, Hassan Mahmoud and A. Alshanyour,Performance evaluation and analytical modeling of noveldynamic call admission control scheme for 3G and beyond celluarwireless networks, AICCSA 2007, May 2007, pp. 631-638.

[5] Keith W. Ross and Danny H. K. Tsang, The stochastic knapsackproblem, IEEE Transactions on Communications, Vol. 37, No. 7,pp.740-747, July 1989.

[6] M. E. R. Khan, Fair admission control policies for femtocelloperation, ICUFN 2012, July 2012, pp. 175-179.

[7] Seung Que LEE, RYU Bying Han and Nam-Hoon PARK, Calladmission control for hybrid access mode femtocell system,WiMob 2011, Oct. 2011, pp. 512-516.

[8] A. Golaup, M. Mustapha and L. B. Patanapongpobul, Femtocellaccess control strategy in UMTS and LTE, IEEE Commun. Mag. ,Vol. 47, No. 9, pp. 117-123, Sept. 2009.

[9] Stefania sesia, Issam Toufik, Matthew Baker, LTE-The UMTSLong Term Evolution: From Theory to Practice, Wiley, 2011.

[10] M. E. Harmon and S. S. Harmon, Reinforcement learning: Atutorial, 2000.

[11] C. J. Watkins and P. Dayan, Technical note: Q-learning,Machine Learning, Vol. 8, pp. 279-292, 1992.

[12] M. Simsek et al, An LTE-femtocell dynamic system levelsimulator, in proc. IEEE smart antennas (WSA), 2010

international ITG Workshop, pp. 66-71, Feb. 2010.[13] IEEE 802.16m Evaluation Methodology Document (EMD).[14] Giuseppe Piro et al., Simulating LTE cellular systems: an open

source Framework, IEEE Trans. Veh. Technol., Vol. 60, No. 2, pp.498-513, Oct. 2010.

[15] Video trace library, [OnLine] Available: http://trace.eas.asu.edu/.[16] Hjelm.B, admission control in future multi-service wideband

direct-sequence CDMA systems, IEEE VTC-FALL 2000. Vol. 3,pp. 1086-1093, Sept. 2000.

0 0.05 0.1 0.15 0.2 0.250

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

0.5

handoff rate

average

co

st

rigid division-based GC

proposed dynamic GC

338