A Channel Allocation Mechanism for Cellular Networks - MDPI

inventions Article

A Channel Allocation Mechanism for Cellular Networks Chi-Hua Chen 1,2,3, *, Bon-Yeh Lin 1 , Che-Hao Lei 4 and Chi-Chun Lo 5 1 2 3 4 5

*

Chunghwa Telecom Laboratories, Chunghwa Telecom Co., Ltd., Taoyuan 326, Taiwan; [email protected] Department of Electrical and Computer Engineering, National Chiao Tung University, Hsinchu 300, Taiwan Department of Industrial Engineering and Engineering Management, National Tsing Hua University, Hsinchu 300, Taiwan Data Communication Business Group, Chunghwa Telecom Co., Ltd., Taipei 100, Taiwan; [email protected] Department of Information Management and Finance, National Chiao Tung University, Hsinchu 300, Taiwan; [email protected] Correspondence: [email protected]; Tel.: +886-3-424-4091

Academic Editors: Giovanni Acampora and Chien-Hung Liu Received: 17 February 2017; Accepted: 12 April 2017; Published: 17 April 2017

Abstract: In cellular networks, call blocking causes lower customer satisfaction and economic loss. Therefore, the channel allocation for call block avoidance is an important issue. This study proposes a mechanism that considers the real-time traffic information (e.g., traffic flow and vehicle speed) and the user behavior (e.g., call inter-arrival time and call holding time) to analyze the adaptable number of communication calls in the specific cell for channel allocation. In experiments about call block probabilities (CBP), this study simulated two cases that are the situations of the whole day and traffic accident. The simulation results show that all CBPs in the scenario of whole day are less than 21.5% by using the proposed mechanism, which is better than using the static channel allocation (SCA) mechanism. Moreover, all CBPs in the scenario of traffic accidents are less than 16.5% by using the proposed mechanism, which is better than using the SCA mechanism. Therefore, the proposed mechanism can decrease the number of CBPs effectively. Keywords: cellular network; call drop; call block; channel allocation

1. Introduction In recent years, the rise in economic growth and information technology advancement has improved the quality of personal communication systems (PCS). Moreover, people now pay more attention to the quality of service in cellular systems. As the number of mobile stations (MS) rises, how to provide the high quality of service in cellular systems has become a big challenge. The cellular system is composed of cells. A cell is the specific coverage of the base station (BS), and it overlaps the coverage of neighboring cells. BS are fixed and interconnected through a fixed network. The structure of the cellular network is the connective coverage of the cell. Moreover, every cell only has limited channels to provide communication services. Among cellular systems, Global System for Mobile Communications (GSM) and Universal Mobile Telecommunications System (UMTS) are well known[1–4]. They can decide whether to accept the call or not depending on the available number of the channels. When an MS wants to communicate with another user, it must first obtain a free channel provided by BS. If the BS has no free channel, the call will fail. Therefore, the number of channels for Handover In BSs that are reserved is an important issue for cellular systems. The number channels provided by BS is limited, and almost everyone has at least one cellphone at the moment, so the channels of BS are usually fully connected.

Inventions 2017, 2, 8; doi:10.3390/inventions2020008

www.mdpi.com/journal/inventions

Inventions 2017, 2, 8

2 of 19

When the channels of BS are fully occupied, the MSs will encounter the situation of call drop or call block [5]. If the MS wants to communicate with another MS, it must request a free channel from the BS to generate a new call. If a channel is free, it is assigned to the MS. If the BS has no free channel to allocate to the MS, the BS will drop the new call request from MS. If the communicating call moves from the old cell to the new cell, it has to request from the new cell a free channel. If the new cell has no free channel, the call block occurs. The call block causes more loss than the call drop because it stops the ongoing call. This study solves the problem through reserving channels for Handover In BS, but how many channels of the BS to reserve is an important issue. Therefore, there is much research that is working on the quality of cellular telephone system improvement. A number of studies have investigated the method to reserve the number of channels. These can be divided into two methods: one method is a static algorithm for channel allocation and another method is a dynamic algorithm [4,6–13]. However, the static allocation algorithms are unsuitable in dynamic environments. This paper will provide a dynamic channel allocation (DCA) mechanism. This study considers the traffic information to analyze the communication behavior and proposes a method to design the DCA mechanism in personal communication systems. This study uses the communication behavior of MSs and the MS movement to estimate the status of channel use in cells. Moreover, this study uses the above information to provide a method that decides how many channels have to be reserved in the cellular telephone system. Therefore, this study uses the proposed mechanism to dynamically adjust the number of channels reserved in each BS according to the variation of traffic information. Simulation results show that the call block probability (CBP) by using the DCA mechanism is lower than by the static channel allocation (SCA) mechanism. In cellular systems, the CBP is an important factor for measuring the quality of service [14,15]. Therefore, this study will consider the factor of CBP and prove that the proposed mechanism is better than the SCA mechanism. 2. Related Work In this section, this study will introduce some concepts of PCS. In order to design a dynamic channel mechanism, the knowledge of PCS (e.g., GSM and UMTS) and the channel allocation mechanism in PCS is described in the following subsections. 2.1. Cellular Network Architecture The rapid growth in the demand for PCS has led the industry to intense research and development efforts towards a new generation of cellular systems. Nowadays, people can’t live without cellular systems. The MS can perform the communication via the cellular system when the specific cell has a free channel. In this part, we will introduce two major systems in present cellular systems. The architectures of GSM and UMTS are described in the following subsections. 2.1.1. Global System for Mobile Communications The GSM is a digital wireless network specified by standardization committees from major European telecommunications operators and manufacturers. Through the standard, all mobile users worldwide can use a common set of compatible services. Figure 1 illustrates the architecture of GSM [16–18]. The network system of GSM at least includes three sections: an MS (Figure 1a), a Base Station Subsystem (BSS, Figure 1b), and a Network and Switch Subsystem (NSS, Figure 1c). Moreover, it has to build the standard communication interface between any two components in GSM to transmit information and control command. Through the standard interface, the components in GSM can communicate with each other and complete the capability of communication. BSS consists of the Base Transceiver Station (BTS, Figure 1d) and the Base Station Controller (BSC, Figure 1e). NSS consists of the Mobile Switching Center (MSC, Figure 1f), Visitor Location Register (VLR, Figure 1g) and Home Location Register (HLR, Figure 1h). The work of BTS is listening to the order from BSC to communicate with the MS via the radio interface, while the BSC communicates with the MSC via the A interface

Inventions 2017, 2, 8

3 of 19

Inventions 2017, 2, 8

3 of 20

BSCMSC communicates with the MSC the A interface (Figure of 1i).circuit MSC is a special that 2017, 3 of 20 for (FigureInventions 1i). is2,a8 special switch thatvia executes the capability switch andswitch is responsible executes the capability of circuit switch and is responsible for recording to the billing system. HLR recording to the billing system. HLR is responsible for recoding the information of user subscribed BSC communicates with the via theofAuser interface (Figure MSCincluding is a special is responsible for recoding theMSC information subscribed our1i). system, whatswitch servicethat our system, including what service user subscribed and theforlocation ofthe MS. VLR issystem. responsible executes the capability circuit switch recording the billing HLR for user subscribed and theoflocation of MS. and VLRisisresponsible responsiblefor recording to information of user recording the information of user inside its domain, including the state and location of user. its domain, including the the information state and location of user. isinside responsible for recoding of user subscribed our system, including what service

user subscribed and the location of MS. VLR is responsible for recording the information of user inside its domain, including the state location of user. b Base Stationand System

b

MS

BSC

BSC

MS

BTS

e

a

d

BSC

d

i

e

BTS

a

c

Base Station System

i

c

A

i

Network and Switching Subsystem

Network VLR and Switching HLR Subsystem

g

MSCg VLR

A

A

i

h

HLR

h

f

MSC

MS

BTS

MS

Figure 1. The GSM network architecture [18].

A

f

BSC

Figure 1. The GSM network architecture [18].

BTS

2.1.2. Universal Mobile Telecommunications System

2.1.2. Universal Mobile Telecommunications System Figure 1. The GSM network architecture [18].

The UMTS is the third generation (3G) and is the standard of mobile communication system. UMTS hasisa the feature which is compatible with existing Radio Service (GPRS), system. The UMTS third generation (3G) and is theGSM/General standard ofPacket mobile communication 2.1.2. Universal Mobile Telecommunications System so it is considered the best cellular system in mobile communication systems. Figure 2 illustrates the UMTS has a feature which is compatible with existing GSM/General Packet Radio Service (GPRS), architecture of the UMTS domain [16–18]. The network system of The UMTS is the thirdCircuit-Switched generation (3G)(CS) and service is the standard of mobile communication system. so it is considered theUser bestEquipment cellular system in mobile systems. Figure 2 illustrates UMTShas includes (UE, Figure andcommunication UMTS Terrestrial Radio Access Network UMTS a feature which is compatible with 2a), existing GSM/General Packet Radio Service (GPRS), the architecture of the2b). UMTS Circuit-Switched (CS) service domain [16–18]. The network system Figure Moreover, it has to build the standard communication interface between any the so(UTRAN, it is considered the best cellular system in mobile communication systems. Figure 2 illustrates two components in UMTS to transmit information and control command. Through the standard, all of UMTS includes User Equipment (UE, Figure 2a), and UMTS Terrestrial Radio Access Network architecture of the UMTS Circuit-Switched (CS) service domain [16–18]. The network system of mobile users canituse common set of compatible services. URTAN consists of Node B any two (UTRAN, Figure 2b).worldwide Moreover, hasa to build the standard communication interface between UMTS includes User Equipment (UE, Figure 2a), and UMTS Terrestrial Radio Access Network (Figure 2d) and Radio Network Controllers (RNC, Figure 2e). UE is equal to user end equipment (UTRAN, Figure to 2b). Moreover, it has to build standard communication betweenall any components in UMTS transmit information and the control command. Throughinterface the standard, mobile such as MS. NodeinBUMTS is responsible for capability of physical layer,command. including controlling information two components to transmit information and control Through the standard, all 2d) users worldwide can use a common set of compatible services. URTAN consists of Node B (Figure of spread spectrum and modulation. RNC is responsible for radio resource management and the mobile users worldwide can use a common set of compatible services. URTAN consists of Node B and Radio Network Controllers (RNC, Figure of2e). UE is equal togeneral, user end such as MS. capability to connect to MSC via the interface IuCS (Figure 2a). In any equipment coverage of Node (Figure 2d) and Radio Network Controllers (RNC, Figure 2e). UE is equal to user end equipment Node B Bisisresponsible for iscapability including controlling information of spread called Cell. HLR responsibleof forphysical recoding layer, the information of users subscribed to the system, such as MS. Node B is responsible for capability of physical layer, including controlling information including what service users subscribed to and the location of MS. spectrum and modulation. RNC is responsible radio resource management and the capability to of spread spectrum and modulation. RNC isfor responsible for radio resource management and the connectcapability to MSC via the interface of IuCS (Figure 2a). In general, any coverage of Node B is called Cell. to connect to MSC via the interface of IuCS (Figure 2a). In general, any coverage of Node b UTRAN HLR isBresponsible recoding the information subscribed to the system, is called Cell.for HLR is responsible for recodingof theusers information of users subscribed to including the system,what including what service users subscribed to and the location of MS. service users subscribed to and the location of MS. c Network and Switching Subsystem

MS(UE) a

MS(UE) a MS(UE)

b UTRAN

Node B d

Node B d B Node

RNC e

RNC e RNC

i

IuCS

ii

IuCS IuCS

i

VLR

c

HLR

g

h

Network and Switching Subsystem

MSC VLR

f g

HLR

h

MSC

IuCS Figure 2. The UMTS network architecture [18].

f

RNC MS(UE)

Node B Figure 2. The UMTS network architecture [18].

Figure 2. The UMTS network architecture [18].

2.2. Channel Allocation Mechanisms In cellular networks, when one MS wants to communicate with another MS, it must get a free channel provided from BS to serve it. Thus, if the BS has a free channel, the connection between two

Inventions 2017, 2, 8

4 of 19

MSs will be built successfully; otherwise, the connection of communication fails. Therefore, to allocate the adaptable number of channels for communication is a critical issue. In this subsection, we will introduce two types of channel allocation mechanisms, which are the traditional SCA mechanism and the DCA mechanism, respectively. 2.2.1. Static Channel Allocation The mechanism of channel allocation in PCS adopts the Static Channel Allocation (SCA) mechanism; when the channels which are allocated, the BS can only use these channels to provide the service of communication. In this case, there is a definite relationship between the channels and the cells that can be used at any time. The obvious disadvantage of using the SCA mechanism can be explained using an example. Imagine two adjacent cells with their allocated channels. At any time, one of the cells may have all of its channels occupied, and another new call requests a free channel in this cell. This call will be blocked even if the adjacent cell has a free channel at this moment. The situation will result in the customer satisfaction and economic loss. Channel reuse is an essential feature of cellular system. However, the discussion of directional cell sites explained the economic incentive for minimizing the ratio of σ, and the parameter of σ can be expressed as a ratio of the physical distance between the center of cells (D) divided by the radius (R) of cells. The co-channel reuse ratio has an impact on both the communication quality and the ender customer capacity of the system [19]. Because the co-channel ratio actually has impact on the communication quality, the ratio will decide the number of channels in each channel sets. The ratio also can limit the communication capacity of cells by deciding the number of channels in each channel set. In the SCA mechanism, when the number of channel (N) is assigned to each cell, the number of channels (N) is a fixed number and permanently allocated to each cells [20]. When the number of channel is decided, the BS only can use these channels to provide service of communication. In general, the number of channels (N) in each cell can be expressed as Formula (1) [6,10,21]:  N=

1 3

 σ2 .

(1)

Here, we define σ as D/R, where D is the physical distance between the two cell centers and the R is the radius of the cell. Therefore, the number of N can only be an integer value. For example, N is 3, 4, etc. In the SCA mechanism, because the number of channels in each cell is permanent, the solution is suitable for the stable situation of traffic information. The CBP will increase with the traffic flow increases. 2.2.2. Dynamic Channel Allocation The SCA mechanism can’t achieve high efficiency of channels using the variation of traffic in the cellular system. In order to overcome this problem, the Dynamic Channel Allocation (DCA) mechanism in recent years has been studied widely. In contrast to the SCA mechanism, the DCA mechanism has no fixed relationship with each cell and channel [21]. All channels are in the central pool. When any cell has a new call arrival, it will allocate one free channel to that cell [10,22–25]. After the call is complete, the channel will be return to the central pool. In DCA, a channel can be used in any cell provided that the signal interference constraints are satisfied. Generally, more channels can be assigned to the cells’ required channel by the central pool. The main idea of the DCA mechanism is that it minimizes the cost to select and use the candidate channels provided that the interference constraints are satisfied. The selections of cost functions are designed with the different schemes. The factors when we design the function include the usage frequency of the candidate channel, the reuse distance, channel occupancy distribution under current traffic conditions, and radio channel measurements of individual MSs of the average CBP of the system [12]. In this part, the DCA mechanism can dynamically allocate free channels

Inventions 2017, 2, 8 Inventions 2017, 2, 8

5 of 19 5 of 20

[12]. Inchannel, this part, the DCA on mechanism can dynamically allocate free channels the to the to the the system cell-required depending the variation of traffic information. Therefore, DCA cell-required channel, depending on the variation of traffic information. Therefore, the DCA mechanism is suitable for dynamical environment. We design the mechanism for channel allocation mechanism is suitable for dynamical environment. We design the mechanism for channel allocation based on the DCA scheme in PCS. Finally, depending on the type of control, the DCA mechanism can based on the DCA scheme in PCS. Finally, depending on the type of control, the DCA mechanism be divided into centralized and distributed mechanisms [10]. can be divided into centralized and distributed mechanisms [10].

Centralized DCA Mechanism

Centralized DCA Mechanism

In the centralized DCA mechanism, the channels in cells are allocated by the central pool in PCS In the centralized DCA mechanism, the channels in cells are allocated by the central pool in for temporary use, and the channels return to the central pool when the use is over. The difference PCS for temporary use, and the channels return to the central pool when the use is over. The between these mechanisms is the specific cost functions adopted for selection the candidate channel difference between these mechanisms is the specific cost functions adopted for selection the forcandidate allocation. Generally, the factors of cost functions are Availableare (FA) and Locally (FA) Optimized channel for allocation. Generally, the factors of First cost functions First Available and Dynamic Assignment (LODA). In the centralized DCA mechanism, the FA is the simplest Locally Optimized Dynamic Assignment (LODA). In the centralized DCA mechanism, the FAstrategy is the to simplest select thestrategy candidate channel for use. Inchannel FA, thefor first available channel within channel the reuse distance to select the candidate use. In FA, the first available within the σ selected during the channel search is assigned to the cell. In LODA, the selected cost function is based reuse distance σ selected during the channel search is assigned to the cell. In LODA, the selected oncost the function future blocking adjacent toprobability the cell where a calltoisthe initiated. is basedprobability on the future blocking adjacent cell where a call is initiated. Distributed DCA Mechanism Distributed DCA Mechanism Due toto economic quality of of service serviceininPCS, PCS,DCA DCAmechanisms mechanisms Due economicgrowth growthand andthe the demand demand of the quality have been studied in recent years. Several simulation results and analyses have shown that have been studied in recent years. Several simulation results and analyses have shown that thethe centralized channelallocation, allocation,but butit itwill will cause centralizedDCA DCAmechanism mechanismcan can provide provide near near optimum optimum channel cause thethe immense management costs in central the central control. Therefore, distributed DCA mechanism is immense management costs in the control. Therefore, distributed DCA mechanism is proposed. proposed. The difference between the centralized DCA mechanism and the distributed DCA The difference between the centralized DCA mechanism and the distributed DCA mechanism is mechanism is thatDCA the mechanism distributed DCA mechanism doesn’tpool havefor the centralcontrol. pool for channel that the distributed doesn’t have the central channel The number The numberand of available channels and signal strength measurements are considered and of control. available channels signal strength measurements are considered and used by the proposed used by the proposed distributed DCA mechanism [10]. distributed DCA mechanism [10]. 3. Channel AllocationMechanism Mechanismin inCellular Cellular Networks Networks 3. Channel Allocation this section,this thisstudy studywill willanalyze analyze the the relation relation of ofof MSs and thethe In In this section, of communication communicationbehavior behavior MSs and status of traffic. Moreover, this study uses this traffic information to provide a method that can help status of traffic. Moreover, this study uses this traffic information to provide a method that can help cellular telephonesystem systemtotodecide decidethe the number number of 3 shows thethe cellular telephone of channels channelsthat thatcan canbebereserved. reserved.Figure Figure 3 shows the architecture of our proposed mechanism. The goal in the proposed mechanism is to get the the architecture of our proposed mechanism. The goal in the proposed mechanism is to get the number number of communicating calls (Ci) that is derived from four factors. The factors include the number of communicating calls (Ci ) that is derived from four factors. The factors include the number of Call of Call Arrival (Ai), the number of Handover In (Ii), the number of Handover Out (Oi), and the Arrival (Ai ), the number of Handover In (Ii ), the number of Handover Out (Oi ), and the number of number of Call Departure (Di). Moreover, this study uses Formula (2) to estimate the number of Call Departure (Di ). Moreover, this study uses Formula (2) to estimate the number of communicating communicating call (Ci) for Celli. In the following, this study will discuss the four factors, call (Ci ) for Celli . In the following, this study will discuss the four factors, respectively: respectively:

C Cii == AAii ++ IIi i−−OOi i −− DDii.

.

Car Flow

The number of Call Arrival(Ai)

The number of Handover In(Ii)

The number of Handover Out(Oi)

The number of Call Departure(Di)

The number of communicating calls(Ci)

Figure proposedmechanism. mechanism. Figure3.3.The The architecture architecture of proposed

(2) (2)

Inventions 2017, 2, 8

6 of 19

Inventions Inventions 2017, 2017, 2, 2, 88

66 of of 20 20

3.1. The Number of Call Arrival (Ai )

3.1. 3.1. The The Number Number of of Call Call Arrival Arrival (A (Aii))

An MS (i.e., (a) in Figure 4) can perform a call in the service area (i.e., cell), which is provided by An MS perform aa call service area (i.e., cell), which is provided by or Node B in MSBcan can perform networks call in in the the service area cell), (i.e., which provided byifBSs BSs Node BSs or An Node in cellular (e.g., GSM and(i.e., UMTS) (b)isin Figure 4) theorcell hasBainfree cellular networks (e.g., GSM and UMTS) if the cell has a free channel. The scenario and timing cellularThe networks (e.g., and UMTS) if has processes a free channel. scenario andcovered timing by channel. scenario andGSM timing diagrams of the callcell arrival on theThe road segments diagrams of call arrival processes on the road segments covered by the specific Cell i is showed in diagrams of call arrival processes on the road segments covered by the specific Cell i is showed in the specific Celli is showed in Figures 4 and 5. The first call set-up at time t0 is performed by a MS (in Figures 4 and 5. first set-up time tt00 is by (in Figure 5), and MS keeps Figures andthe 5. The The first call call set-up at at is performed performed by aa MS MS Figure and the the Figure 5), 4and MS keeps moving to time the specific cell coverage at (in time t1 (in5), Figure 5).MS Thekeeps second moving to the specific cell coverage at time t 1 (in Figure 5). The second call set-up at time t2 is moving to the specific cell coverage at time t 1 (in Figure 5). The second call set-up at time t2 is call set-up at time t2 is (in performed by the MS (in Figure 5) beforecoverage leaving thetime specific cell coverage at performed performed by by the the MS MS (in Figure Figure 5) 5) before before leaving leaving the the specific specific cell cell coverage at at time tt33 (in (in Figure Figure 5). 5). time t3This (in Figure 5). This study study assumes assumes that that the the call call inter-arrival inter-arrival time time (t) (t) is is exponentially exponentially distributed distributed with with the the mean mean This study assumes that the call inter-arrival time (t) is of exponentially distributed with the mean 1/λ 1/λ [26–33] [26–33] to to generate generate the the Call Call Arrival. Arrival. Because Because the the distance distance of the the cell cell coverage coverage is is llii and and the the speed speed of of 1/λ [26–33] to generate the Call Arrival. Because the distance of the cell coverage is l and the speed i the the car car is is V Vii,, the the time time that that the the car car moves moves from from one one side side to to the the other other side side is is llii/V /Vii.. This This approach approach ofconsiders the car is Vi , MS the timehas that the car moves from one side to the other time side is li /Vi . the This approach considers the the MS that that has twice twice the the number number of of Call Call Arrivals, Arrivals, and and the the time entering entering the cell cell and and considers the MS that has twicethe theprobability number of of Call Arrivals, and thestudy time entering the cell and leaving leaving cell to estimate Call Arrival. This gets the number leaving the cell to estimate the probability of Call Arrival. This study gets the number of of Call Call the cell to estimate theroad probability Call Arrival. gets the number of Call Arrivals Arrivals (A segments traffic flow (F by of Arrivals (Aii)) on on the the road segmentsofthrough through traffic This flowstudy (Fii)) multiplied multiplied by the the probability probability of Call Call(Ai ) It can be expressed as Formula (3) [27–33]: onArrival. the road segments through traffic flow (F ) multiplied by the probability of Call Arrival. It can be i Arrival. It can be expressed as Formula (3) [27–33]: expressed as Formula (3) [27–33]: λl i − λl i −

l x + lii ∞  lli  1 − e VVii λli (3) x +V l − λt ∞  i λe − λt dtdx = F × 1 − e i  dx = F × Z  Z Z A F x t x Pr = × < < + i V i Aii = Fii ×∞∫∫x = 0 Pr x < t < x + V dtλtdx = Fii × λ1 − e− Vi (3) li dx = Fii × ∫∫xx==∞00 ∫∫tt==xxx+λVei − xPr = 0 x < t < x + V i dx = Fi × λe dtdx = Fi × λ Ai = Fi × . (3)  Vi i  λ .. x =0 t = x x =0 ∞ ∞

bb

BSC/RNC BSC/RNC

Cell Cellii FFi i

Road Road

aa enter Cell at Call Call Arrival Arrival enter tCell at t11 at at tt0 0

Call Call Arrival Arrival leave leave Cell Cell at at at at tt22 tt3 3

Figure The scenario diagram for vehicle movement and call arrivals on the road segment Figure 4.The Thescenario scenariodiagram diagramfor forvehicle vehiclemovement movementand andcall callarrivals arrivals on on the the road road segment segment covered covered by Figure 4.4. covered by the specific Cell i [27–33]. the specific i [27–33]. thebyspecific CelliCell [27–33].

nd Call Arrival 22nd Call Arrival

11stst Call Call Arrival Arrival tt enter enter Cell Cellii xx tt00

leave leave Cell Cellii llii/V /Vii

tt11

tt22

tt33

Figure 5. The timing diagram diagram forcall call arrival on the road segment coveredthe by the specific Celli [27–33]. Figure Figure 5. 5. The The timing timing diagram for for call arrival arrival on on the the road road segment segment covered covered by by the specific specific Cell Cellii [27–33]. [27–33].

Inventions 2017, 2, 2, 88 Inventions 2017, Inventions 2017, 2, 8

7 of 2019 7 of 7 of 20

3.2. The Number of Handover In (Ii) 3.2.The TheNumber NumberofofHandover HandoverInIn(I(Ii )i) 3.2. When a communicating MS in the car moves from the source cell to the target cell, the handover When communicating MS(i.e., in the moves from thecar source cell target cell, When aacommunicating (a) car in Figure in the moves from the source cellthe to the target procedure can be performed ifMS the target cell has a 6) free channel. There istoathe road covered by ahandover set of procedure can be performed if the target cell has a free channel. There is a road covered by a set cell, the handover procedure can be performed if the target cell has a free channel. There is a road cells and a communicating MS in a car on the road. For instance, Figures 6 and 7 illustrate the of cells and communicating MS in a car on road. For 6 andat 7time illustrate covered byaatiming set of diagrams cells and afor communicating MS specific in a car oninstance, the instance, Figures andathe 7 scenario and Handover In the the Cell i. Aroad. call Figures isFor performed t06 by scenario and timing diagrams for Handover In the specific Cell i . A call is performed at time t0 byata illustrate the7). scenario and timing diagrams Handover the Celli . 7) A and call aisfree performed MS (in Figure Furthermore, the MS moves for from Celli−1 to In Cell i atspecific t1 (in Figure channel MS (in Figure 7). Furthermore, the MSofmoves from CellBSC i−1 toor Cell i at t1 to (inCell Figure 7) and a free channel t0 by a MS (inhandover Figure 7).procedure Furthermore, theMS MS from Cell istime allocated for the the bymoves the RNC. i −1 i at t1 (in Figure 7) and a is allocated for the handover procedure of the MS by the BSC or RNC. free channel is allocated for the handover procedure of the MS by the BSC or RNC (i.e., (b) in Figure 6).

b

BSC/RNC BSC/RNC

b

Handover Zone Handover Zone Celli-1 Celli-1

a

Fi

Celli Celli Fi

a

Road Road

Cell at Call Arrival enter enter Cell at t Call Arrival 1 at t0 t 1 at t0

leave Cell at t2 leave Cell at t2

Figure 6.6. The scenario diagram forfor vehicle movement and Handover i [27–33]. Figure The scenario diagram vehicle movement and HandoverInInthe thespecific specificCell Cell i [27–33]. Figure 6. The scenario diagram for vehicle movement and Handover In the specific Cell i [27–33].

Call Arrival Call Arrival

Call Departure Call Departure

τ enter Celli enter Celli x t0

leave Celli

li/Vi li/Vi

x t1

t0

τ leave Cell i

t2

t1

t2

t3

t3

Figure The timing diagram Handover the specificCell Cell Figure 7.7.The timing diagram forfor Handover InIn the specific i [27–33]. i [27–33]. Figure 7. The timing diagram for Handover In the specific Celli [27–33].

For ),i ),the Forestimating estimatingthe thenumber numberofofHandover HandoverInIn(Ii(I thecall callholding holdingtime time(τ)(τ)isisassumed assumedtotobebe For estimating the number of Handover In (I i), the call holding time (τ) is assumed to be exponentially distributed with thethe mean 1/μ1/µ [26–33]. In this study, the car and vehicle speedspeed are exponentially distributed with mean [26–33]. In this study, theflow car flow and vehicle exponentially distributed withthe thedistance mean 1/μ [26–33]. In this study, the car time flow difference and vehiclebetween speed are assumed as F i and V i . Moreover, of the cell coverage is l i , so the are assumed as Fi and Vi . Moreover, the distance of the cell coverage is li , so the time difference assumed asi F(i.e., i and Vi. Moreover, the distance of the cell coverage is li, so the time difference between entering t1) and leaving i (i.e.,Cell t2) i is(i.e., measured as li/Vi. as The variable x is the time betweenCell entering Cell (i.e., t1 ) andCell leaving t2 ) is measured li /V i . The variable x is the enteringbetween Celli (i.e., t1i)arrival and leaving Cell i (i.e., t2) is measured as li/Vi. The variable x is the time difference call (i.e., t 0 ) and entering Cell i (i.e., t 1 ). When the call holding timetime (τ) is(τ) time difference between call arrival (i.e., t0 ) and entering Celli (i.e., t1 ). When the call holding difference between callx,arrival (i.e., t0) and enteringwill Cellbe i (i.e., t1). When the call holding time (τ) is longer than the variable the handover procedure performed. Therefore, the number is longer than the variable x, the handover procedure will be performed. Therefore, the numberofof longer than the variable x, the handover procedure will be performed. Therefore, the number of Handover ) of Cell i can be expressed asas Formula (4): HandoverInIn(Ii(I ) of Cell can be expressed Formula (4): Handover In (Ii i) of Celli i can be expressed as Formula (4): Z∞∞ Z∞ ∞ ∞Z ∞ F − µτ ( ) IIii == FFi × × ∫ ∞Pr > x dx = F × dx = =i FFii. τ τdτdx Pr(τ > x )dx = Fi i ×∫ ∞∫ ∞µe µe−−dµτ µτ I i = Fi ×xx==∫0 0 Pr (τ > x )dx = Fi ×x =x∫0=0τ =∫xτ =x µe dτ dx =µ µ. x =0

x =0 τ = x

µ.

(4) (4) (4)

Inventions 2017, 2, 8

8 of 20

3.3. The Number of Handover Out (Oi) For estimating the number of Handover Out ( Oi ), two cases including (1) Call Arrival and (2) 19 Handover In are discussed and analyzed. For instance, when a communicating MS in the car8 of leaves Celli, the procedure of Handover Out can be performed in Figures 7 and 8. For the case of Call Arrival, call is performed MS in Celli at t2 and the communicating MS leaves Celli at t3 (in 3.3. Thea Number of Handover by Outa(O i) Inventions 2017, 2, 8

A

can be estimated (5) in Figure For 8). estimating Therefore, the thenumber numberofofHandover Handover Out ( Otwo i ) cases Out (Oi ), including (1) by CallFormula Arrival and (in instance, Figure 8). For athe case of Handover In, car a handover accordance with probability ofanalyzed. (τ > li/Vi)For (2) Handover In the are discussed and when communicating MS in the leaves Celli , the Handover Out can be performed in iFigures 7 and 8. ForMS theenters case of Cell Call i,Arrival, process is procedure performedof by a communicating MS in Cell at t1 when the and the I a call is performed by a MS in Cell at t and the communicating MS leaves Cell at t (in Figure 2 3 i i can communicating MS leaves Celli at t2 (in Figure 7). Therefore, the number of Handover Out ( Oi )8). A Therefore, the number of Handover Out (Oi ) can be estimated by Formula (5) in accordance with the be estimated by Formula (6) in accordance with the probability of (τ > x + li/Vi) (in Figure 7). probability of (τ > li /Vi ) (in Figure 8). For the case of Handover In, a handover process is performed is expressed as Formula (7): Therefore, the number MS of Handover ( Oi )the by a communicating in Celli at Out t1 when MS enters Cell i , and the communicating MS leaves Celli at t2 (in Figure 7). Therefore, the number of Handover Out (OiI ) can be estimated byli Formula (6) −µ Vi in accordanceAwith the probability (τ > x + li /Vi ) (inli Figure 7). −Therefore, the number of Handover ∞ ∞  lof e − 1 µτ i (5) µe dτ dx = Fi × Oi = Fi ×as Formula Prτ > (7): − x dx = Fi × Vi li Out (Oi ) is expressed x =0 τ = − x x =0 µ Vi Vi   ,   li Z ∞ Z li Z ∞ − µ Vi 1 − e li li Vi A −µτ Oi = Fi × Pr τ > , (5) − x dx = Fi × µe dτdx = Fi × −µ l x =0 ∞ x =0∞ τ =∞Vi − x  Vi li  e µVi i I − µτ (6) Pr τ > x + dx = Fi × Oi = Fi × dτ dx = Fi × l µe x =0 τ = x + i x =0 µ Vi  li Z ∞ Z ∞ Z ∞ Vi l e−µ Vi , OiI = Fi × Pr τ > x + i dx = Fi × µe−µτ dτdx = Fi × , (6) li Vi µ x =0 x =0 τ = x + V

∫ ∫

∫

∫

∫ ∫

i 1 = Ii 1 Oi = OiA + OiI = Fi × µ = Ii ..

Oi = OiA + OiI = Fi ×

(7) (7)

µ

Call Departure

Call Arrival τ enter Celli

leave Celli li/vi

x t1

t2

t3

t4

Figure The timingdiagram diagramfor forHandover Handover Out Out the Arrival [27–33]. Figure 8. 8. The timing the specific specificCell Celli iderived derivedfrom fromCall Call Arrival [27–33].

3.4. The NumberofofCall CallDepartures Departures(D (Di)i ) 3.4. The Number Forestimating estimating the the number Call Departure (Di ), (two (1) Call Arrival (in Figures For numberofof Call Departure twoincluding cases including (1) Call Arrival 9(in Di ),cases and 10); and (2) Handover In (in Figures 11 and 12) are discussed and analyzed. For the case of Call Figures 9 and 10); and (2) Handover In (in Figures 11 and 12) are discussed and analyzed. For the Arrival, a call is performed by a MS in Celli at t1 and the call departure of the MS is performed in Celli case of Call Arrival, a call is performed by a MS in Celli at t1 and the call departure of the MS is at t2 (in Figures 9 and 10). In Case 1, the number of Call Departure is expressed as DiA . For the case of performed in Celli at t2 (in Figures 9 and 10). In Case 1, the number of Call Departure is expressed as Handover In, a handover process is performed by a communicating MS in Celli at t1 when the MS A Denters thei , case of Handover In, of a handover process isin performed by Figures a communicating MS in Cell and the call departure the MS is performed Celli at t2 (in 11 and 12). In Case 2, i i . ForCell I number Departure is expressed as Di . Therefore, the of Call Departure the of MSCall enters Celli, and the call departure of the MS is number performed in Cell i at t2 (in(D Figures at the t1 when i ) is I as Formula (8). The scenario is called Call Departure derived from Handover In for Celli . the number of 11expressed and 12). In Case 2, the number of Call Departure is expressed as Di . Therefore, Furthermore, Oi can be estimated by Formula (10) in accordance with Formula (7) and Formula (9): Call Departure ( Di ) is expressed as Formula (8). The scenario is called Call Departure derived from Di =be DiAestimated + DiI , by Formula (10) in accordance (8) with Handover In for Celli. Furthermore, Oi can Formula (7) and Formula (9): Ai + Ii = Di + Oi , (9)

Di = DiA + DiI ,− λlVi

Ai = Di = Fi ×

1−e λ

i

.

(8) (10)

A + iI i = Di i + Oi ,i , − Vi 1− e Ai = Di = Fi × λlλi l −λ − i 11−−ee ViVi . AiA=i =DD i i==FF i × i × λλ . .

(9) (9)

i i

Inventions 2017, 2, 8

b bb

(10)

(10) (10) 9 of 19

BSC/RNC BSC/RNC BSC/RNC

Cell Cell Cell

Road

a

Road Road

aa Call Arrival Call Departure leave Cell at t3 enter Cell at t0 at t 1 Call Arrival at t2 Call Arrival CallDeparture Departure leaveCell Cellatatt3t3 Call enter Cell leave enter Cell at at t0 t0 at t at t1 1 at at t2t2

Figure 9. The scenario diagram for vehicle movement and call departure in the specific Celli derived Figure scenario diagram vehicle movementand andcall calldeparture departurein thespecific specificCell Cell iderived derived from call arrival [27–33]. Figure 9. 9. The scenario diagram forfor vehicle ininthe the specific Cell Figure 9.The The scenario diagram for vehiclemovement movement and call departure i iderived from call arrival [27–33]. from call arrival [27–33]. from call arrival [27–33].

enter Celli enterCell Cell enter i i

leave Celli leaveCell Celli i leave

li/Vi l i l /V Call Arrivali /V i iCall Departure Call Arrival CallDeparture Departure Call Arrival Call τ ττ x xx t0 t0 t0

t1 t1t1

t2t2

t2

t3t3

t3

Figure 10.10. The departureininthe thespecific specificCell Cell i derived from arrival [27–33]. Figure Thetiming timingdiagram diagram for for call call departure i derived from callcall arrival [27–33]. Figure 10.10. The timing diagram forfor call departure Figure The timing diagram call departureininthe thespecific specificCell Celli iderived derivedfrom from call call arrival arrival [27–33]. [27–33].

b b b Handover Handover Handover Zone Zone Zone

Celli-1 Cell Cell i-1i-1

Fi Fi Fi

a aa

BSC/RNC BSC/RNC BSC/RNC

Celli Cell Cell i i Road RoadRoad

leave Cell at Call Arrivalenter Cell at t1 leave Cell at at leave t Cell Call CallArrival Arrival Cell at at t1 tCall at t0 enter enter Cell Departure t 3 1 3 t 3 atatt0t0 Call Departure Call Departure at t2 at tat 2 t2 Figure 11. The timing diagram for Call Departure in the specific Celli derived from Call Arrival [27–33]. Figure 11. timing diagram for Call Departure the specific Cell i derived from CallCall Arrival [27–33]. Figure 11. timing diagram for Call Departureininin the specific Cell from Call Arrival [27–33]. Figure 11.The TheThe timing diagram for Call Departure the specific Cell i derived from Arrival [27–33]. i derived

Inventions 2017, 2, 8 Inventions 2017, 2, 8 Inventions 2017, 2, 8

10 of 19 10 of 20 10 of 20

Call Arrival Call Arrival

τ enter Celli τ enter Celli

Call Departure Call Departure leave Celli leave Celli

li/Vi li/Vi

x x t0 t0

t1 t1

t2 t2

t3 t3

Figure12. 12.The Thetiming timingdiagram diagramfor forcall calldeparture departureininthe thespecific specificCell Celli derived fromHandover HandoverInIn[27–33]. [27–33]. Figure i derivedfrom Figure 12. The timing diagram for call departure in the specific Celli derived from Handover In [27–33].

4.4.Numerical NumericalAnalysis Analysis 4. Numerical Analysis This investigatesthe the communication behaviors of handovers and call arrivals by This section section investigates communication behaviors of handovers and call arrivals by proposed This section investigates the communication behaviors of handovers and call arrivals by proposed model through numerical analysis. The input parameters are considered as follows: model through numerical analysis. The input parameters are considered as follows: proposed model through numerical analysis. The input parameters are considered as follows: • The Thevariable variableliliisisthe thedistance distanceofofthe theroad roadsegment segmentcovered coveredby bythe thecell celli,i,  The variable l i is the distance of the road segment covered by the cell i, • the is the average speed of car, thevariable variableVVi which i which is the average speed of car,  the variable V i which is the average speed of car, • the thevariable variableFFi iwhich whichisisthe thecar carflow, flow, thecall variable Fi which is the carhas flow,  the inter-arrival time t (h) exponential distribution with mean 1/λ, • the call inter-arrival time t (h) has exponential distribution with mean 1/λ,  the call inter-arrival time t (h) has exponential distribution with mean  the call holding time τ(h) has exponential distribution with mean 1/μ. 1/λ, the • the call call holding holding time time τ(h) τ(h) has has exponential exponential distribution distribution with with mean mean 1/µ. 1/μ. The variable Ai is the count of call arrival per hour. The variable Oi is the count of Handover Out variable the ofcall call arrival per hour. variable isi count the count of Handover The variable AAi iisisthe arrival per hour. The variable Oi isOthe of Handover Out iD per hour. The variable Ii count iscount the of amount of Handover In.The The variable is the amount of call Out per hour. The variable I is the amount of Handover In. The variable D is the amount of per hour. The variable Ii study is ithe investigates amount of Handover In.traffic The variable Di is thecommunication amount of call call i and department per hour. This the effects of information department per This investigates information and communication department per hour. This study investigates the effects of traffic communication behaviors on the Ai and Oi. behaviors on A andAO O behaviors the Ai the and i.. Effectson of the Fi on i iand Oi: Figure 13 shows that the numbers of Ai and Oi are increased in Effects of F on the A and : Figure thatgenerate the numbers of increased the Ai iand OO i:are 1313 shows numbers AiAand iO are increased iFigure i and i are accordance withFiiFon i. Several MSs carried inshows Cellthat i to the the of signals ofOcall arrivals andin in accordance with F . Several MSs are carried in Cell to generate the signals of call arrivals and accordance with F i . Several MSs are carried in Cell i to generate the signals of call arrivals and i i handovers with traffic congestion. Therefore, the factor of car flow (Fi), which is important. handovers with traffic congestion. Therefore, the factor of car flow (F ), which is important. car flow (Fii

Figure 13. Effect of Fi on the Ai and Oi. Figure 13. 13. Effect Effect of of FFii on on the the A Aii and Figure and O Oii..

Effects of Vi on the Ai and Oi: Figure 14 shows that the number of Ai is decreased in accordance of V Vnumber i on the A i: Figure 14 shows that study the number of on Ai is in accordance with VEffects i, but the ofi and Oi isOnot affected by Vi. This focuses thedecreased channel allocation for Effects of i on the Ai and Oi : Figure 14 shows that the number of Ai is decreased in accordance with V i , but the number of O i is not affected by V i . This study focuses on the channel allocation for handover, so the the factor of of theOaverage speed of car (Vi) is not important. with Vi , but number i is not affected by Vi . This study focuses on the channel allocation for handover, so the factor of the average speed of car (V i) is not important. handover, so the factor of the average speed of car (V ) is not important. i

Effects of λ on the Ai and Oi : Figure 15 plots Ai and Oi against λ, which indicates that Ai and Oi are unchanged as λ increases. This phenomenon is explained as follows. When the call inter-arrival

Inventions 2017, 2, 8

11 of 19

Inventions 2017, 2, 8

11 of 20

time (λ) increases, the influence of call inter-arrival time (λ) on Ai and HOC are invalid. Though Inventions 2017,factor 2, 8 11 of 20 Figure 15, the of call inter-arrival time (λ) is not important for channel allocations.

Figure 14. Effect of Vi on the Ai and Oi.

Effects of λ on the Ai and Oi: Figure 15 plots Ai and Oi against λ, which indicates that Ai and Oi are unchanged as λ increases. This phenomenon is explained as follows. When the call inter-arrival time (λ) increases, the influence of call inter-arrival time (λ) on Ai and HOC are invalid. Though Figure14. 14.time Effect of Vii on on the Figure Effect the A Aii and and Oi.i . Figure 15, the factor of call inter-arrival (λ)of is V not important forOchannel allocations.

Effects of λ on the Ai and Oi: Figure 15 plots Ai and Oi against λ, which indicates that Ai and Oi are unchanged as λ increases. This phenomenon is explained as follows. When the call inter-arrival time (λ) increases, the influence of call inter-arrival time (λ) on Ai and HOC are invalid. Though Figure 15, the factor of call inter-arrival time (λ) is not important for channel allocations.

Figure 15. Effect of λ on the Ai and Oi.

Figure 15. Effect of λ on the Ai and Oi . Effects of μ on the Ai and Oi: Figure 16 plots Ai and Oi against μ, which indicates that Ai is unchanged Oi A increases holding phenomenon is explained Effects of µ onand the Figure 16 time plots(μ)Aincreases. against µ, which indicatesasthat Ai is i and Oas i : call i and Oi This follows. the call holding (μ) increases, the amounts of handover become higher.isInexplained the unchanged andWhen Oi increases as calltime holding time (µ) increases. This phenomenon as other words, no matter how the call holding time (μ) changes, the influence of the call holding time follows. (μ) When the call holding time (µ) increases, the amounts of handover become higher. In the on Oi is invalid. In Figure 16, the factor of the call holding time (μ) is important for channel other words, no matter how the call holding time (µ) changes, the influence of the call holding time (µ) allocations. on Oi is invalid. Figure 16, the factor the call of holding (µ)Ois allocations. InventionsIn 2017, 2, 8 12 of 20 Figureof15. Effect λ on thetime Ai and i. important for channel

Effects of μ on the Ai and Oi: Figure 16 plots Ai and Oi against μ, which indicates that Ai is unchanged and Oi increases as call holding time (μ) increases. This phenomenon is explained as follows. When the call holding time (μ) increases, the amounts of handover become higher. In the other words, no matter how the call holding time (μ) changes, the influence of the call holding time (μ) on Oi is invalid. In Figure 16, the factor of the call holding time (μ) is important for channel allocations.

Figure 16. Effect of μ on the Ai and Oi.

Figure 16. Effect of µ on the Ai and Oi . 5. Simulation Analyses Trace-driven experiments were designed and performed to analyze relationship of the traffic information and communication behaviors in cellular networks. For vehicle movement trace generation, this study simulated the highway scenario by VISSIM (V.5.20; Planung Transport

Inventions 2017, 2, 8

Figure 16. Effect of μ on the Ai and Oi.

12 of 19

5. Simulation Analyses

5. Simulation Analyses Trace-driven experiments were designed and performed to analyze relationship of the traffic information and experiments communication cellular networks. For vehicle movement trace Trace-driven werebehaviors designed in and performed to analyze relationship of the traffic generation, this study simulated the highway scenario by VISSIM (V.5.20; Planung Transport information and communication behaviors in cellular networks. For vehicle movement trace generation, Verkehr AG, Karlsruhe, assigned the position of each cell and the position of AG, the this study simulated theGermany) highway and scenario by VISSIM (V.5.20; Planung Transport Verkehr handover. The vehicle andposition MS communication are generated by a traffic Karlsruhe, Germany) andmovement assigned the of each cell andtraces the position of the handover. The simulation program VISSIM. This study considered a highway scenario that is characterized by the vehicle movement and MS communication traces are generated by a traffic simulation program VISSIM. Wiedemann “psycho-physical” car-following model and lane model [34–36]. The This study considered a highway scenario that is characterized by thechanging Wiedemann “psycho-physical” conceptual development and limited available data can be considered by the Wiedemann to car-following model and lane changing model [34–36]. The conceptual development andmodel limited generate traffic steam data for highways and freeways [27,28]. For MS communication trace available data can be considered by the Wiedemann model to generate traffic steam data for highways generation, study that the distribution of callthis holding is exponential with the and freewaysthis [27,28]. Forassumed MS communication trace generation, studytime assumed that the distribution mean 1/μ and the distribution of call inter-arrival time is exponential with the mean 1/λ [26–33]. The of call holding time is exponential with the mean 1/µ and the distribution of call inter-arrival time is communications of each MS were simulated according to the assumptions. Finally, this study exponential with the mean 1/λ [26–33]. The communications of each MS were simulated according to assumed that an Finally, MS is carried in aassumed car, so the movement and communication the assumptions. this study thatvehicle an MS is carried intraces a car, so theMS vehicle movement traces were combined in accordance with car IDs (shown in Figure 17). traces and MS communication traces were combined in accordance with car IDs (shown in Figure 17). Road Conditions and Vehicle Movement Behaviors

MS Communication Behaviors

Vehicle Movement Trace Generation

MS Communication Trace Generation

Vehicle Movement And MS Communication Trace Generation

Figure 17. The architecture of simulation [28]. Figure 17. The architecture of simulation [28].

5.1. Simulation Simulation Case Case Design Design and and Performance Performance Metrics Metrics 5.1. In experiments, study designed two cases of simulations, which are the whole daywhole simulation In experiments,this this study designed two cases of simulations, which are the day and the traffic respectively. Case 1 isCase the whole simulation. This study simulation and accident the trafficsimulation, accident simulation, respectively. 1 is theday whole day simulation. This obtained the peak and off-peak of traffic flow from this case. Through the variation of actual traffic flow, this study analyzed the use of channels and handover occurrence. Case 2 is the traffic accident simulation. This study simulated the vehicle movement and communication behavior before and after traffic accident. In Case 2, this study analyzed the surging amount of traffic flow and discussed the relation of traffic flow and the number of handovers. 5.1.1. Case 1: Whole Day In Case 1, the traffic information is obtained from the actual VDs that were built on the 42 KM milepost on National Freeway No. 1 on 31 July 2008. This study used the real traffic information to simulate the vehicle movement and MS communication behaviors. The length of a three-lane highway is 10 km was considered in experimental environments. In this case, 10 cells were sequentially distributed on the 10-km length highway from 0 km to 10 km, and a 1 km length road segment could be covered by a cell. Therefore, 11 handover points were distributed on the 0 km length highway. Furthermore, 11 Data Collection Points (DCPs) were assumed to be located at these 11 handover points. When a car passed a DCP, the passing time was recorded. It means that the coverage area of Cell1 is the area between the first DCP1 and the second DCP2 . In this case, this study obtained the traffic information for the whole day and analyzed the CBP during traffic peak hours and traffic off-peak hours.

Inventions 2017, 2, 8

13 of 19

5.1.2. Case 2: Traffic Congestion This study designed a traffic accident to simulate the MS communication behavior and handover events in the situation of traffic congestion. This study observed the number of channels used in a cellular system during traffic congestion occurrence and ease. The setting of road environments is the same as Case 1. In Case 2, the parameters of road information and communication behaviors were assumed as these parameters in Case 1. However, the traffic flow was assumed as 5000 vehicles per hour in Case 2 for the estimation of traffic congestion. The desired speed of each vehicle was assumed between 85 and 120 km/h when the status of traffic condition is free flow. The total simulation time is 1.5 h. This study simulated a traffic accident that impacts two lanes on the road, and the mileage is between 5.05 and 5.35 km from simulation time from the 15th min to the 45th min. When vehicles go through this road segment, the average speed of car is dropped to 4–6 km/h. Until simulation time to the 45th min, the traffic accident is removed and the traffic conditions returns to a free flow. 5.1.3. Performance Metrics In a cellular system, the CBP (Pi ) in Celli is an important performance metric for measuring the quality of service. The CBP is denoted as Formula (11). The number of call block (Bi ) divided by the number of the Handover In (Ii ) in Celli . The probability represents the fraction of the number of handovers that are blocked: B (11) Pi = i . Ii 5.2. Simulation Results and Analyses In this section, this study introduced the simulation results and analysis. In the simulations, this study analyzed two cases that are for the whole day simulation and traffic accident simulations. Therefore, this study separately analyzed the results derived from both cases. 5.2.1. Case 1: Whole Day In the whole day simulation, this study simulated the traffic information (e.g., traffic flow and vehicle speed) in Cell6 in the whole day shown as Figure 18. In experiments, this study analyzed the relation of traffic flow and the number of Handover In (Ii ) in each cell. This study observed the CBP (P6 ) in Cell6 during the traffic peak hour (i.e., 8:00 a.m.) and the traffic off-peak hour (i.e., 4:00 a.m.). In Figure 18, this study can discover that the minimum number of traffic flow is off-peak at 4:00 a.m. because the time is deep into the light in Taiwan. Figure 19 shows the number of Handover In (I6 ) in Cell6 for the whole day. In this situation, the number of Handover In (I6 ) is lower with the lower traffic flow. When the time is going into the morning, this study can discover that the maximum amount of traffic flow at 8:00 a.m. reaches the peak of the whole day. The number of Handover In (I6 ) is higher with the higher traffic flow during the traffic peak hour. These simulation results can prove that the number of handovers increase with the amount of traffic flow increase in Case 1. According to the simulation results, the proposed mechanism can dynamically allocate the channels to use the source of channels effectively.

traffic traffic flow. flow. When When the the time time is is going going into into the the morning, morning, this this study study can can discover discover that that the the maximum maximum amount of traffic flow at 8:00 a.m. reaches the peak of the whole day. The number of Handover amount of traffic flow at 8:00 a.m. reaches the peak of the whole day. The number of Handover In In (I (I66)) is higher with the higher traffic flow during the traffic peak hour. These simulation results can prove is higher with the higher traffic flow during the traffic peak hour. These simulation results can prove that that the the number number of of handovers handovers increase increase with with the the amount amount of of traffic traffic flow flow increase increase in in Case Case 1. 1. According According to the simulation results, the proposed mechanism can dynamically allocate the channels the to the simulation results, the proposed mechanism can dynamically allocate the channels to to use use the Inventions 2017, 2, 8 14 of 19 source of channels effectively. source of channels effectively.

Traffic Flow Flow(F (Fi)(car/hr) i)(car/hr) Traffic

7000 7000 6000 6000 5000 5000 4000 4000 3000 3000 2000 2000 1000 1000 00

00 11 22 33 44 55 66 77 88 99 10 10 11 11 12 12 13 13 14 14 15 15 16 16 17 17 18 18 19 19 20 20 21 21 22 22 23 23

Real Real Time Time (hr) (hr)

The number number of ofHandover HandoverIn In The (I i ) (Ii)

Figure 18. The traffic flow in Cell Figure in aaa whole whole day. day. Figure18. 18.The The traffic traffic flow flow in in Cell Cell666 in in whole day.

120 120 100 100 80 80 60 60 40 40 20 20 00

λλ == 11 call/hr call/hr 1/ 1/µ µ == 11 min/call min/call 00 11 22 33 44 55 66 77 88 99 10 10 11 11 12 12 13 13 14 14 15 15 16 16 17 17 18 20 21 21 22 22 23 23 18 19 19 20

Reat Reat Time Time (hr) (hr) Figure 19. The number of Handover In (I Cell66 in a whole day. Figure19. 19.The Thenumber numberof ofHandover Handover In In (I (I666))) in in Figure in Cell Cell6 in in aawhole wholeday. day.

5.2.2. Case Traffic Congestion 5.2.2.Case Case2:2: 2:Traffic TrafficCongestion Congestion 5.2.2. In the traffic accident simulation, this study analyzed the variability of traffic thetraffic trafficaccident accidentsimulation, simulation, this this study study analyzed analyzed the the relation relation of of InInthe the relation of the the variability variability of of traffic traffic information and the number of Handover In (I i). Figure 20 shows the traffic information in the information and the number of Handover In (I i). Figure 20 shows the traffic information in the information and the number of Handover In (Ii ). Figure 20 shows the traffic information in the simulation. At the simulation time 16th min, the traffic at the mileage of 5.05 km in simulation.At Atthe thesimulation simulation time time 16th 16th min, min, the the traffic traffic accident accident occurs occurs simulation. accident occurs at at the the mileage mileageof of 5.05 5.05km kmin in simulation, and the traffic jam occurs in Cell 5. The cars move slowly and line up in Cell5, and they simulation,and andthe the traffic jam occurs in Cell 5. The cars move slowly and line up in Cell5, and they simulation, traffic jam occurs in Cell . The cars move slowly and line up in Cell , and they can’t 5 5 can’t go into the Cell can’t Cell66.. For For Cell Cell66,, the the amount amount of of traffic traffic flow flow will will drop drop substantially substantially and and the the average average go intogo theinto Cellthe 6 . For Cell6 , the amount of traffic flow will drop substantially and the average vehicle speed also drops substantially after traffic accident occurrence. When the traffic accident is removed at the simulation time 45th min, the traffic condition will return to a free flow. Therefore, this study can discover that the amount of traffic flow increases rapidly when the traffic accident is removed. This study analyzed the number of Handover In (Ii ) before and after the traffic accident occurrence. However, the traffic accidents impact the number of Call Arrival, Handover In, Handover Out, and Call Departures in cellular systems. If this study applies the traditional SCA mechanism to support the handover events in cellular system, it will waste the number in the dynamic environment and cause the higher CBP with the amount of traffic flow to increase rapidly. Figure 21 shows that the number of Handover In (I6 ) drops substantially during the traffic accident in Cell6 between the simulation time 16th min and the 45th min. The number of Handover In (I6 ) increases substantially in Cell6 when the traffic accident is removed at the simulation time 46th min.

environment cause the higherinCBP with the amount traffic flowthe to increase 21 support the and handover events cellular system, it of will waste numberrapidly. in the Figure dynamic shows that theand number of Handover Inwith (I6) drops substantially trafficrapidly. accidentFigure in Cell21 6 environment cause the higher CBP the amount of trafficduring flow tothe increase between the the simulation 16th min In and 45th substantially min. The number of the Handover In (I6) increases shows that number time of Handover (I6the ) drops during traffic accident in Cell6 substantially Cell6 when the16th traffic accident removed thenumber simulation time 46th In min. between the in simulation time min and theis45th min. at The of Handover (I6) increases Inventions 2017, 2, 8 15 of 19 substantially in Cell6 when the traffic accident is removed at the simulation time 46th min. 200

Traffic Flow (Fi)(car/min) Traffic Flow (Fi)(car/min)

180 200 160 180 140 160 120 140 100 120 80 100 60 80 40 60 20 40 0 20 0

0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 63 66 69 72 75 78 81 84 87

simulation time (min) 0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 63 66 69 72 75 78 81 84 87 simulation time (min) Figure 20. The traffic flow in Cell6. Figure 20. The traffic flow in Cell6 . Figure 20. The traffic flow in Cell6.

number of Handover TheThe number of Handover In (IIn i) (Ii)

7 7 6

λ = 1 call/hr 1/µ = 1 min/call λ = 1 call/hr 1/µ = 1 min/call

6 5 5 4 4 3 3 2 2 1 1 0 0

0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 63 66 69 72 75 78 81 84 87

simulation time (min) 0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 63 66 69 72 75 78 81 84 87 simulation time (min)

Figure 21. The number of Handover Out (O6 ) in Cell6 . Figure 21. The number of Handover Out (O6) in Cell6.

Figure 21. The number of Handover Out (O6) in Cell6.

5.2.3. Comparisons between SCA and Proposed Mechanism 5.2.3. Comparisons between SCA and Proposed Mechanism InComparisons this subsection, this studyand analyzed the CBPs that are derived from the traditional SCA 5.2.3.In between Proposed this subsection, this SCA study analyzed theMechanism CBPs that are derived from the traditional SCA mechanism and the proposed mechanism under the two cases. The CBP (Pi ) is denoted as the number mechanism the proposed mechanism the two The CBP (Pi)the is traditional denoted as SCA the In this and subsection, this study analyzedunder the CBPs thatcases. are derived from of the call blocks (Bi ) divided by the number of handover In (Ii ) in Celli . The traditional SCA mechanism number of the callthe blocks (Bi) divided by the number of handover (Ii) in Cell(P i. The mechanism and proposed mechanism under the two cases.InThe CBP i) is traditional denoted asSCA the doesn’t consider the specific channel propagation conditions. Katzela and Naghshineh define the mechanism doesn’t consider specific channel propagation conditions. Katzela and Naghshineh number of the call blocks (Bi)the divided by the number of handover In (Ii) in Cell i. The traditional SCA parameter D and R [10], where D is the physical distance between the two cell centers equal to 1 km mechanism doesn’t consider the specific channel propagation conditions. Katzela and Naghshineh and the R is the radius of the cell equal to 0.5 km in this case. They only assumed that the distance (D/R) which depends on the cellular system environment is allowed to reuse the same channel in cells. Therefore, if the channel is allocated to Celli , , it cannot be reused in the same cell at the same time because of unacceptable co-channel interference. In SCA, the number of channels for handover events that are calculated by Formula (12) is equal to 2. Otherwise, in the proposed mechanism, the number of channels for Handover In in each cell is dynamically adjusted according to the traffic information in each cell. The comparison between SCA mechanism and the proposed mechanism is introduced as follows: 1 D N = σ2 , where σ = (12) 3 R In Case 1, this study analyzed the numbers of Handover In (Ii ) during peak hour and off-peak hour. Figure 22 shows that the minimum amount of traffic flow and the minimum number of Handover

to the traffic information in each cell. The comparison between SCA mechanism and the proposed mechanism is introduced as follows:

1 D N = σ 2 , where σ = 3 R.

(12)

Inventions 2017, 2, 8

16 of 19

In Case 1, this study analyzed the numbers of Handover In (Ii) during peak hour and off-peak hour. Figure 22 shows that the minimum amount of traffic flow and the minimum number of In (I6) in Cell6 are during 4:00 a.m. (i.e., off-peak hour). Otherwise, the maximum number In (IHandover 6 ) in Cell6 are during 4:00 a.m. (i.e., off-peak hour). Otherwise, the maximum number of the traffic the the traffic flow and number the maximum number of In (I 6) in Cell6 are during 8:00 a.m. (i.e., flowofand maximum of Handover In Handover (I6 ) in Cell 6 are during 8:00 a.m. (i.e., peak hour). peak hour). Figure 23 shows that the CBPs (P 6) are derived by using SCA and the proposed Figure 23 shows that the CBPs (P6 ) are derived by using SCA and the proposed mechanism in Cell6 mechanism in Cell6 during simulation time 4:00 a.m. and 8:00 a.m.. For Handover In events, the during simulation time 4:00 a.m. and 8:00 a.m.. For Handover In events, the number of channels in number of channels in the traditional SCA mechanism is set to 2 and the number of channels in the the traditional SCA mechanism is set to 2 and the number of channels in the proposed mechanism is proposed mechanism is dynamically adjusted depending on the actual traffic information. Through dynamically adjusted depending on the actual traffic information. Through Figure 23, this study can Figure 23, this study can discover that the all CBPs in this case are less than 21.5% by using the discover that all CBPswhich in thisiscase arethan lessusing than the 21.5% bymechanism. using the proposed mechanism, proposed the mechanism, better SCA The results indicate thatwhich the is better than using the SCA mechanism. results that the proposed mechanism is suitable proposed mechanism is suitable for theThe normal caseindicate in real life.

for the normal case in real life.

the number of Handover In (Ii)

數列2 λ = 1 call/hr 1/µ = 1 min/call 100 80 60 40 20 0 4 8 the time of whole day (hr)

Inventions 2017, 2, 8

Figure 22. The number of Handover In (I6 ) at 4:00 a.m. and 8:00 a.m. in Cell6 .

17 of 20

Figure 22. The number of Handover In (I6) at 4:00 a.m. and 8:00 a.m. in Cell6.

the call block probability (Pi)

Our proposed

SCA

λ = 1 call/hr 1/µ = 1 min/call

30% 25% 20% 15% 10% 5% 0% 4

8 the time of whole day (hr)

Figure 23. The CBPs (P6 ) are derived by SCA and the proposed mechanism during 4:00 a.m. and Figure 23. The CBPs (P6) are derived by SCA and the proposed mechanism during 4:00 a.m. and 8:00 8:00 a.m. in Cell6 . a.m. in Cell6.

In In Case 2, 2, the at the the mileage mileagepoint pointfrom from 5.05 to 5.35 between Case thetraffic trafficaccident accident occurs occurs at 5.05 to 5.35 km km between the the simulation time 15th and 45th min.This Thisstudy studyanalyzed analyzedthe therelation relation of of the the traffic traffic information simulation time 15th and 45th min. informationand and the the number of Handover InCell (I6) 6inbefore Cell6 and before andthe after the accident traffic accident occurrence. Through number of Handover In (I6 ) in after traffic occurrence. Through Figures 20 and can 21, discover this studythat canthe discover that the amount of traffic flow and the numberInof andFigures 21, this20 study amount of traffic flow and the number of Handover (I6 ) in Handover In (I 6) in Cell6 decrease rapidly when traffic accidents occur. Moreover, the number of Cell6 decrease rapidly when traffic accidents occur. Moreover, the number of Handover In (I6 ) in Cell6 Handover In (I6) in Cell6 dramatically increases when accidents are removed. Depending on the variation of traffic information, this study can expect that the number of Handover In (Ii) increases dramatically with the amount of traffic flow increase. Therefore, if the cellular system adopts the traditional SCA mechanism, it can’t afford the substantial increase of the number of channels for handover events, and it will cause the higher CBP. Figure 24 shows that the CBPs are derived by using the SCA mechanism and the proposed mechanism. This study can observe that the cars move

Inventions 2017, 2, 8

17 of 19

dramatically increases when accidents are removed. Depending on the variation of traffic information, this study can expect that the number of Handover In (Ii ) increases dramatically with the amount of traffic flow increase. Therefore, if the cellular system adopts the traditional SCA mechanism, it can’t afford the substantial increase of the number of channels for handover events, and it will cause the higher CBP. Figure 24 shows that the CBPs are derived by using the SCA mechanism and the proposed mechanism. This study can observe that the cars move slowly and line up in Cell5 and the amount of traffic flow is less in Cell6 when the traffic accident occurs between the simulation time 16th min and 45th min. During the traffic accident period in Cell6 , this study discovered that the CBP derived by using the proposed mechanism and SCA mechanism have little difference because the amount of traffic flow and the number of handovers are rare. Otherwise, this study observed that the number of handovers dramatically increases with the amount of traffic flow increase when the traffic accidents are removed between the simulation time between the 46th and 60th min. The cellular system that adopts the proposed mechanism can dynamically adjust the number of allocating channels in each cell for handover events, and it can effectively avoid the call block when the amount of traffic flow increases. Otherwise, the cellular system that adopts the SCA mechanism can’t dynamically adjust the number of allocating channels in each cell for handover events, so using the SCA mechanism will cause the higher CBP. Through Figure 24, all CBPs in the scenario of traffic accidents are less than 16.5% by using the proposed mechanism that is better than using the SCA mechanism. The simulation results show that the fixed number of channels by using the SCA mechanism can’t be suited to the dynamic environment. Moreover, the can accurately predict the number of channels that are required for handover. Inventions 2017, 2, mechanism 8 18 of 20 Therefore, the proposed mechanism can decrease the number of CBPs effectively.

The call block probability (Pi)

Our proposed

SCA

λ = 1 call/hr 1/µ = 1 min/call

70% 60% 50% 40% 30% 20% 10% 0% 0~15

16~30 31~45 46~60 61~75 simulation time (min)

76~90

Figure 24. The CBPs (P6 ) are derived by using SCA and the proposed mechanism in Cell6 . Figure 24. The CBPs (P6) are derived by using SCA and the proposed mechanism in Cell6.

5.2.4. Discussions 5.2.4. Discussions In experiments, this study designed two cases to simulate the traffic information and In experiments, thisThrough study the designed two cases simulate the2,traffic information and communication behavior. CBPs derived from to Case 1 and Case this study could discover communication behavior. Through the CBPs derived from Case 1 and Case 2, this study could that the CBPs of the cellular system adopting the proposed mechanism are always less than the discover that the mechanism. CBPs of the cellular system adopting the proposed mechanism are always less than traditional SCA Therefore, the simulation results indicate that the proposed mechanism the traditional SCA mechanism. Therefore, the simulation results indicate that the proposed can allocate the number of channels more precisely than the SCA mechanism. In Case 2, the maximum mechanism can allocate number of channels more precisely thanwhich the SCA mechanism. Casethe 2, CBP in a cellular systemthe using the proposed mechanism is 16.28%, is less than thatIn using the maximum CBP in a cellular system using the proposed mechanism is 16.28%, which is less than SCA mechanism. that using the SCA mechanism. 6. Conclusions According to the proposed mechanism, this study can analyze the communication behavior and the status of traffic. Moreover, this study can adopt the information of traffic flow to solve the problem of channel allocation in PCS. Through the simulation results, this study obtained the results derived from the whole day simulation and the accident simulation, respectively. The simulation

Inventions 2017, 2, 8

18 of 19

6. Conclusions According to the proposed mechanism, this study can analyze the communication behavior and the status of traffic. Moreover, this study can adopt the information of traffic flow to solve the problem of channel allocation in PCS. Through the simulation results, this study obtained the results derived from the whole day simulation and the accident simulation, respectively. The simulation results show that all CBPs in the scenario of the whole day are less than 21.5% by using the proposed mechanism, which is better than using the SCA mechanism. Moreover, all CBPs in the scenario of traffic accidents are less than 16.5% by using the proposed mechanism, which is better than using the SCA mechanism. Therefore, if the cellular system adopts the proposed mechanism, it can provide better quality of service in PCS. In this study, the highway in Taiwan is adopted in the simulation environment. The coverage of the cells is simpler than in the urban environment, and one car only has one MS. In the future, the simulation environment can change from the freeway to urban and one car can have multiple MSs. Author Contributions: Chi-Hua Chen and Bon-Yeh Lin conceived and designed the experiments; Che-Hao Lei performed the experiments; Chi-Chun Lo and Che-Hao Lei analyzed the data; Chi-Hua Chen and Che-Hao Lei contributed reagents/materials/analysis tools; and Che-Hao Lei wrote the paper. Conflicts of Interest: The authors declare no conflict of interest.

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15.

16.

Capone, A.; Redana, S. Call Admission Control Techniques for UMTS. In Proceedings of the IEEE 56th Vehicular Technology Conference, Atlantic City, NJ, USA, 7–11 October 2001. Chen, C.H.; Lin, B.Y.; Lin, C.H.; Liu, Y.S.; Lo, C.C. A Green Positioning Algorithm for Campus Guidance System. Int. J. Mobile Commun. 2012, 10, 119–131. [CrossRef] Lin, B.Y.; Chen, C.H.; Lo, C.C. A Novel Speed Estimation Method Using Location Service Events Based on Fingerprint Positioning. Adv. Sci. Lett. 2011, 4, 3735–3739. [CrossRef] Zhao, C.; Gan, L. Dynamic Channel Assignment for Large-Scale Cellular Networks Using Noisy Chaotic Neural Network. IEEE Trans. Neural Netw. 2011, 22, 222–232. [CrossRef] [PubMed] Fang, Y.; Zhang, Y. Call Admission Control Schemes and Performance Analysis in Wireless Mobile. IEEE Trans. Veh. Technol. 2002, 51, 371–382. [CrossRef] Del Re, E.; Fantacci, R.; Giambene, G. Handover and Dynamic Channel Allocation Techniques in Mobile Cellular Networks. IEEE Trans. Veh. Technol. 1995, 44, 229–237. [CrossRef] Esmailpour, A.; Nasser, N. Dynamic QoS-Based Bandwidth Allocation Framework for Broadband Wireless Networks. IEEE Trans. Veh. Technol. 2011, 60, 2690–2700. [CrossRef] Hoang, A.T.; Liang, Y.C.; Islam, M.H. Power Control and Channel Allocation in Cognitive Radio Networks with Primary Users’ Cooperation. IEEE Trans. Mob. Comput. 2010, 9, 348–360. [CrossRef] Jiang, H.; Rappaport, S. CBWL: A New Channel Assignment and Sharing Method for Cellular Communications Systems. In Proceedings of the IEEE 43th Vehicular Technology Conference, Secaucus, NJ, USA, 18–20 May 1993. Katzela, I.; Naghshineh, M. Channel Assignment Schemes for Cellular Mobile Telecommunication Systems: A Comprehensive Survey. IEEE Pers. Commun. 1996, 3, 10–31. [CrossRef] Sidi, M.; Starobinski, D. New Call Blocking versus Handoff Blocking in Cellular Networks. ACM J. Wirel. Netw. 1997, 3, 15–27. [CrossRef] Tekinay, S.; Jabbari, B. Handover and Channel Assignment in Mobile Cellular Networks. IEEE Commun. Mag. 1991, 29, 42–46. [CrossRef] Zhang, Y. Handoff Performance in Wireless Mobile Networks with Unreliable Fading Channel. IEEE Trans. Mob. Comput. 2010, 9, 188–200. [CrossRef] Everitt, D.; Manfield, A. Performance Analysis of Cellular Mobile Communication Systems with Dynamic Channel Assignment. IEEE J. Sel. Areas Commun. 1989, 7, 1172–1180. [CrossRef] Hong, D.; Rappaport, S.S. Traffic Model and Performance Analysis for Cellular Mobile Radio Telephone Systems with Prioritized and Nonprioritized Handoff Procedures. IEEE Trans. Veh. Technol. 1986, 35, 77–92. [CrossRef] Lin, Y.B.; Pang, A.C. Wireless and Mobile All-IP Networks; John Wiley & Sons: Hoboken, NJ, USA, 2005.

Inventions 2017, 2, 8

17. 18. 19. 20. 21. 22. 23. 24. 25.

26. 27. 28. 29. 30.

31. 32. 33. 34. 35. 36.

19 of 19

Mac Donald, V.H. Advanced mobile phone service: The cellular concept. Bell Syst. Tech. J. 1979, 58, 15–41. [CrossRef] Lin, Y.B.; Sou, S.I. Charging for Mobile All-IP Telecommunications; John Wiley & Sons: Hoboken, NJ, USA, 2008. Sin, J.; Georganas, N. A Simulation Study of a Hybrid Channel Assignment Scheme for Cellular Land-Mobile Radio Systems with Erlang-C Service. IEEE Trans. Commun. 1981, 29, 143–147. [CrossRef] Zhang, M.; Yum, T.-S.P. Comparisons of channel-assignment strategies in cellular mobile telephone systems. IEEE Trans. Veh. Technol. 1999, 38, 211–215. [CrossRef] Kahwa, T.; Georganas, N. A Hybrid Channel Assignment Scheme in Large-Scale, Cellular-Structured Mobile Communication Systems. IEEE Trans. Commun. 1978, 26, 432–438. [CrossRef] Ngo, C.Y.; Li, V.O.K. Fixed channel assignment in cellular radio networks using a modified genetic algorithm. IEEE Trans. Veh. Technol. 1998, 47, 163–172. [CrossRef] Antonopoulos, A.; Verikoukis, C. Traffic-Aware Connection Admission Control Scheme for Broadband Mobile Systems. IEEE Commun. Lett. 2010, 14, 719–721. [CrossRef] Alsamhi, S.H.; Rajput, N.S. Implementation of call admission control technique in HAP for enhanced QoS in wireless network deployment. Telecommun. Syst. 2016, 63, 141–151. [CrossRef] Antonopoulos, A.; Verikoukis, C. Dynamic Call Admission Control for Enhanced GoS of UGS Connections during “Busy Hour” in WiMAX. In Proceedings of International Conference on Mobile Lightweight Wireless Systems, Barcelona, Spain, 10–12 May 2010. Bolotin, V.A. Modeling Call Holding Time Distributions for CCS Network Design and Performance Analysis. IEEE J. Sel. Areas Commun. 1994, 12, 433–438. [CrossRef] Lin, B.Y.; Chen, C.H.; Lo, C.C. A Traffic Information Estimation Model Using Periodic Location Update Events from Cellular Network. Commun. Comput. Inf. Sci. 2011, 135, 72–77. Chang, H.C.; Chen, C.H.; Lin, B.Y.; Kung, H.Y.; Lo, C.C. Traffic Information Estimation Using Periodic Location Update Events. Int. J. Innov. Comput. Inf. Control 2013, 9, 2031–2041. Chen, C.H.; Chang, H.C.; Su, C.Y.; Lo, C.C.; Lin, H.F. Traffic Speed Estimation Based on Normal Location Updates and Call Arrivals from Cellular Networks. Simul. Model. Pract. Theory 2013, 35, 26–33. [CrossRef] Cheng, D.Y.; Chen, C.H.; Hsiang, C.H.; Lo, C.C.; Lin, H.F.; Lin, B.Y. The Optimal Sampling Period of a Fingerprint Positioning Algorithm for Vehicle Speed Estimation. Math. Probl. Eng. 2013, 2013, 306783. [CrossRef] Chang, M.F.; Chen, C.H.; Lin, Y.B.; Chia, C.Y. The Frequency of CFVD Speed Report for Highway Traffic. Wirel. Commun. Mob. Comput. 2015, 15, 879–888. [CrossRef] Wu, C.I.; Chen, C.H.; Lin, B.Y.; Lo, C.C. Traffic Information Estimation Methods from Handover Events. J. Test. Eval. 2016, 44, 656–664. [CrossRef] Lai, W.K.; Kuo, T.H.; Chen, C.H. Vehicle Speed Estimation and Forecasting Methods Based on Cellular Floating Vehicle Data. Appl. Sci. 2016, 6, 47. [CrossRef] Planung Transport Verkehr AG. VISSIM User Manual—V.5.20; Planung Transport Verkehr AG: Karlsruhe, Germany, 2009. Gomes, G.; May, A.; Horowitz, R. A Microsimulation Model of a Congested Freeway using VISSIM. Transp. Res. Rec. 2004, 1876, 71–81. [CrossRef] Higgs, B.; Abbas, M. Segmentation and Clustering of Car-Following Behavior: Recognition of Driving Patterns. IEEE Trans. Intell. Transp. Syst. 2015, 16, 81–90. [CrossRef] © 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).