US20040165561A1

US20040165561A1 - System for constructing a mobility model for use in mobility management in a wireless communication system and method thereof

Info

Publication number: US20040165561A1
Application number: US10/369,678
Authority: US
Inventors: Ta-gang Chiou; Constantin Chiscanu; Ching-Jui Hsiao
Original assignee: Chiou Ta-Gang; Constantin Chiscanu; Ching-Jui Hsiao
Current assignee: GROUNDHOG TECHNOLOGIES Inc
Priority date: 2003-02-21
Filing date: 2003-02-21
Publication date: 2004-08-26
Also published as: US8031676B2; AU2003293406A8; WO2004077845A2; WO2004077845A3; AU2003293406A1; US20100118725A1

Abstract

A method for constructing a mobility model for use in optimizing mobility management in a wireless communication system is disclosed, which comprises steps of collecting a plurality of network traffic statistic data of a plurality of desired items, wherein the desired items are related to the behavior of a plurality of subscribers of the wireless communication system; constructing the mobility model according to the network traffic statistic data, wherein the mobility model is for modeling the mobility behavior of the subscribers; and optimizing mobility management according to the mobility model.

Description

BACKGROUND OF THE INVENTION

1. Field of the Invention

The invention relates in general to a system for use in optimizing mobility management in a wireless communication system and the method thereof, and more particularly to a system for constructing a mobility model for use in optimizing mobility management in a wireless communication system and the method thereof.

2. Description of the Related Art

In order to satisfy the convenience, efficiency requirement of the modern people, wireless communication systems have been developed rapidly. All kinds of wireless communication devices, such as cell phones or mobile stations, are widely used in the world.

FIG. 1 shows the rough architecture of a sample wireless communication network. Using a cellular system as an example, the geographical coverage area of the wireless communication system is partitioned into cells, served by base stations. Each mobile station of an individual subscriber is connected to the wireless network via the base stations. The coverage of cells differs greatly according to various factors, such as the power of the base station, the geographical features (e.g. mountains, valleys, rivers) within the cells, the area (e.g. city, suburb) of the cells, the architecture (e.g. tall buildings, railroad, highway) within the cells, etc. One or more cells are respectively combined to a location area (LA), also known as Paging Area, Routing Area, or Registration Area in some systems. Basically, a location area is a region in which subscribers can move arbitrarily without requiring a location area update, which incurs central database update (e.g. in a Home Location Register or Visitor Location Register) of the location area information that is utilized for describing the current location area of subscribers. The size of a location area is defined to cover the demands raised by traffic density and flow, population density and subscriber mobility, etc.

Mobility management enables the wireless network to find the locations of mobile stations so as to deliver incoming calls, messages, or packets to mobile stations. Mobility management includes location update, paging, and other operations, such as handover, that are related to the location or mobility of subscribers. Since subscribers are free to move within the service area of the system, the system can only maintain the approximate location of each subscriber. When a connection needs to be established for a particular subscriber, the system has to determine the subscriber's exact location, to the accuracy of a cell, within the location area. When a subscriber leaves the border of the specific location area, the mobile station must register its new location area through signaling the location area information to the system. This procedure is called updating (location area update), or location registration. The updating procedure is for informing the system about the current location area of the subscriber. Besides location area update, there are also other types of location update that will be described later in this specification. When the system tries to deliver a phone call or message to a subscriber by first finding the location of the specific subscriber, the system can search among the cells within the current location area of the mobile station. This procedure is called paging. The paging procedure is for determining the exact location, to the accuracy of a cell, of the subscriber.

Because there are many tradeoffs and high complexity involved, the parameters involved in mobility management are difficult to define in an optimal manner. For example, how to define the scope, including size and the border cells, of location areas so as to decrease the overall cost of the wireless communication system is an important issues for optimizing mobility management. Since a location area is composed of cells, the size and the border of each location area can be defined by deciding which cells are collected into the location area. If the size of the location area is too small, mobile stations cross the location area frequently. As a result, the mobile stations perform location area update frequently and the location accuracy is to a smaller region, the system can thus have lower paging load. However, the system must waste its resources by performing frequent location area update, and the mobile station must waste its power transmitting the location area update signal. On the other hand, if the size of the location area is too large, mobile stations cross the border of the location areas rarely and do not perform location area update frequently. However, a large coverage area has to be paged when a call or a message arrives, which waste resource of the system. In addition, the border of the location area is also an important factor in defining the scope of the location area. If the border of the location area is set parallel to and close to major highways, or in heavy traffic regions where population and mobility behavior of the subscribers are high, the subscribers may result in much location area update. Furthermore, the subscribers may cross the border of a specific location area back and forth, thereby causing much location area update, if the border of the location area is not properly set. As a result, the system wastes its resources by processing frequent location area update procedure, and the mobile stations waste power transmitting the location area update signal.

Various conventional mobility models, such as fluid flow model, gravity mode, random walk model, etc, are presented as a basis for studying issues resulted from subscribers' behavior. For further discussion, please refer to “Location Management for Next-Generation Personal Communications Networks” (pp.18˜pp.24, IEEE Network, September/October 2000) incorporated herein by reference. Those conventional mobility models are more used for studying issues resulted from subscribers' behavior, than optimizing mobility management for a live wireless communication system. They lack enough precision and accuracy to practically optimize mobility management due to the following reasons.

First, each of these conventional models is based on certain intuitions and assumptions and might not correctly model realistic use of a live system. Taking the fluid flow model as an example, the fluid flow model is used for simulating the aggregate mobility behavior of the fluid in the flow or any other systems which the aggregate mobility behavior of the components in the system is like the fluid in the flow. It is obvious that the feature of the fluid flow is different from the subscribers of the wireless communication system in a crowded city. Therefore, the fitness of applying fluid flow model to model the aggregate mobility behavior of the subscribers in a crowded city is questionable. In the same manner, the fitness of other conventional mobility models is questionable for widespread use in optimizing mobility management in a live wireless communication system, since the real model is usually different from a hypothetical one.

Second, since the subscribers' mobility behavior in a wireless communication system is complicated, some of the conventional mobility models are oversimplified that they only specifically put stress on certain factors and significantly simplify the complexity of the mobility behavior of the subscribers. Therefore, those conventional mobility models lack enough accuracy to reflect the real mobility behavior of subscribers for optimization mobility management.

Third, since data of real wireless communication systems are difficult to acquire, most of the conventional mobility models are proposed or built upon simulated data, not a plurality of data obtained from real wireless communication systems. Due to the above-described reasons, the reliance and the accuracy of the simulation result of those mobility models are questionable if they are applied to optimizing mobility management.

In the present time, there is no optimal method or algorithm which is proven to be able to minimize the overall cost of mobility management based on realistic mobility model of subscribers. From time to time, mobility management parameters need to be redefined. For example, from time to time the scope of location area needs to be redefined, such as splitting it into two new location areas when equipment of the old location area reaches its paging capacity. In practice, mobility management parameters are defined according to the subjective experience and rough judgment of the wireless operators. It is obvious that this kind of methods have difficulty minimizing the overall cost of mobility management. Although there are optimization algorithms that can be used, as will be introduced and incorporated later in this specification, without a realistic mobility model their practicability and accuracy are highly limited. The conventional approaches either has difficulty to find the optimal strategy out of numerous possible configurations, or do not concisely take into account the mobility characteristics of the entire system. This results in extra or unbalanced resource consumption, and leads operators to more capital expenditure on hardware expansion as mobility management traffic grows.

SUMMARY OF THE INVENTION

It is therefore an objective of the present invention to provide a system for constructing a mobility model for use in optimizing mobility management in a wireless communication system and a method thereof so as to achieve the objectives, which the mobility model is constructed according to network traffic statistic data collected from the wireless communication system to realistically reflect the mobility behavior of the subscribers. In this manner when optimizing mobility management according to the mobility model of the present invention, the overall cost can be reduced. Besides, the extra or unbalanced resource consumption, and the unnecessary capital expenditure on hardware expansion as mobility management traffic grows can be avoided.

According to the objectives of the present invention, a system is provided for constructing a mobility model for use in optimizing mobility management in a wireless communication system. The system comprises a processor and a storage medium. The processor further includes a mechanism for collecting a plurality of network traffic statistic data of a plurality of desired items, wherein the desired items are related to the behavior of a plurality of subscribers of the wireless communication system, a mechanism for constructing the mobility model according to the network traffic statistic data, wherein the mobility model is for modeling the mobility behavior of the subscribers, and a mechanism for optimizing mobility management according to the mobility model. The storage medium coupling to the processor is for storing the network traffic statistic data, data of the mobility model, and data of optimizing mobility management.

A method is also provided for constructing a mobility model for use in optimizing mobility management in a wireless communication system. The method comprises collecting a plurality of network traffic statistic data of a plurality of desired items, wherein the desired items are related to the behavior of a plurality of subscribers of the wireless communication system; constructing the mobility model according to the network traffic statistic data, wherein the mobility model is for modeling the mobility behavior of the subscribers; and optimizing mobility management according to the mobility model.

Other objects, features, and advantages of the invention will become apparent from the following detailed description of the preferred but non-limiting embodiments. The following description is made with reference to the accompanying drawings.

BRIEF DESCRIPTION OF DRAWINGS

FIG. 1 shows a sample rough architecture of a wireless communication system; [0017]
FIG. 2 illustrates the flow chart of the mobility model construction method for use in optimizing mobility management according to the preferred embodiment of the present invention; [0018]
FIG. 3 shows a diagram used to explain the way to implement mobility model construction method of the present invention; [0019]
FIG. 4 illustrates the flow chart of optimizing a mobility management parameter according to the preferred embodiment of the present invention; and [0020]
FIG. 5 illustrates the system for executing the mobility model construction method for use in optimizing mobility management according to the preferred embodiment of the present invention.[0021]

DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS

The feature of the disclosed method of the present invention is to construct the mobility model of the wireless communication system according to network traffic statistic data collected from the wireless communication system to realistically reflect the mobility behavior of the subscribers first, and then optimize the mobility management related parameters according to the mobility model of the present invention. [0022]
The conventional mobility models are either derived from other models that are based on certain intuitions and assumptions not suitable for many practical situations, or only specifically put stress on certain factors to simplify the complexity of the mobility behavior of the subscribers. Therefore, those mobility models are not able to accurately reflect the real mobility behavior of subscribers for optimizing mobility management. In the present invention, a mobility model construction method is provided in order to construct a mobility model based on the network traffic statistic data collected from different parts of the real wireless communication system. The network traffic statistic data may reflect different aspects and metrics of subscribers' behavior of the wireless network. The mobility model constructed according to the network traffic statistic data combines all the different aspects and metrics of subscribers' behavior, and realistically reflects the mobility behavior of the subscribers in the system. In addition, the mobility model of the present invention needs not be set up through oversimplifying the complexity of the wireless communication system. It can be utilized to provide the objective optimal solution to minimize the overall cost of mobility management when defining or redefining mobility management related parameters, such as the scope of a location area. The capital expenditure of the system operators can thus be reduced. [0023]
FIG. 2 illustrates the flow chart of the mobility model construction method according to the preferred embodiment of the present invention. The mobility model construction method according to the preferred embodiment of the present invention comprises the following steps. In the beginning, [0024] step 202 of collecting network traffic statistic data of desired items is executed. The network traffic statistic data can be generated by different devices of the system, such as MSC, VLR of the Network Switching Subsystem (NSS) and/or the BSC of the Base Station Subsystem (BSS). The network traffic statistic data can, but not necessarily, be collected through the Operations Support Subsystem (OSS) or Operation and Maintenance Center (OMC) of the wireless communication system.
In the preferred embodiment of the present invention, the desired items include: Location Update or Total Location Update (LU, which includes different types of location update such as location area update), Periodic Location Update (PLU), Attaching Location Update (ATA); Sum of Mobility Rate (ΣD); Handover (HO), which may be directional handover (HO[0025] _ji) or handover due to a specific cause (e.g. Power Budget Handover, PBGT HO_i); calls, which may be Mobile Terminating Call (MTC) or Mobile Originating Call (MOC); ratio of mobility behavior causing handover (A_i), ratio of calls causing handover (B_i); SMS (Short Message Service) counter; Paging (PN), et cetera. Each of the desired items may help reflect a specific aspect of mobility behavior of the subscribers. In a preferred embodiment, the Location Update (LU) and Handover (HO) are the necessary items. Other items are optional items for implementing the better mode of the present invention. Detail of each item will be described in the following specification.
The step of collecting network traffic statistic data of desired items can further include a step of pre-processing the network traffic statistic data if needed. Since the network traffic statistic data are generated by different devices of the system, the format may not be coherent. The pre-processing step can be executed to unify the format of the network traffic statistic data so as to apply the statistic data to construct the mobility model. For example, the statistic data of some desired items may be hourly based and that of other items may be daily based. The time unit of the statistic data of all desired items can be unified through executing the pre-processing step. In addition, the corresponding geographical area of the statistics of all desired items can be determined and unified as well through executing the pre-processing step. [0026]
Then, step [0027] 204 of constructing a mobility model for modeling mobility behavior of subscribers in the system is executed. The mobility model implies knowledge such as how many subscribers move from one cell to another cell at different time. To understand the construction of mobility model, analogy in computer vision may be helpful: With photos of a certain object from several different angles, a computer with advanced computer vision algorithms is able to construct the 3D model of the object based on these 2D photos. Similarly, the construction of mobility model is based upon network traffic statistic data containing different aspects and metrics of subscribers' behavior from different parts of the wireless network. As mentioned above, the contributing information sources may include time series of counters and success ratio of location update, handover, paging, calls, etc. By itself, each of these observations reveals only a slice of the complete state of the subscribers' mobility behavior. The present invention interweaves the data gathered from these disparate information sources to construct the mobility model of subscribers. One equation, simultaneous equations, inequations, or nonlinear mathematics (e.g. neural network) corresponding to the network traffic statistic data can be solved to generate the mobility model, which is an accurate spatial-temporal model of subscribers' mobility behavior, over time and between every pair of cells.
In the preferred embodiment, in order to construct the mobility model, the behavior of the subscribers corresponding to each cell is determined first. Please refer to FIG. 3, which shows a diagram used to explain the way to implement mobility model construction method of the present invention. Taking CELL[0028] ₁as an example, it belongs to the location area LA1 and is a border cell of LA1. CELL₁is neighbor to CELL₂and CELL₃, which belongs to the location area LA2 and LA3 respectively, as shown in FIG. 3.
In the preferred embodiment, when constructing the mobility model of the system, the mobility behavior of the subscribers in each border cell of the location area are determined first, then the mobility behavior of the subscribers in the inner cells, which are not border cells, of the location area are determined. [0029]
It should be noted that not only when the subscriber crosses the border of the specific location area to another location area will the mobile station execute the location update procedure. For example, when a subscriber has stayed in the specific location area over a specific period of time without sending information to the wireless network, or when the subscriber turns on the mobile station in the specific location area, the mobile station may execute the location update procedure as well. [0030]
The Location Update (LU[0031] ₁) of CELL₁represents the total number of location update recorded in CELL₁during a predetermined time period. In the preferred embodiment, the magnitude of Location Update (LU_i) can be determined by the following items. (a). Sum of Mobility Rate (ΣD_ji) which represents the total number of the subscribers which move from the neighbor cells (e.g. CELL₂and CELL₃) belonging to another location area (e.g. LA2 and LA3 respectively) into CELL_i(e.g. CELL₁). (b). Periodic Location Update (PLU_i) which represents the Periodic Location Update triggered by mobile stations that have stayed in CELL_iover a specific period of time without sending information to the network. (c). Attaching Location Update (ATA_i) which represents the times of mobile stations being turned on in CELL_iduring the predetermined time period. The relation among Sum of Mobility Rate (ΣD_ji), Periodic Location Update (PLU_i), and Attaching Location Update (ATA_i) selected from the set of network traffic data can be shown as the mobility equation (Eq.1) as follows:
LU _i =f _l(Σ_{j, cell j is adjacent to cell i, and cell j belongs to a different LA than cell i,} D _ji , PLU _i, ATA _i) (Eq. 1)
A reasonable f[0032] _lcan be chosen according to practice. For example, an applicable form of Eq.1 and f_lis:
LU _i=Σ_{j, cell j is adjacent to cell i, and cell j belongs to a different LA than cell i,} D _ji +PLU _i +ATA _i (Eq. 1′)
It should be noted that Sum of Mobility Rate (ΣD[0033] _ji) represents the total number of mobile stations moving into CELL_ifrom cells belonging to a different location area from that of CELL_i. Sum of Mobility Rate (ΣD_ji) is a necessary term in the right hand side of equation (Eq.1), and other terms in the right hand side may be skipped in certain embodiments, in exchange for faster processing while lower accuracy. If terms in the right hand side besides Sum of Mobility Rate (ΣD_ji) are skipped, Location Update (LU) are considered to be close to Location Area Update described in the Description of the Related Art of the present specification. In addition, all other situations than the above-disclosed situations which the mobile station will execute location update can be put into consideration when constructing the mobility model.
When a mobile station moves from the coverage of [0034] CELL 1 to that of CELL 2, the connection between the mobile station and the network system must be changed from via the base station of CELL 1 to that of CELL 2. The process is called handover. There are different causes of handover, such as power budget, quality, interference, and level of signals being better in the new cell. Note that not only the mobility behavior of subscribers but also other causes can cause handover. According to our experience, Power Budget Handover (PBGT HO) is usually more frequently caused by mobility behavior of subscribers than handover of other causes. Nevertheless, a substantial portion of handover represents the times of mobile stations moving from other neighbor cells, such as from CELL_jto CELL_i. In a typical wireless communication system, there can be different types of handover statistic data. For instance Directional Handover (HO_ji) statistic data record the times of subscribers moving from CELL_jto CELL_i., with knowledge of both the source cell (CELL_i) and destination cell (CELL_j). In certain wireless communication systems, Power Budget Handover (PBGT HO_i) statistic data may record the total number of handover caused by power budget, from all neighboring cells into the destination cell (CELL_i), without differentiating the source cell. In some other wireless communication systems, Power Budget Handover (PBGT HO_ij) statistic data may contain more specific directional information by recording the number of handover caused by power budget, from the source cell (CELL_i) to the destination cell (CELL_j).
The magnitude of handover can relate to factors such as: (a). Mobility Rate (D[0035] _ji) which represents the times of the mobile stations moving from CELL_jto CELL_i; (b). Call Rate (e.g. C_ior C_i), which is a function of Mobile Terminating Call (e.g. MTC_ior MTC_j) and Mobile Originating Call (e.g. MOC_ior MOC_j). Mobile Terminating Call (e.g. MTC_ior MTC_j) represents the times of subscribers located within a cell (e.g. CELL_ior CELL_j) receiving calls; Mobility Originating Call (e.g. MOC_ior MOC_j) represents the times of subscribers located within a cell (e.g. CELL_ior CELL_j) calling out.
In one embodiment, the relation among Directional Handover (HO[0036] _ji), Mobility Rate (D_ji), Call Rate (C_ij), Mobile Terminating Call (MTC_ior MTC_j), Mobility Originating Call (MOC_ior MOC_j) can be defined by the Mobility Equations (Eq2 and Eq3) as follows:
HO _ji =f _h(D _ji , C _ij) (Eq2)
C _ij =f _c(MTC _i , MOC _i , MTC _j , MOC _j) (Eq3)
Eq2 and f[0037] _h, as well as Eq3 and f_c, can further be implemented with a reasonable selection of linear or nonlinear equations. For instance, one of the simplest forms applicable is:
HO _ji =D _ji =*A _i +C _ij *B _i (Eq2′)
C_ij=MTC_i (Eq3′)
A[0038] _iis the ratio of mobility behavior causing handover, and B_iis the ratio of calls causing handover. The parameters A_iand B_i, as well as D_ji, of border cells can be derived by applying corresponding network statistic data (e.g. LU_i, PLU_i, ATA_i, MTC_i, MOC_i, MTC_j, MOC_j, HO_jiof border cells) to the equations (Eq1′, Eq2′, and Eq3′) and see what A_iand B_igenerates the best result (e.g. with minimum least square error.) Therefore, the mobility behavior of the subscribers corresponding to CELL_ithat are border cells can be determined. There can be different A_iand B_ifor different time, when the CELL_iis having different loading, or between different pairs of cells, given enough statistic data for calculation. Therefore, A_iand B_ican be further specified as A_itand B_it, A_il, and B_il, or A_ijand B_ij. In addition, all other situations than the above-disclosed situations which the mobile station will execute handover or whose counter is related to that of handover can be put into consideration when construction the mobility model.
Referring to FIG. 3 again, the mobility behavior of the subscribers corresponding to all border cells of LA[0039] 1, such as CELL₁, CELL₄, CELL₅, CELL₆, CELL₇, and CELL₈can be calculated in the same manner. Since A_iof all border cells of LA1 are available, the A_iof inner cells of LA1, such as CELL₉, can be calculated by that of the neighboring cells according to the following Mobility Equation Eq4:
A _g =f _a(A ₁ , A ₂ , A ₃ , A ₄ , A ₅ , A ₆ , A ₇ , A ₈) (Eq4)
Since B[0040] _iof all border cells of LA1 are available, the B_iof inner cells of LA1, such as CELL₉, can be calculated by that of the neighboring cells according to the following Mobility Equation Eq5:
B _g =f _b(B ₁ , B ₂ , B ₃ , B ₄ , B ₅ , B ₆ , B ₇ , B ₈) (Eq5)
Eq4 and f[0041] _a, as well as Eq5 and f_b, can further be implemented with a reasonable selection of linear or nonlinear equations. For instance, one of the simplest forms applicable is:
A ₉=(A ₁ *HO ₁₉ +A ₂ *HO ₂₉ +A ₃ *HO ₃₉ +A ₄ *HO ₄₉ +A ₅ *HO ₅₉ +A ₆ *HO ₆₉ +A ₇ *HO ₇₉ +A ₈ *HO ₈₉)/(HO ₁₉ +HO ₂₉ +HO ₃₉ +HO ₄₉ +HO ₅₉ +HO ₆₉ +HO ₇₉ +HO ₈₉) (Eq4′)
B ₉=(B ₁ *C ₁₉ +B ₂ *C ₂₉ +B ₃ *C ₃₉ +B ₄ *C ₄₉ +B ₅ *C ₅₉ +B ₆ *C ₆₉ +B ₇ *C ₇₉ +B ₈ *C ₈₉)/(C ₁₉ +C ₂₉ +C ₃₉ +C ₄₉ +C ₅₉ +C ₆₉ +C ₇₉ +C ₈₉) (Eq5′)
Another more accurate but computationally intensive example of implementation of Eq4 and f[0042] _ais:
A ₉=(A ₁ *D ₁₉ +A ₂ *D ₂₉ +A ₃ *D ₃₉ +A ₄ *D ₄₉ +A ₅ *D ₅₉ +A ₆ *D ₆₉ +A ₇ *D ₇₉ +A ₈ *D ₈₉)/(D ₁₉ +D ₂₉ +D ₃₉ +D ₄₉ +D ₅₉ +D ₆₉ +D ₇₉ +D ₈₉) (Eq4″)
The simultaneous equations of Eq1˜Eq5 of cells in the location area can thus be listed, and A[0043] _i, B_i, D_jican be solved with standard statistical methods given enough data points of network statistic data. The mobility behavior of the subscribers can thus be determined. The mobility model based on various network traffic statistic data collected from the real network system to model the mobility behaviors of the subscribers in the real network system can thus be constructed.
Every single cell in the location area contributes to the total paging number of the location area (PN[0044] _L1). Taking CELL₁of LA1 as an example, Paging Rate (PR₁) of CELL₁represents the number of paging of PN_L1contributed by paging subscribers in CELL₁. PR₁can be calculated according to the mobility Equation (Eq6) as follows:
PR ₁=(MTC ₁ /MTC _L1)*PN _L1 (Eq6)
MTC[0045] ₁represents the times subscribers, located within CELL₁, receiving phone calls. MTC_L1represents the times of subscribers, located within LA1, receiving phone calls. Therefore, MTC_L1=ΣMTC_i, for all CELL i belonging to LA1. Paging Rate of others cells can thus be calculated in the same manner.
It should be noted that all other situations than the above-disclosed situations which the mobile station will execute paging can be put into consideration when constructing the mobility model. [0046]
As another example, mobility model can also be constructed starting with the following equations: i,j are indices to CELLs, [0047]
LU _i=(Σ_{j, cell j is adjacent to cell i, and cell j belongs to a different LA than cell i,} D _ji)*K _i (Eq7)
HO _ij =D _ij *A _i +C _i *B _i (Eq8)
PBGT HO _i=(Σ_{j, cell j is adjacent to cell i} D _ij)*E _i +C _i *F _i (Eq9)
2*C _i =MTC _i +MOC _i (Eq10)
Eq. 7 is similar to Eq. 1′, and Eq. 8 is similar to Eq. 2′, except certain terms are ignored because they are small in magnitude relatively. Among all different types of handover, Eq. 9 utilizes statistic data of certain handover due to a specific cause which may most likely be resulted from subscribers' mobility behavior. For example, PBGT HO, which stands for power budget handover, is handover due to the cause of power budget. In our experience, PBGT HO is more likely to be resulted from subscribers' mobility behavior, as compared to handover caused by quality, interference, or level. In some wireless communication systems, PBGT HO[0048] _icannot differentiate the source cells, and in some other wireless communication systems, PBGT HO_ijcan differentiate the source cells. In this embodiment, PBGT HO, which cannot differentiate the source cells, is used as an example. Because PBGT HO_ican be caused by mobility behavior of subscribers, the first term in Eq. 9 is proportional to the times of subscribers moving into CELL_i. However, PBGT HO_ican also result from random noises or interferences that decrease the power level of signals such that a handover is needed to maintain the signal level, hence the second term in Eq. 9 reflects the fact that a portion of the calls will suffer from a decrease in signal level that results in a power budget handover.
The mobility model construction can also be extended with first applying a step of smoothing the call rate C[0049] _i. This step is to smoothen the noises in the data to get the underlying pattern. Therefore, instead of Eq. 10, we have the following:
C _i=[(MTC _i +MOC _i)/2+λ*Σ_{j, cellj is adjacentto cel i,}(MTC _j +MOT _j)/2]/(1+λ*Σ_{j, cell j is adjacent to cell i}) (Eq11)
where λ is a smoothing factor. λ can be assumed a value or obtained by optimization based on cross-validation, which is to be described. [0050]
The mobility model construction can further be extended by obtaining a preliminary estimate of the mobility model. In this example, C[0051] _iis substituted back into Eq8 to obtain a preliminary estimate of the mobility model. In the preferred embodiment, the value of B_iis obtained by doing a regression analysis with HO_ijbeing the regressand and C_ibeing the regressor. B_iis in general i dependent. In one embodiment, B_iis assumed to be constant over all cells but it can also be obtained by doing regression analysis on subsets of the cells and optimized using cross-validation. A preliminary estimate of D_ijcan thus be calculated:
D _ij *A _i =HO _ij −C _i *B _i (Eq 12)
Again, A[0052] _ican be assumed a value or obtained by optimization based on cross-validation.
The next step is to use this preliminary D[0053] _ijto compute Σ_jD_jiand Σ_jD_ijin Eq7 and Eq9. Similar to the case of B_i, coefficients K_i, E_i, and F_iare calculated by doing linear regression analyses on Eq7 and Eq9 with Σ_jD_ji, Σ_jD_ij, and C_ibeing the regressors, and LU_iand PBGT HO_ibeing the regressands. The regressions analyses also yield two root-mean-square errors σB_iand σE_ifor Eq7 and Eq9 respectively. Finally, all the data are put together to compute the mobility model:
Dfinal_ij =[D _ij+(B _i /σB _i *Σ _j D _ij/number of cells adjacent to cell i and residing in a different LA than cell i)+(E _i /σE _i *Σ _j Dij/number of cells adjacent to cell i)]/(1+B _i /σB _i +E _i /σE _i) (Eq13)
The accuracy of a model can be verified by cross-validation. The constructed mobility model D[0054] _ijcan be substituted into the right-hand side of Eq7, and we can compare the resulting number with the available data LU_i. This comparison yields a measure of the accuracy of the model. The values of λ and A_ican be optimized based on this measure. Modifications and combinations of the above two methods of mobility model construction can be done to yield various mobility models. For example, a preliminary mobility model D_ijcan be computed using the first method instead, and then the final estimator for D_ijcan be obtained by doing regression analyses on Eq7 and Eq9 and evaluating Eq13. Cross-validation can be carried out over all the cells or on local cells to indicate the degree of validity for each of the mobility model. Therefore, a most accurate mobility model can be found for each pair of cells.
After mobility model is constructed, step [0055] 206 of optimizing mobility management according to the constructed mobility model is executed. Since the mobility model is constructed based on the practical mobility behavior of the subscribers in the real world, optimizing mobility management based on the mobility model can minimize the overall cost of mobility management.
FIG. 4 illustrates the flow chart of optimizing mobility management parameters according to the preferred embodiment of the present invention. In this embodiment, the K-L (Kernighan Lin) algorithm is used for determining the scope of location area. It should be noted that the parameters of mobility management are not limited to the scope of the location area. Other parameters of mobility management that can be optimized include, but are not limited to, number and scope of location areas; value of different timers such as circuit-switch/packet-switch Periodic Location Update timer and RRC Connection Release timer, and Inactivity timer; network topology and connectivity of different network elements such as MSC, BSC, and BTS; paging parameters such as paging duration, paging retry times, paging scope, sequence of paging; method of handover (e.g. hard, soft, or softer handover), hysteresis factor of handover; coverage of cells, et cetera. In addition, the algorithm used for adjusting parameters of mobility management is not limited to the K-L algorithm. Other applicable algorithms include, but are not limited to, greedy algorithm, F-M (Fiduccia Mattheyses) algorithm, Genetic Algorithm, and simulated annealing algorithm. For information on K-L algorithm, please refer to “An Efficient Heuristic Procedure for Partitioning Graphs” (The Bell system technical journal, 49(1):291-307, 1970). For information on greedy algorithm, please refer to “Introduction to Algorithms: A Creative Approach, [0056] chapter 7” (pp.210˜pp.211, Addison-Wesley Publishing Company, 1989). For information on F-M algorithm, please refer to “A Linear-Time Heuristic for Improving Network Partitions” (Proc. of DAC, 1982). For information on Genetic Algorithm, please refer to “A Genetic Algorithm For Optimizing Multiple Part Placement To Reduce Build Time” (Proceedings of the Fifth International Conference on Rapid Prototyping, Dayton, Ohio, June 1994). For information on simulated annealing algorithm, please refer to “Location Area Planning in Cellular Networks Using Simulated Annealing” (Proceedings of IEEE Infocom, The Conference on Computer Communications 2001, Anchorage, Ak., Apr. 22-26, 2001). The above-mentioned publications are incorporated herein by reference. All other optimization algorithms can be used as well.
In the beginning, [0057] step 402 of defining the scope of each location area is executed. The number of location areas and the scope of each location area can be determined according to the number of cells and the network's rules/constraints, such as the paging capacity and channel capacity. For example, when collecting the cells into one location area, it should be noted that the sum of Paging Rate (PR) of all cells cannot exceed the paging capacity of the location area. The Paging Rate (PR) of each cell can be determined according to the mobility equation (Eq6) disclosed above. Many rules/constraints, such as capacity limit of various equipments, RF planning experience (i.e. rule-of-thumb), geographical conditions, and so on, are taken into account at this step.
Then, step [0058] 404 of computing the overall costs of the system network is executed. Since the mobility model of the present invention can model the mobility behavior of all subscribers, the overall costs of mobility management of the system can be determined. The mobility model of the present invention can be determined according to the mobility equations disclosed above. It should be noted that in the practical wireless communication network system, the cost of every single paging, location update, and handover operation might not be the same for different network elements. When executing mobility management, we take into account the difference in cost overhead of operations involving different network elements, such as the RF costs of the Base Station Subsystem, A-interface loading, and computation costs of the Network Switching Subsystem. For instance, inter-VLR and intra-VLR location area updating results in different costs for certain network elements, and inter-BSC handover and intra-BSC handover contribute to different costs for MSC. Therefore, not only the number of location update and paging, but also the cost of every single operation, are put into consideration when computing the overall cost of the system.
Then, in one embodiment, step [0059] 406 of redefining the scope of location areas according to K-L algorithm is executed. Taking CELL₁as an example, CELL₁belongs to LA1 in the original network system, as shown in FIG. 3. When executing the K-L algorithm, the location area of CELL₁is changed to LA2 and the location area of one cell belonging to LA2, such as CELL₂is changed to LA1 and then computing the overall cost of the system. In the same manner, all other possible ways to exchange the location area of every two cells in different location areas are executed and the overall costs of the resulting system are evaluated. When redefining the location area scope formed by executing K-L algorithm, the overall cost of the system may be either increased or decreased. The desired scope of location areas is the one for which the overall cost of the system is minimal compared to the overall cost of other ways of location area scope definition.
Then, step [0060] 408 of determining whether the new way of location area scope definition is the better solution than the original way of location area scope definition is executed. Step 408 can be executed through comparing the system's overall cost of the new location area scope form to that of the original location area scope form. If the system's overall cost of the new location area scope form is lower than that of the original location area scope form, step 410 of replacing the original way of location area scope definition with the new one is executed then the algorithm returns to step 406 to execute the K-L algorithm again. If the system's overall cost of any other location area scope configurations are all higher than that of the original location area scope configuration, step 412 is executed that the original way of location area scope definition is proven to be the best solution to mobility management through executing the K-L algorithm.
In this example, cell is the smallest unit for the scope of location area. Practically, the smallest unit can be different according to different systems. For example, some operators may prefer to use a Base Station, a Node-B, or an Access Point as the smallest unit for the scope of location area. [0061]
Besides K-L algorithm, other optimization algorithms, such as F-M (Fiduccia, C. M. and R. M. Mattheyses) algorithm, greedy algorithm, genetic algorithm, simulated annealing, etc, can be executed as well, when optimizing mobility management of the wireless communication system. Since the mobility model of the present invention is constructed based on the actual mobility behavior of all subscribers, not the subjective experience of the system operators, the result of mobility management through executing the method disclosed in the specification can be proven to lower the overall cost of the system. In this manner, the capital expenditure and hardware expansion of the system operators can thus be decreased. [0062]
The method disclosed in the present invention is for use in all different kinds of wireless network, including not only the second generation wireless networks such as GSM, CDMA, and PDC, but also more advanced systems such as 2.5G systems like GPRS and EDGE, 3G systems like WCDMA, CDMA2000, and TD-SCDMA, and other wireless communication systems such as PHS and Wireless LAN/IPv6 networks. The terms used in different systems may be different but the concept is similar. For example, the concept of location area of GSM is similar to the routine area of packet-switched systems, zone and paging area of PDC and PHS, as well as UTRAN Registration Area of WCDMA. The concept of the network topology and connectivity of different network elements such as NSS, MSC, BSC, BSS, BTS, A-Interface in GSM system is similar to Core Network, SGSN/GGSN, RNS, RNC, Node-B, I[0063] _uPS-lnterface in WCDMA packet-switched system, or Core Network, WRT, ELU, Radio Network, Cell Station (CS), WRT-ELU interface in PHS system. Therefore, the method disclosed in the present invention can be applied to other kinds of wireless communication systems as well.
FIG. 5 illustrates the system for executing the mobility model construction method for use in optimizing mobility management according to the preferred embodiment of the present invention. The system includes a [0064] processor 502 and a storage medium 504. The processor 502 further includes a mechanism 506 for collecting a plurality of network traffic statistic data, a mechanism 508 for constructing the mobility model according to the network traffic statistic data, and a mechanism 510 for optimizing mobility management according to the mobility model. In addition, the mechanism 506 for collecting network traffic statistic data can further include a mechanism 512 for pre-processing the network traffic statistic data if needed. The mechanism 508 for constructing mobility model can further include a mechanism 514 for obtaining the preliminary estimate of the mobility model, and a mechanism 516 for verifying the accuracy of the mobility model. The operations of these mechanisms are disclosed in the above specification. In addition, each of the above-disclosed mechanisms can be implemented in the form of either a hardware circuit or a software program.
The [0065] storage medium 504 coupling to the processor is for storing the network traffic statistic data, data of the mobility model, and data of optimizing mobility management. The storage medium 504 can be any kind of storage device used in computer system.
While the invention has been described by way of example and in terms of the preferred embodiment, it is to be understood that the invention is not limited to the disclosed embodiment. On the contrary, it is intended to cover various modifications and similar arrangements and procedures, and the scope of the appended claims therefore should be accorded the broadest interpretation so as to encompass all such modifications and similar arrangements and procedures. [0066]

Claims

What is claimed is:

1. A method for constructing a mobility model for use in optimizing mobility management in a wireless communication system, wherein the wireless communication system includes a plurality of location areas and each of the location area includes a plurality of cells, comprising:

collecting a plurality of network traffic statistic data of a plurality of desired items, wherein the desired items are related to the behavior of a plurality of subscribers of the wireless communication system;

constructing the mobility model according to the network traffic statistic data, wherein the mobility model is for modeling the mobility behavior of the subscribers; and

optimizing mobility management according to the mobility model.

2. The method according to claim 1, wherein the step of collecting network traffic statistic data of the desired items further comprising pre-processing the network traffic statistic data of the desired items, wherein each of the desired items is generally corresponding to specific behavior of the subscribers.

3. The method according to claim 2, wherein the desired items at least include location update.

4. The method according to claim 2, wherein the desired items at least include handover.

5. The method according to claim 1, wherein the mobility model is constructed by solving simultaneous equations corresponding to the network traffic statistic data.

6. The method according to claim 1, wherein the mobility model is constructed by solving nonlinear mathematics corresponding to the network traffic statistic data.

7. The method according to claim 1, wherein the step of constructing the mobility model further comprising obtaining a preliminary estimate of the mobility model.

8. The method according to claim 1, wherein the step of constructing the mobility model further comprising verifying the accuracy of the mobility model.

9. The method according to claim 1, wherein the cells of the location areas at least include a border cell and an inner cell, wherein the mobility model at least includes a first submodel for modeling the behavior of the subscribers belonging to border cells and a second submodel for modeling the behavior of the subscribers belonging to inner cells.

10. The method according to claim 1, wherein the step of optimizing mobility management is executed for adjusting a plurality of parameters of mobility management.

11. The method according to claim 10 wherein the parameters of mobility management at least include the scope of each location area.

12. The method according to claim 10, wherein the parameters of mobility management at least include the magnitude of a plurality of timers.

13. The method according to claim 10, wherein the parameters of mobility management at least include network topology and connectivity of different network elements.

14. The method according to claim 10, wherein the parameters of mobility management at least include paging parameters.

15. The method according to claim 10, wherein the step of optimizing mobility management is executed by applying the K-L (Kernighan and Lin) algorithm.

16. The method according to claim 10, wherein the step of optimizing mobility management is executed by applying the F-M (Fiduccia, C. M. and R. M. Mattheyses) algorithm.

17. The method according to claim 10, wherein the step of optimizing mobility management is executed by applying the genetic algorithm.

18. The method according to claim 10, wherein the step of optimizing mobility management is executed by applying the simulated annealing algorithm.

19. A method for constructing a mobility model of a wireless communication system, wherein the wireless communication system includes a plurality of location areas and each of the location area includes a plurality of cells, comprising:

collecting a plurality of network traffic statistic data of a plurality of desired items, wherein the desired items are related to the behavior of a plurality of subscribers of the wireless communication system; and

constructing the mobility model according to the network traffic statistic data, wherein the mobility model is for modeling the mobility behavior of the subscribers.

20. The method according to claim 19, wherein the step of collecting network traffic statistic data of the desired items further comprising pre-processing the network traffic statistic data of the desired items, wherein each of the desired items is generally corresponding to specific behavior of the subscribers.

21. The method according to claim 20, wherein the desired items at least include location update.

22. The method according to claim 20, wherein the desired items at least include handover.

23. The method according to claim 19, wherein the mobility model is constructed by solving simultaneous equations corresponding to the network traffic statistic data.

24. The method according to claim 19, wherein the mobility model is constructed by solving nonlinear mathematics corresponding to the network traffic statistic data.

25. The method according to claim 19, wherein the step of constructing the mobility model further comprising obtaining a preliminary estimate of the mobility model.

26. The method according to claim 19, wherein the step of constructing the mobility model further comprising verifying the accuracy of the mobility model.

27. The method according to claim 19, wherein the cells of the location areas at least include a border cell and an inner cell, wherein the mobility model at least includes a first submodel for modeling the behavior of the subscribers belonging to border cells and a second submodel for modeling the behavior of the subscribers belonging to inner cells.

28. A system for constructing a mobility model of a wireless communication system, comprising:

a processor, wherein the processor includes:

a mechanism for collecting a plurality of network traffic statistic data of a plurality of desired items, wherein the desired items are related to the behavior of a plurality of subscribers of the wireless communication system; and

a mechanism for constructing the mobility model according to the network traffic statistic data, wherein the mobility model is for modeling the mobility behavior of the subscribers; and

a storage medium coupling to the processor for storing the network traffic statistic data and data of the mobility model.

29. The system according to claim 28, wherein the mechanism for collecting network traffic statistic data of the desired items further comprising a mechanism for pre-processing the network traffic statistic data of the desired items, wherein each of the desired items is generally corresponding to specific behavior of the subscribers.

30. The system according to claim 29, wherein the desired items at least include location update.

31. The system according to claim 29, wherein the desired items at least include handover.

32. The system according to claim 28, wherein the mobility model is constructed by solving simultaneous equations corresponding to the network traffic statistic data.

33. The system according to claim 28, wherein the mobility model is constructed by solving nonlinear mathematics corresponding to the network traffic statistic data.

34. The system according to claim 28, wherein of the mechanism for constructing the mobility model further comprising a mechanism for obtaining a preliminary estimate of the mobility model.

35. The system according to claim 28, wherein of the mechanism for constructing the mobility model further comprising a mechanism for verifying the accuracy of the mobility model.

36. The system according to claim 28, wherein the cells of the location areas at least include a border cell and an inner cell, wherein the mobility model at least includes a first submodel for modeling the behavior of the subscribers belonging to border cells and a second submodel for modeling the behavior of the subscribers belonging to inner cells.

37. A system for constructing a mobility model for use in optimizing mobility management in a wireless communication system, comprising:

a processor, wherein the processor includes:

a mechanism for constructing the mobility model according to the network traffic statistic data, wherein the mobility model is for modeling the mobility behavior of the subscribers;

a mechanism for optimizing mobility management according to the mobility model; and

a storage medium coupling to the processor for storing the network traffic statistic data, data of the mobility model, and data of optimizing mobility management.

38. The system according to claim 37, wherein the mechanism for collecting network traffic statistic data of the desired items further comprising a mechanism for pre-processing the network traffic statistic data of the desired items, wherein each of the desired items is generally corresponding to specific behavior of the subscribers.

39. The system according to claim 38, wherein the desired items at least include location update.

40. The system according to claim 38, wherein the desired items at least include handover.

41. The system according to claim 37, wherein the mobility model is constructed by solving simultaneous equations corresponding to the network traffic statistic data.

42. The system according to claim 37, wherein the mobility model is constructed by solving nonlinear mathematics corresponding to the network traffic statistic data.

43. The system according to claim 37, wherein the mechanism for constructing the mobility model further comprising a mechanism for obtaining a preliminary estimate of the mobility model.

44. The system according to claim 37, wherein of the mechanism for constructing the mobility model further comprising a mechanism for verifying the accuracy of the mobility model.

45. The system according to claim 37, wherein the cells of the location areas at least include a border cell and an inner cell, wherein the mobility model at least includes a first submodel for modeling the behavior of the subscribers belonging to border cells and a second submodel for modeling the behavior of the subscribers belonging to inner cells.

46. The method according to claim 37, wherein the mechanism for optimizing mobility management is executed for adjusting a plurality of parameters of mobility management.

47. The method according to claim 46, wherein the parameters of mobility management at least include the scope of each location area.

48. The method according to claim 46, wherein the parameters of mobility management at least include the magnitude of a plurality of timers.

49. The method according to claim 46, wherein the parameters of mobility management at least include network topology and connectivity of different network elements.

50. The method according to claim 46, wherein the parameters of mobility management at least include paging parameters.

51. The method according to claim 46, wherein the mechanism for optimizing mobility management is executed by applying the K-L (Kernighan and Lin) algorithm.

52. The method according to claim 46, wherein the mechanism for optimizing mobility management is executed by applying the F-M (Fiduccia, C. M. and R. M. Mattheyses) algorithm.

53. The method according to claim 46, wherein of the mechanism for optimizing mobility management is executed by applying the genetic algorithm.

54. The method according to claim 46, wherein step of optimizing mobility management is executed by applying the simulated annealing algorithm.