Improving the Efficiency of Call Admission Control in Wireless Cellular Communication Networks by Frequency Sharing Techniques

8/12/2019 Improving the Efficiency of Call Admission Control in Wireless Cellular Communication Networks by Frequency Sha

1/14

International Journal of Computer Trends and Technology (IJCTT) volume 9 number 3 Mar 2014

ISSN: 2231-2803 www.internationaljournalssrg.org Page 133

Improving the Efficiency of Call Admission Control in WirelessCellular Communication Networks by Frequency Sharing

Techniques

Vikas Solanki 1, M. Qasim Rafiq 2 1( Department of Computer Science Engineeri ng & I T, Mangalayatan Uni versity, Aligarh, I ndia)

2 (Department of Computer Science Engineeri ng, AM U, Ali garh, I ndia)

ABSTRACT : In this paper, a two-tier cellularcommunication wireless network is characterizedby overlapping the service area for managing thenew calls users having different mobility speed.The overlapping property of the two-tier system

provides the advantages that share the traffic loadto improve the efficiency of new calls subscriberwith reservation of channels (guard channels) in

cell to handle the ongoing old calls (handoff calls). Microcells are used to provide the services to slow- speed, high-intensity traffic area users andmacrocells are overlaid over more than onemicrocells cater mainly too lower density, high-

speed users. The two-tier of microcells andmacrocells provide the secondary resource which

provide the service to new calls as well as handoffcalls with guard channels by overflow the slow

speed users in macrocell and sharing the frequencyin vertical as well as in horizontal directions in theupper layer. In this paper, we tried to optimize useof resources by using advantage of overlapping

coverage of two-tier wireless cellular networks and frequency sharing techniques like VDFS and HDFS. The call lose probability of new calls aredeveloped through numerical analysis. The VDFSand HDFS schemes are proposed and comparedwith the existing schemes of CAC. The result showsthe new proposed schemes are more efficient.

Keywor ds - cellular network, two-tiercommunication system, load redirection, guardchannel, frequency sharing.

1. INTRODUCTION

In order to provide the efficient services against thehuge demand of the wireless cellularcommunication networks, is the challenging issue.The radio spectrum range is limited and cant becatch up with increase of users demand for thewireless cellular communication networks. Torelease that stress one way is to design microcells.Moreover, microcells are not advantageous in the

service area where user population is sparse withslow and high speed subscribers. Small cellsystems induce an increase handovers by highspeed mobile subscribers. Micro-macrocell overlaystructures can overcome with these difficulties.Overlapping property of two-tier structure providesthe advantages that share traffic load to improvethe efficiency of call admission control. Call

admission control (CAC) schemes are critical forthe success of future generation of wirelessnetworks. It provides the users to access a wirelessnetwork service. On the other hand, these are thedecision making part of the network carriers that

provide services to users with guaranteed qualityand achieving maximum possible resourceutilization. It is therefore conceivable that CAC

policy is one of the critical design considerations inany wireless network. CAC schemes provide theservices for both i.e. new as well as old call users.

New call users are those who make the newconnection in current cell for communication and

required the frequency for connectionestablishment. On the other hand, old call users arethose subscribers whose connections are alreadyestablished, but due to mobility they have tochange the cell for continuous communication.Therefore, they further require the frequency inanother neighbouring cell for connectionestablishment, without breaking the connection.Earlier, CAC scheme called New ConnectionEstablishment (NCE) or in short New Calls andlatter CAC scheme is known as Handoff calls.Therefore, CAC schemes are categorized in two

parts: one is for new calls and the second is for on

going (old) calls.

A two-tier cellular network is characterized bymacrocells and microcells overlapping in theservice area. Larger cells, reside on the top layerare called macrocells, while smaller cells reside onthe bottom layer are called microcells. Subscribersare assigned to microcell or macrocell based ontheir mobility speed.
http://c/Documents%20and%20Settings/User/My%20Documents/Downloads/www.internationaljournalssrg.orghttp://c/Documents%20and%20Settings/User/My%20Documents/Downloads/www.internationaljournalssrg.orghttp://c/Documents%20and%20Settings/User/My%20Documents/Downloads/www.internationaljournalssrg.org


2/14



Systems employing multitier cells have beenconsidered in a number of publications. Severalmethods, for handling new calls and handoff trafficof the defined mobile subscriber speed classes are

proposed and performance measures such as the

probability of new call blocking, forcedtermination, and traffic capacity have beendetermined. In the case of a speed-insensitiveselection mechanism, call originations are assignedto a default cell layer which is, in most cases, thelowest (microcell) layer [1-2]. It is proposed todirect a new/handoff for the appropriate tier basedon its previous speed [3-4]. However, when there isno available channel on the preferred tier, the callwill be directed to the other (un-preferred) tier.This is called an overflow. If a speed sensitiveselection mechanism is used, arriving calls can bedirected to the specific cell layer that depends on

the speed class of the mobile station. Many worksare also directed in the direction to optimize the performance of the system based on the factorssuch as roaming speed of users, level of cloudinessof an area, location management, and channelmanagement etc. [5-11]. Two-way overflows areconsidered in between both tier and a take-backscheme is also proposed in which call is redirectfrom an un-preferred tier to the preferred tier at thetime of handoff take place [12] and a channelrearrangement scheme is proposed by forcing ahandset in the overlapping area to take an earlyhandoff permanently [13].

In this paper, a two-tier system is proposed withGuard-Channels that are reserved to handle thehandoff calls only. The aim of this work is toimprove the performance of new calls by usingoverlapping property of the two-tier system that

provides the advantage to share the traffic loadwith frequency sharing techniques in betweenmicro-macrocell. By using the overlapping

property of two-tier system the load of the cell may be transferred from lower tier to upper tier andvice-versa.

2. RESULT AND DISCUSSION

Let us assume that a macrocell is overlappingwith n microcells, and a macrocell neighbouring tok macrocells. When a channel request arrives at themacrocell, and if the macrocell has no freechannels, then the system force one of the slowcalls exist in the macrocell to move into one of theother n overlapping corresponding empty

microcells. On the contrary, when a channelrequest arrives at a microcell, and if the microcellhas no free channels, then it can be either overflowto the macrocell, or force one of the calls in themacrocell to move into one of the other n-1

overlapping empty microcells. Thus, a channel becomes vacant in the macrocell and new call can be overflowed to the overlaid macrocell. It isobserved that such frequency sharing provides lotof flexibilities to shift the load among the cells onthe two-tiers in vertical direction, so the scheme iscalled the Vertical Direction Frequency Sharing(VDFS). In this paper, we restricted the fast speeduser to overflow in microcell and if slow speedusers overflowed to un-preferred tier then it willnot return automatically, until it forced to overflowfor serving the new/handoff calls in macrocell. Inthis system a fast moving calls do not shift to

microcells and avoid the more handoff. Thus,overall system can avoid the more call dropping probability. The other scheme Horizontal DirectionFrequency Sharing (HDFS) [13] works only inupper-tier with k neighbouring cells, when VDFSscheme fails and not able to provide the service forarriving calls. In the proposed system, somechannels are reserved called Guard Channels, thatare used only for providing the services to handoffcalls. Therefore, guard channels cant be used forserving new call or overflow the new call in un-

preferred tier. The proposed system used VDFSscheme prior to HDFS scheme to share the

frequency in both tier for handle the traffic load.Table 1 shows the redirection of load in differentschemes. Numerical analysis shows the

performance comparison of different schemes and proposed schemes.

Table 1: Redirection of load in different schemes

SchemeReference

Strategy No. ofredirect

12 Overflow+Take Back 1

13 Rearrange+Overflow k+1

ProposedVDFS

VDFS with Guardchannels

n

ProposedHDFS

HDFS with Guardchannels

n+k


3/14



2.1 Proposed VDFS and HDFS Strategy

Let us consider that there is a channel requestarriving at the macrocell M or one of n microcellsmi. If no channel is available in the cell to satisfy

that request then proposed VDFS and HDFS strategy will take place, trying to find a channel forserving the said request. The VDFS strategy tries tofind a channel by shifting calls in the verticaldirection, i.e., from one tier to the other tier. WhenVDFS strategy fails to serve the request, HDFSstrategy tries to find a channel by shifting calls inthe horizontal direction on the higher tier.Simulation results shows that the proposedstrategies improve the efficiency of wirelesscellular networks even it contains the guardchannels [14] to improve the performance of thehandoff calls. In the following two sub-sections, we

discuss channel sharing in vertical and horizontaldirections. The first sub-section is for the VDFSstrategy, while the combination of the two sub-sections is for the later strategy i.e. for HDFSstrategy.

2.1.1 Vertical Direction Frequency Sharing(VDFS)

In VDFS strategy, the system transfers thecalls between the two-tiers. The tiers may be eitherhomogeneous or heterogeneous. In this work, tiers

are considered as homogeneous. In the following,we separate the discussion into calls arriving at thelower tier and at the higher tier. When there is achannel request arriving to microcell (lower tier)mi, 1 i n, the operation is showing by theflowchart A ( Fig. 1). On the contrary, flowchartB ( Fig. 2) shows the operation encounteredduring the request arriving to macrocell M (Uppertier).

As shown in Fig. 3, a slow new subscriberS1 arrives at microcell m 1, as represented by Vm1,that has no free channel (Free channels means, the

number of available channels except the GuardChannels to handle the request for serving the

proposed schemes/new calls in a cell).

The new arrival request Vm1 is handling as peralgorithm shown by the flow chart A. Thereforeas per flowcha rt A the operation 3 (Op. 3) isapplied and subscriber will be overflowed tomacrocell M because microcell m1 has no free

channel to serve the request. But the macrocell Mis also full, therefore the frequency sharing invertical direction takes place and tries to pick theslow subscriber, which is currently served inmacrocell M and then said subscriber can be

handoff to corresponding microcell with a freechannel. In the given example, shown in Fig. 3, theuser P1 can be shifted to a channel in microcell m 2.After performing the operation 4 (Op.4), thechannel is released by P1 user and macrocell M can

be used by S1new subscriber (Op. 5).

2.1.2 Horizontal Direction FrequencySharing (HDFS)

In this strategy, the calls are transferredhorizontally with in upper tier. If previous strategyVDFS fails then HDFS take place. This strategytakes place by forcing the subscriber on macrocellM i, 1 i k to early handoff in neighbouring cell asshown in flowchart C( Fig. 4). Two newsubscribers S1 and S2 arrive at microcell m1 andmacrocell M respectively; both cells have alreadyrun out of channels (See Fig. 5). Subscribers P5and P6 are two possible candidates who can takeearly handoff to M3 and M2 macrocellsrespectively. Suppose that user S1 arrive first tomicrocell m1that has no free channel(s), and thenvertical handoff occurs to upper layer macrocell M,which is also run out of channels. But Fig. 5 shows

that all corresponding microcell, to overlaidmacrocell M also goes to out of channels, sooperation 4 and 5 of flowchart A is not possible. Also, the failure of our VDFS implies that there isno free channels in microcells m 1, m2, m3,., m n, covered by macrocell M. In such situation, HDFSscheme is applied in upper tier (Fig. 4). Thesubscriber P5 now takes early handoff to macrocellM3 to vacate the channel and may be allocated thatchannel to user S1.

3. ANALYSIS MODEL DESCRIPTION

It is assumed that cells are circular inshape and each macrocell covers n microcells.Moreover, all cells in the same tier are statisticallyidentical, and thus we can focus on the behaviourof only one cell and its interaction withneighbouring cells.


4/14



No

No No

Yes Yes Yes

Figure 1: Flowchart A showing Operations occurred on Call request arrived to microcell

Flowchart A: Operations occurred on Call request arrived to microcell

No No

Yes Yes

Figure 2: Flowchart B showing Operations occurred on Call request arrived to macrocell

Op.1:Arrival of newcall request in m i

IsChannel available

in m microcell

Start

Op.2: Assign the channelin m i to the request

End

IsChannel available in

M macrocell

Op.3:Overflow the request& assign the channel in

M to the request

Op.4:Pick any slow user call in M such

that calls corresponding microcell m j tovacate the channel in M

Op.5:Overflow the request in Mmacrocell


any m j microcell

Op.6:Call Blocked or useHDFS scheme(if available)

Op.7:Arrival of newcall request in M

macrocell

IsChannel available

in macrocell

Start

Op.8:Assign thechannel in M to the

request

End

Op.9:Pick any slow user call in M suchthat calls corresponding microcell m j to

vacate the channel in M

Op.10:Assign the channel in M tothe request


any m j microcell underthe coverage of slow

user call in M

Op.11:Call Blocked or useHDFS scheme (if available)


5/14



Macrocell

Fast User

Slow -User

microcell : Representing, cell has free channel(s)

: Representing, cell is fully occupied Sn: Call Request

Pn : Present other subscribers

Figure 3: Channel sharing in the vertical direction for slow users

The radius of the circle for a subscriber totake a normal handoff on a macrocell is R n, andthat on a microcell is r n. However, since there will

be some overlapping between twomacrocells/microcells, the radius of the circle for asubscriber to take an early handoff on a macrocellis R e, and that on a microcell is r e as shown in Fig.6.

The arrival rates of new calls and handoffcalls for both low and high speed users are assumedto be Poisson processes. The Poisson distribution isa suitable approach to represent handoff calls [15].

Handoff calls have burst characteristics, sooccur at different rates. The cell dwelling time(CDT) is the time a mobile user spends in a cell

before it is handed off to another cell. It depends onthe speed of the mobile user and the size of the cell.The cell dwelling time [12] can be calculated asfollows:

1 r

=

d 2V

P4 P1

M

P3 P2

m 3

m 2 m 1

m 4

S2

S1

Vm1

Vm2Vm3

Vm1


6/14



Where V is the velocity of subscriber and r is theradius of the cell.

Note: Subscript s and f are indicating slow andfast user respectively while superscript m and M

represent microcell and macrocell respectively.

The inverse of the cell dwell time is thecell cross-over rate (CCOR) and therefore, for slowand fast subscribers in macrocell are and

respectively as follows:

;

The handoff probabilities [16] for fast andslow subscribers in a macrocell are

Where is the inverse of the mean

unencumbered call duration time. Theunencumbered call duration (UCD) of a call is theamount of time that the call may remain in progressif it can continue to complete without beingdropped, and it also follows a negative exponentialdistribution. The mean channel occupancy time(COT) / session duration is the mean of minimumof unencumbered call duration and cell dwell time[16], which should be

for fast and slow subscribers respectively.

All the derivations that are discussedabove for macrocell can be recalculating formicrocell and can be represented as follows:

Some more traffic parameters that areused in performance analysis of VDFS and HDFSare as follows:

A. For macrocell

Description subscriberfast slow

1. Traffic rate of new calls -2. Traffic rate of handoff calls3. Traffic rate of overflow calls -4. Traffic rate of vertical -

direction calls5. Traffic rate of horizontal

direction calls6. Aggregate traffic rate

B. For microcell

Description subscriberfast slow

1. Traffic rate of new calls2. Traffic rate of handoff calls3. Traffic rate of overflow calls

4. Traffic rate of verticaldirection calls

5. Traffic rate of horizontaldirection calls

6. Aggregate traffic rate


7/14



4. PERFORMANCE MEASUREMENTAND ANALYSIS

In this section, analysis is made of proposed model described above. The traffic flowsinclude new, handoff calls, and those incurred byvertical and vertical-horizontal direction frequencysharing. These traffic flows are all assumed tofollow the Poisson process. It is assumed that C T and C R are the total number of available channels

and guard channels for handoff calls in each cellrespectively. Therefore, vacant number of channelsto handle the new calls are C A=CT-CR . If the freechannels in a cell are greater than C R then there will

be no problem to handle handoff as well as call set

up of new calls. But, if any cell has less or equalnumber of channels than C R , a free channel performing only handoff and though new calls, oroverflow the slow speed user/forced handoff inupper tier are handled by proposed VDFS and /orHDFS schemes.

No No

Yes Yes

Figure 4: Flowchart C showing Operations occurred on Call request to HDFS

Op.13: Pick any slow user call in M i such that calls corresponding

microcell m j to vacate the channelin M i

Start

Op.14: pick M i neighbouringmacrocell that has at least one

vacate channel to HDFS for handoffthe subscriber Pj from M resident

in the area covered by both M & M i to early handoff

Op.12:when VDFS fails than callrequest to HDFS arrived

IsAny macrocell Mi has

at least one vacatechannel for HDFS

Op.15:enforce the subscriber Pj toearly handoff in Mi for vacate the

channel in macrocell M, and assignthe vacated cahnnel to the request

or overflow the request to M

IsChannel available in any

m j microcellcorresponding to M i

macrocell

Op.16:Request Blocked

Stop


8/14



Macrocell

Fast User

Slow User

microell

: Representing, cell has free channel(S)

: Representing, cell is fully occupied Sn: Call requestPn: Present other suscriber

Figure 5: Channel sharing in the horizontal direction for slow subscribers and fast subscribers

microcell/macrocell

r n / R n Early handoff area Overlapping area

r: microcell radius

r e / R e R:macrocell radius

Handoff area Figure 6: Radii of the circles for the subscribers to take normal handoff and an early handoff

S1

HM3HM2

M2P4 P1

M P6

P3 P2

M3

HM1 P5

M4

S2

HM3

m 6 m 5

m 8m 3

m 2 m 1

m 4

P7


9/14



4.1 Performance of Using Vertical DirectionFrequency Sharing

The target is to give the efficientscheme for handling the new calls, so that system

will reduce the call lose probability of new calls.Therefore, the calls l ose probabil iti es and ofnew calls for fast and slow subscribers respectivelyare derived. These are the probability for a channelrequest being refused after the vertical directionfrequency sharing.

Therefore call loss probabilities are:

Where is the failure probability of the sharingthe vertical direction

(1)

Where (1 ) is the probability of microcell hasat least one free channel and is the probability

having a slow subscriber that is currently served bythe macrocell and also covered by the microcell toserved them.

(2)

Where represents the probability

of all slow subscriber serving in macrocell are notlocated in this particular microcell and is the

probability of slow subscribers located in themacrocell for the same area.

The , and is the probability thata mobile subscriber has no free channel in amacrocell and microcell respectively.

(3)

(4)

Here and are the traffics

contributed by the subscribers on macrocell andmicrocell respectively for available channels.

Let variables , , and denotes the

arrival rates and , , and , denotes the

service rates.

Next, it is need to calculate the aggregatetraffic , , and by analysis model [17].These traffics are composed of new calls, handoffcalls, overflow calls, and channel sharing calls.Variable is aggregate traffic rate incurred by

new calls and handoff calls into macrocell by thefast subscribers:

Where

Here means the handoff rate, is the aggregatetraffic rate itself successfully stays in the macrocell

times with the handoff probability

. Similarly, is the aggregate traffic rateincurred by overflow calls and handoff calls into amacrocell by slow mobile subscribers.

Where


10/14



Here means the overflow rate incurred byoverflow from the n microcells covered by themacrocell and is the handoff calls into amacrocell by slow mobile subscribers, whichequals the slow subscribers successfully staying on

the high tier times the handoff probability , that is

The traffic rate of is incurred by new calls,handoff calls, and calls caused by channel-sharingfor slow subscribers:

Where is the handoff calls equals the slowsubscriber successfully handoff on lower tier

and is caused by vertical direction frequencysharing strategy,

Here is the load caused by the verticaldirection frequency sharing by slow subscriber inthe physical area covered by a macrocell (includingone macrocell and n microcell) and times is the

probability that a subscriber in macrocell can berearranged to a microcell but only a fraction 1/n ofthe load will be injected to the microcell. The rate

can be derived as follows:

,

It equals the new call arrival rate andhandoff call rate of slow subscribers into the nmicrocells , is times probabilitythat they see no free channel in the local microcell,and times probability that they see no freechannel in the macrocell.

Finally the term is the ratio of vertical

direction frequency sharing flows by slowsubscribers into microcells.

4.2 Performance of Using Horizontal DirectionFrequency Sharing

In this section, analysis is made of the proposed Horizontal Direction Frequency Sharingscheme. If a mobile subscriber call request foundno free channel in its local cell then previouslydiscussed VDFS take place first. HDFS schemetakes place, if VDFS scheme fail to perform. Thegoals is to drive the call lose probabili ties and

of new calls for fast and slow subscribersrespectively.

The call loss probabilities are:

Where probability is the failure probability ofvertical direction sharing as given in 1, and is thefailure probability of horizontal direction sharing isas follows:

Where is the failure probability of verticaldirection sharing for macrocell M i as given in 1,and is the probability for staying at least onesubscriber in early handoff area given as:

, and can be derived as (3) and (4)

respectively, but their values are different fordifferent aggregate traffic rates. The horizontaldirection sharing affects only traffic flows onmacrocell, therefore in microcell are same asthat in the vertical direction frequency sharing asdiscussed in section 4.1, but their value aredependent on , and when derived as forhorizontal direction sharing.


11/14



In macrocell, the aggregate rateincurred by new calls, handoff calls, and horizontaldirection sharing calls for fast subscriber:

+

caused by horizontal direction sharing can becalculated as:

It equals the new call arrival rate and handoff rateinto macrocell , times probability thatthey found no free channel in the macrocell ,times probability that they fail in vertical directionsharing , and times probability that at least one

subscriber staying in early handoff area .Similarly is the aggregate traffic rate incurred

by overflow calls, handoff calls, and horizontaldirection sharing in macrocell by slow subscribers:

Where is horizontal direction sharing by slowsubscriber,

which equals the new call arrival rate and handoffrate of slow subscribers into the n microcells, times the probabilities that they

found no free channel in the local microcell ,and neither in the macrocell , times the

probability that they fail in vertical directionsharing , and times probability that at least onesubscriber staying in early handoff area .

5. NUMERICAL EXAMPLES ANDDISCUSSION

We consider 4 cases denoted from (a) to(d) for comparison as follows:

a) The call loses probabilities of Take Back (TB) scheme in reference [12] for fast and slowsubscribers are denoted by , andrespectively.

b) The call loses probabilities of ChannelRearrangement (CR) scheme in reference [13]for fast and slow subscribers are denoted

by , and respectiely.c) The call loses probabilities of Vertical

Direction Frequency Sharing (VDFS) schemefor fast and slow subscribers are denoted by, and respectively.

d) The call loses probabilities of HorizontalDirection Frequency Sharing (HDFS) schemefor fast and slow subscribers are denoted by

, and respectively.

It is assumed that the total traffic to the entire areafollows the Poisson process with the rate and thefraction q of this traffic from slow mobilesubscribers.

To compare the different strategies listed (a) to (d),we assumed some parameters as shown in Table 2

In Table 2, q is tune the amount of slowsubscriber in an area and n takes care of sizedifference between macrocell and microcell. Hereassume that n=4, q=0.5 and the mean holding timefor a call is 140 seconds.

Figure 7: Comparison of numerical analysis on callloss probability with (.2 to.6) call arrival rate forfast subscribers


12/14



Table 2: List of parameters taken for performance comparison

S.No. Parameter name Value

Macrocell Microcell1 Radius 800 m 400 m2 Average velocity 40 km/hr 4km/hr3 The call arrival rate of slow subscriber q q/n 4 The call arrival rate of fast subscriber (1 - q) 05 Number of channels 37 96 Guard channels 8 2

Table 3: Number of assumed channels in the different schemes to be compared

S.No. Scheme Guard Channels(microcell,macrocell)

Available number of channels- C A Slow subscriber Fast subscriber

1 TB scheme 0 , 0 9 372 CR scheme 0 , 0 9 373 VDFS 2 , 8 7 294 HDFS 2 , 8 7 29

Figure 8: Comparison of numerical analysis on callloss probability with (.2 to.6) call arrival rate for

slow subscribers

To different schemes mentioned above from (a) to(d), the calls lose probability is calculated andcompared (Fig. 7 and 8) for fast and slowsubscribers. It is assumed that for fast and slowsubscribers, the available channels in the cellshown in Table 3.

It is found that, for fast subscribers, the CR schemeredirects the traffic only to neighbouring macrocell,if the current macrocell M is fully occupied. As perFig. 7, it performs the worst as it does not releasethe traffic effectively. The TB scheme overflows acall to the overlaid microcell, with take-back

strategy, at the time of handoff and thus performs better than CR for fast subscribers. The VDFSscheme performs better than CR because it pushesslow subscribers on the macrocell to its overlaidmicrocells even the VHFS scheme has less numberof channels as compared to CR schemes. Thechannels are reserved (guard channels) for handoffcalls in proposed schemes. There are multiplemicrocells to take the load of a macrocell. Fig. 7shows that VDFS scheme has less call loss

probability compare to CR scheme. HDFS scheme performs better than VDFS, because it alsotransfers the load to neighbouring.

For slow subscribers the CR scheme perform better than TB, as it has channelrearrangement strategy with overflow scheme, thathas more redirecting choices as compare to TB(Fig. 8). The VDFS and HDFS schemes performsimilar trend for both fast and slow subscribers.According to Fig. 8, the call loss probability of CRand HDFS scheme is same even the HDFS scheme


13/14



has less channels, reserve as the guard channels forhandle the handoff calls.

Figure 9: Comparison of numerical analysis on callloss probability with (.7 to1.1) call arrival rate forfast subscribers

Figure 10: Comparison of numerical analysis oncall loss probability with (.7 to1.1) call arrival rate

for slow subscribers

The proposed schemes are suggested to provide the service to slow speed users with highintensity traffic area and macrocells are overlaidover more than one microcell cater mainly to lowerdensity with high speed subscribers. The VDFSand HDFS schemes performs similar trend for fastand slow subscriber. Now, as soon as the total

traffic rate is increased to the entire area, the valueof tune the strength of slow subscribers is increasedto q=0.7, It is also observed that the results for

proposed schemes are too good in comparison toTB, and CR schemes (Fig. 9, 10). HDFS scheme

performs even better than VDFS scheme. The callloss probability of the proposed scheme is,therefore, better than TB, and CR schemes in thearea where services to slow speed users with highintensity traffic area, and high speed subscriberswith lower density area, even that the proposedschemes reserve the guard channels for the handoffcalls (Fig. 9 and 10).

6. CONCLUSION

In this paper, the VDFS & HDFS schemes perform

better for new arrival calls; even it contains lesschannels in comparison to conventional strategiesfor handle the new calls. The proposed schemeshave guard channels as a reserve channels tohandle the handoff calls only. An analytical modelhas been developed to derive some useful

performance indices. The methods to improve thecall loss probability in the service area that containsuser population in sparse with slow and high speedusers is proposed by taking the advantage ofoverlapping coverage of two-tier system. Thus, it is

possible to share the load in between two-tier ofhighly populated area for slow subscribers and less

dense to fast subscribers. It is also found that asignificant reduction is obtained in call loss probability for VDFS and HDFS scheme incomparison to TB and CR schemes. The VDFS andHDFS schemes performs similar trend for fast andslow subscribers also. The proposed schemesshows better performance in comparison to TB andCR schemes for both kinds of subscribers, even thetotal traffic rate is increased to entire area, value oftune the strength of the slow subscribers isincreased.

REFERENCES

[1] W. M. Jolley and R. E. Warfield, Modeling and analysis oflayered cellular mobile networks, in Teletraffic and DataTraffic in a Period of Change, ITC-13 , pp. 161 166, 1991.

[2] X. Lagrange and P. Godlewski, Performance of ahierarchical cellular network with mobility-dependent hand-overstrateg ies, in Proc. IEEE Vehicular Technology Conference(VTC 96) , 1996.


14/14



[3] C. W. Sung and W. S. Wong, User speed estimation anddynamic channel allocation in hierarchical cellular system, in

Proc. IEEE Vehicular Technology Conference (VTC 94) , pp.91 95, 1994.

[4] K. L. Yeung and S. Nanda, Optimal mobile -determined

micro macro cell selection, in Proc. IEEE VehicularTechnology Conference (VTC 95) , 1995.

[5] A. S. Anpalagan and L. Katzela, Overlaid cellular systemdesign with cell selection criteria for mobile wireless users, in

IEEE Canadian Conference on Electrical and Computer Engineering , pp. 24 28, 1999.

[6] M. Benveniste, Cell selection in two -tiermicrocellular/macrocellular systems, in GlobeCom 95 , pp.1532 1536, 1995.

[7] I. Chih-Lin , L. J. Greenstein, and R. D. Gitlin, Amicrocell/macrocell cellular architecture for low- and high-

mobility wireless users, in IEEE J. on Selected Areas inCommunication -11 , pp.885 891, 1993.

[8] D. Kim, B. W. Lim, and D. G. Jeong, An efficient pagingscheme for overlaid microcell/ macrocell systems, in 5th IEEE

International Conference on Universal PersonalCommunications , pp. 961 964, 1996.

[9] Y. I. Kim, K. J. Lee, and Y. O. Chin, Effect of handoff areavariation on PCS system traffic, in IEEE InternationalConference on Personal Wireless Communications , pp. 134 139, 1996.

[10] K. L. Yeung and S. Nanda, Channel management in

microcell/macrocell cellular radio systems, in IEEE Trans. onVehicular Technology -45 , pp.601 612, 1996.

[11] K. L. Yeung and S. Nanda, Optimal mobile -determinedmicro- macro cell selection, in IEEE Vehicular TechnologyConference (VTC 95) , pp. 294 299, 1995.

[12] B. Jabbari and W. F. Fuhrmann, Teletraffic Modeling andAnalysis of Flexible Hierarchical Cellular Networks withSpeed- Sensitive Handoff Strategy, in IEEE J. on Selected

Areas in Communication -15(8) , pp.1539 1548, 1997.

[13] S. Marano, C. Mastroianni, and R. Riccardi, Performanceof Micro-Macrocellular System with Overlapping Converageand Channel Rearrangement Techn iques, in Computer andCommunication, pp. 705 710, 1998.

[14] Vikas Solanki et.at., Improving the Performance ofHandoff Calls using Frequency Sharing , IJMNCT, Vol. 2, No.4, August 2012

[15] M. A. Farahani and M. Guizani, Markov ModulatedPoisson Process Model for Hand- off Calls in Cellular Systems,

in IEEE Wireless Communications and Networking Conference(WCNC) , 2000.

[16] B. Jabbari, Teletraffic aspects of evolving and nextgeneration wireless communication networks, in IEEE

Personal Communications , pp. 4 9, 1996.

[17] F. Vanhaverbeke, M. Moeneclaey, and H. Sari, Increasingcdma capacity using multiple orthogonal spreading sequencesets and successive interference cancellation, in IEEE

International Conference on Communications , 3, pp. 1516-1520, 2002.