FI113141B - Subscriber tracking procedure - Google Patents

Description

1, 113141

A method for tracking subscribers

Field of the Invention

The invention relates generally to tracking and subscribing mobile subscribers to value-added service applications.

Technology background

Mobile networks (PLMNs) contain information about the location of their subscribers. The geographical area of the network is made up of partially overlapping cells 10 such that each base station in the network has its own cell. The base station or other unit determines the column where the subscriber is currently located. The column is the area of the network.

Figure 1 shows an example of the shape of a column, where the minimum and maximum radius and angle define the shape. However, the shape of the column can vary between different grid systems and different positioning methods, the shape can be any geometric region.

Figure 2A illustrates key elements used to locate subscribers in a GSM network. The GSM network is one example of PLMN networks. As shown in the figure, the hierarchy of elements among each other is such that each base station controller (BSC) controls its own base stations and • the mobile services switching center (MSC) communicates with the base station controllers and the. ; with the Dyskeskus (GMLC). Tracking or positioning methods, which are not necessarily part of the GSM network, receive GSM subscriber location information for the connection,. through the center.

25 The following description applies to GSM networks, but the same principles can be applied to other mobile networks. The location data is updated when '' the subscriber calls or someone calls the subscriber. In both cases the network to determine the location of the subscriber to form a connection between the client and the other party. The location is a specific column within a given cell. The location data is also updated when the subscriber moves from one base station controller area to another base station controller area. In the GSM standard, the mobile station transmits a signal every two seconds. This allows the location of each subscriber to be continuously known, at least at the base station controller level.

GPS (Global Positioning System) provides another method for tracking mat-35 mobile subscribers. GPS is based on satellites owned by the United States Defense Organization.

2 113141 The GPS-based positioning method requires that each mobile subscriber has a GPS receiver that receives satellite signals and calculates the subscriber's location. This location is then transmitted in some way to the exchange that stores it in the location database.

5 GPS can provide much more accurate location information than a cellular network. Precise location information is required for many services, for example, when distributing an ad to subscribers within a limited area. For example, the advertisement may be sent to the subscribers concerned by SMS.

Known positioning methods are based on gathering information in a cellular network or information provided by GPS. Positioning methods require and receive a subscriber column. At the time of positioning, the probability that the subscriber is within the column is at its maximum value, but the probability decreases with time. Thus, the location method must constantly query the columns, which consumes the capacity and costs of the service providing the information in question.

Known positioning methods have the following disadvantages.

First, the methods lack parameters to customize their operation. They typically use two parameters, one containing the subscriber identifier or the subscriber group identifier and the other specifying how often columns are queried. Prior art positioning methods respond to the question "" What are the subscriber's current coordinates? "By returning the x and y coordinates. However, this issue is irrelevant in many location services.

Second, the methods fail to produce the exact position - >> t>; 25 estimates at a reasonable cost. For this reason it is too. . expensive to implement certain positioning service.

'···' Thirdly, the methods fail to utilize different column information related to the accuracy, size, and geometry of the columns. Networks may produce different column information, but prior art positioning methods are typically only capable of processing certain types of column information, which limits their usability.

. BRIEF SUMMARY OF THE INVENTION

The application positioning method is called a tracker. Tracker '·; · [35 calculates estimates for the subscriber's location. Each estimate contains the reliability value calculated by the coordinate point and the tracker. The coordinate point is a point on the trace coordinate system. Estimates, called matched estimates, form a particular subscriber trail.

The tracker rejects the question: '' What are the current coordinates of the subscriber? '. A better perspective is to trust after the subscriber and update the 5 tracks only when needed. The tracker updates the subscriber trace depending on the values of the accuracy and reliability parameters.

The precision parameter determines how accurately the positioning method should determine the position. The value of the precision parameter is given in meters, a low value means a small area.

10 The reliability parameter determines how reliably the positioning method should determine the position. The value of the reliability parameter is a percentage value from zero to 100. Reliability can also be viewed as a probability, that is, the subscriber's location should be correct with a certain probability ranging from zero to one.

15 In addition to accuracy and reliability parameters, the tracker follows six other parameters. The Value Added application sets values for all parameters. Those parameter values indicate the requirements of a particular user of the service application. The tracker consumes exactly the capacity it needs to meet customer requirements. Thus, the tracker saves the capacity of the network positioning system 20 as compared to prior art positioning methods.

The tracker is computationally more powerful than the prior art:, *. t positioning methods for two reasons.

First, the tracker only updates subscriber traces when it is ·: <· necessary. Each trace consists of estimates of the subscriber's location. The timer 25 sets a certain time in the timer and when that time is up, the timer counts down. a new estimate. The tracker then uses a trust table or trust function to determine whether or not a new column is needed. The tracker is able to evaluate which tracks can be updated later, which greatly reduces the load and cost of the tracking system.

In addition, multiple tracing processes that run simultaneously can use the same traces. Thus, while one process has updated the trace, other processes may use that trace.

. *. · * Secondly, the tracker is computationally efficient because it uses calculation formulas instead of iteration.

35 These formulas provide more accurate estimates of the location of subscribers than the prior art methods. The formulas are derived from an empirical study * * · • »4 113141 and are based on certain predetermined constants.

Characteristic features of the invention are described in the independent claims. The invention has the following objects.

5 The first goal is to improve the accuracy level to approximately 150 meters, making it very useful for service applications.

Another object is to improve capacity to a level where it can be used to track far more subscribers compared to prior art positioning methods.

10 The third objective is to provide tracing that can be optimized for a particular network.

The fourth objective is to provide traceability that is tolerant of changes in network structure or operation.

A fifth object is to provide positioning based on various types of location data such as accuracy, size, and column geometry.

List of figures

The invention will be further described with reference to the drawings, of which 20

Figure 1 shows an example of a column shape.

Figure 2A illustrates GSM network elements used in the positioning process.

: · I Figure 2B illustrates the operating environment of the tracker.

Figure 2C illustrates a simplified version of the tracker function.

·: ·: 25 Figure 3 illustrates location estimates for two heuristic positioning models.

·: ··· Figure 4A shows two columns and sixty coordinate points.

. ··. Figure 4B shows an example of a map consideration.

Figure 5 shows the unmatched estimate and the fitted estimate.

Fig. 6 illustrates the filters used by the tracker.

• · \,! Figure 7A illustrates the reliability values stored in the trust table.

'• y1 Figure 7B illustrates hits on the hit table.

ν ': Figure 8A shows the first part of the execution of the tracing task.

Figure 8B shows the latter part of the execution of the tracing task.

. Figure 9 illustrates the steps of determining the reliability table.

• 1 I •> • · · · 1 * »s 113141

DETAILED DESCRIPTION OF THE INVENTION

Figure 2B shows the operating environment of the tracker. The service application starts the trace task by sending a trace request to the tracker. The tracker contains parameters, calculation formulas with predefined constants, and a reliability table. During the trace task, the tracker sends column requests to the Gateway Mobile Location Center (GMLC) and receives columns from the gateway. The tracker uses columns to generate the report and then sends the report to the utility. There can be one or more reports per tracking task. For example, the report may include an adjusted 10 estimate. Or it could include information such as: '' subscriber 040-1234567 is in the area of interest. '

Figure 2C shows a simplified version of how the tracker works during the trace task. When a tracker starts a trace task, it receives requirements that determine, for example, the accuracy and reliability of the trace task. Requirements may also include a time interval specifying how long the tracker must wait before it calculates a new estimate.

Later, the following will be described: - Tracker parameters, - Calculation formulas and filters, 20 - Reliability table and how the tracker uses it to achieve 4 * •, j eg required reliability and accuracy, and: *, - Predefined Constants for calculation formulas and how - *: * · specific constants are obtained and how the confidence table is constructed.

.... First, some important terms are formally defined.

.... 25 The column is the location data obtained in collaboration with the tracker. . from the mobile network. Each column contains the geometry area and the time indicating the measurement time.

A set of columns consists of a number of columns associated with a particular subscriber. The tracker uses a set of columns and formulas to estimate the position.

· '· The preliminary estimate contains a coordinate point and a value called "reliability value." The tracker generates a set of coordinate points and calculates a set of reliability values for each coordinate point. If the column set contains k columns and the tracker generates m * n coordinate points, the tracker calculates a confidence value of «> · I · r • * 6 113141 m * n * k, thus producing a m * n * k preliminary estimate.

The confidence sum is the sum of the confidence values of the preliminary estimates related to the same coordinate point. The tracker calculates a confidence level of 5 lbs for each coordinate point it generates, so the tracker produces a confidence sum of m * n. One is the highest amount of confidence.

The unmatched estimate contains the coordinate point and the maximum reliability sum.

The fitted estimate is the final position estimate of the tracker. The matched estimate includes the time at which the matched estimate is calculated. In addition, the fitted estimate contains the same reliability value as the unadjusted estimate, i.e., the maximum reliability sum. However, the fitted estimate usually contains a different coordinate point than the unmatched estimate.

The trace consists of adjusted estimates that the tracker has calculated for 15 subscribers. The trace may be the actual path that the subscriber has passed or not.

Accuracy is the parameter value set by the service application that the tracker uses. Accuracy can be viewed as a radius defining a circle centered on a fitted estimate.

20 Reliability is another parameter value set by the service application. Reliability describes the quality of the fitted estimate. High reliability. · Indicates a high probability that the fitted estimate is accurate to the desired accuracy. For example, the probability could be 0.95 and accuracy. could be 50 m, ie the probability that the subscriber is within 50 m

• I

The 25 fitted estimates are 0.95.

A reliability table is a table structure made up of cells where '·> ·' each contains a reliability value. Each cell with its reliability value is associated with a certain accuracy and a certain amount of reliability.

The reliability function results in a reliability value as a function of precision and confidence sum. For example, if the resolution is 50 m and: '·. · Reliability is 1.2, reliability could be 0.7.

The reliability chart and the trust function are for the same application. For example, when the service application has set a precision parameter of 50 m to a value of 0.7, a reliability-35 table or reliability function gives a reliability sum that should be at least 1.2. Otherwise, the adjusted estimate will not meet the required accuracy and reliability • · * 7 113141. In this case, the tracker asks for and receives a new column from the mobile network, updates the set of columns, and calculates a new adjusted estimate, which should fulfill the required accuracy and reliability.

The tracker parameters adjust the tracker function so that the tracker can meet customer requirements. By comparing the tracker with prior art positioning methods, the tracker saves the capacity of the system providing the column information and at the same time is able to compute better position estimates than prior art methods.

The precision parameter determines how accurately the tracker should determine the location and the reliability parameter determines how reliably the tracker should report the location. The trace follows the following parameters: - Trace task The ID parameter identifies the task started by the service application. The result will be the result of the subscriber.

- Subscriber ID / Group ID is another parameter.

15 - Subscriber ID identifies one subscriber and group ID identifies a set ti wide.

- The accuracy parameter defines the accuracy of the position in meters.

- The reliability parameter specifies the location-related reliability as a percentage.

20 - Minimum Interval parameter specifies, in seconds, how often the tracker must at least request a column.

·. / - The Maximum Interval parameter specifies in seconds how often ....: the trace may not request more than a column.

- The area of interest is an optional parameter that defines a rectangular area with a fixed * «25. The parameter consists of two • »· t ·. / coordinate points that define opposite '···' angles of a rectangular area.

- The History parameter determines whether or not the matched estimates are stored.

Calculations and filters are a very important part of the invention. Tracker: 'v can use the cellular information provided by the cellular network in such a way that it can locate subscribers almost as accurately as the positioning method that uses the cellular information provided by the GPS system.

The tracker uses filters to calculate fitted Estimates.

35 Each filter contains one or more calculation formulas.

* »* • I» |

* · I

»» · 8 113141 The tracker is quite complex to operate, so at first two simple heuristic models will be described.

Figure 3 shows an example of an estimate of a GPS system regarding the location of a subscriber and the actual path of the subscriber. GPS estimates are derived from GPS 5 (Global Positioning System) and are marked with a dash. GPS estimates are coordinate points: (3,1), (5,1), (5,3), and (3,3). The dashed line represents the actual path of the subscriber.

According to the first model, the subscriber is assumed to be located in the center of the GPS Estimates at a point (4,2) as shown in Figure 3. The first 10 models can be improved by taking into account the time since the last GPS estimate was received. This time is denoted by t \. If (3,3) is shorter, it is obvious that the subscriber is located closer to the point (3,3) than to the point (4,2), since (3,3) is the latest GPS estimate received.

15 According to another model, the subscriber will move in the same direction and at the same speed as before. It can be assumed that the subscriber has moved from point (5.3) to point (3.3) in time tm. Thus, the subscriber of Figure 3 should be at point (1,3) when the time tm has elapsed since the last GPS estimate.

As can be seen, these two models may produce well 20 different location estimates, such as points (4.2) and (1.3) above.

: When the tracker calculates mismatched Estimates, its function in some way resembles the first model, and when the tracker calculates the matched, ... · ground, its action is in some way reminiscent of the second model. There are also differences between the performance of the tracker and these models. First, the tracker does not use GPS Estimates, but columns derived from the cellular network. Second, the calculation formulas used by the tracker are unknown in the prior art.

«'···' Denote predefined constants A, B, C, D and N. The calculation formulas contain predefined constants A-D: - A is the probability dilution factor within the column.

30 - B is the probability dilution factor outside the column.

C is the factor for the velocity vector, where the previous estimate and: '' * the latest unmatched tracker estimate define the velocity vector.

I t t. The V-Don factor is the so-called. for probability dissipation.

\ y A predefined constant N specifies the default size of the column set. The predefined constants A-D and N are obtained as a result of a generic algorithm or some optimization method.

* _ 113141 δ

In addition, the formulas include the following variables: - The distance variable expresses the shortest path from the coordinate point to the outer boundary of the nearest column.

- The elapsed time variable indicates the elapsed time since the time-5 was received.

The tracker generates a set of coordinate points. Those coordinate points form an area that covers the columns in the set. The tracker uses the predefined constants and variables described above in the following calculation formulas when calculating the reliability values of the preliminary estimates.

10 The original reliability is the standard reliability divided by the area of the column.

If the generated coordinate point is inside a column, the tracker uses the first calculation formula: 15 Reliability Value = Original Reliability * Accumulated Time.

If the generated coordinate point is located outside the column, the tracker uses another calculation formula: readability = initial reliability * B run time * elapsed time * (D / distance2) 20 Figure 4A shows two columns and 60 tracker generated coordinate points. The coordinate points form a rectangular region covering the two columns. Within each column are four I * coordinate points. The tracker uses the first calculation formula when the coordinate point is inside the column. Otherwise, the tracker uses a different calculation formula. Figure 4A also shows an example of a distance variable. The distance variable • * * »is set to the value that is the shortest distance from the coordinate point to the outer boundary of the nearest column.

The first or second formula gives the result of preliminary estimates. Looking at 4A and its 60 coordinate points, the tracker calculates 30 initials for four columns using the first formula and; * ', * 56 a preliminary estimate using another formula. The tracker then counts on the column: '' 2 four preliminary estimates using the first formula and 56 preliminary ones. y estimate using another formula.

Finally, the tracker adds the initial, ·; · <35 estimates for each of the 60 coordinate points and selects the maximum reliability sum. Specifically, the s-tracker adds the reliability values calculated for columns 1 and 2, that is, two reliability values for each coordinate point.

Thus, the tracker produces an unmatched estimate including the coordinate point and the maximum reliability sum.

5 Next, the tracker calculates the fitted estimate using the third formula. The third formula takes into account a predetermined constant C, the direction of the subscriber, and the time ratio dt.

The direction is obtained by constructing a vector from the coordinate point of the previous estimate to the coordinate point of the unmatched estimate.

For calculating the time ratio dt, 1) the time elapsed between the previous estimate and the unadjusted estimate and 2) the time elapsed between the previous estimate and its estimate preceding the preceding estimate. The first time is denoted by U and the second time is denoted by t2. The time ratio dt is calculated as follows: 15 dt = U 112

Figure 4B illustrates an example of how a tracker may observe a map. The tracker looks at the subscriber's speed. In addition, the tracker could look at whether the subscriber is located at sea or on land. For example, when the subscriber is on land and the subscriber's speed is more than 30 km / h but less than 20 230 km / h, the tracker could deduce that the subscriber is using some kind of ground transport. . woe that uses you. The tracker then uses the map and generates a coordinate; points only to the area where land vehicles can move. Of course, off-road vehicles, snowmobiles and very fast trains are exceptions: 'they can move off-road or they can travel faster than 230 25 km / h. However, the tracker could deduce the vehicle with a fairly high probability * * '' and use the map if necessary.

Figure 5 shows the same two columns as Figure 4A. In this example, 60 coordinate points are omitted. Instead, the previous estimate, the unmatched estimate, and the fitted estimate are shown in the figure.

30 The previous estimate is denoted (xp, yp), the unmatched estimate ...: denoted by (xn, yn), the fitted estimate is denoted by (xa, ya) and the estimate preceding the previous estimate is denoted by (xp_i, yp.i). The fitted estimate is obtained from the third formula as follows: • '.0 xa = xn + C * (xp-xp-i) 35 ya = yn + C * (yp-yp-i) * df • · 11 113141

Figure 6 illustrates the filters used by the tracker. Filters are an alternative to the aforementioned calculation charts. Filters contain formulas that produce even better Estimates than those formulas. The filter 2 contains improved versions of the first and second calculation formulas 5. The filter 4 contains an improved version of the third calculation formula. Figure 6 illustrates a case where GPS points were obtained during the tracker optimization step. This optimization step is performed prior to the use of the tracker. The optimization phase description has predefined constants. If GPS points or corresponding coordinate points are not available for the 10 column sets used in filter 1, then the tracker does not use filter 1. Instead, filter 2 receives the latest column and uses the column set instead of the circle estimate set (see Figure 6).

The filter receives an estimate as input. This estimate can be implemented, for example, as a data structure containing the following information: Geometry, Time, Reliability. The geometry can be a column, a circle, or a coordinate point.

FILTER 1 is a GPS correction filter.

The feed geometry is a column.

The filter maintains a set of columns with a maximum size of predefined constant Ni. The filter receives such GPS points as information. . a strain inside at least one set of columns. From; GPS points form a set of GPS points. The filter then forms a circle centered on the midpoint of the GPS set. The radius of the formed * 'circle is such that F percent of the GPS point within the set of GPS points lies inside the formed circle. This F is a '' * '' predefined constant.

\ Output geometry is a point if filter 1 failed to find GPS points for a set of columns. Otherwise, the geometry of the printout is a circle. The time associated with the printout is the same as in the most recent column. The reliability of printout 30 has not yet been calculated.

. ·. : FILTER 2 is a conversion filter.

The feed geometry is column or printout of filter 1.

The filter maintains a set of circle estimates with a maximum size predefined constant N2. The filter passes through either 35 sets of circular estimates or a set of columns. Because circle estimates can also be viewed as columns, the terms "" column "and" "set of columns" can be used. The filter calculates the confidence sum for each column center, that is, 12 113141 calculates the confidence sum for each column center, or center of gravity, as follows:

Let p be the midpoint of the column.

For each column set column 5 {

The initial readability is the confidence almlava = —c — rr region 'where c is the confidence constant associated with each column and' region 'is the region of the column and G is constant 10 If region <1 then trust almlara = c

Reliability at the present time is confidence = lu0- a, ^ ™ 2- - - where H and / are constants, At is the time difference between the time stamp of the column and the current, and d is the distance between p and the midpoint of the column.

}

Now, the confidence totals you trust are calculated for each column Finally, the filter calculates the reliability weighted by the mean midpoint of the columns * * * n · J] trust, * p, 0 ·. 20 p = ^ - *: * ·· ^ trust / = 1 where the points pi are the center of the columns and n is the number of entries in the buffer, n <N2. In these formulas, the direction vector p; · is used to denote a point.

The geometry of the output is a point. The time associated with the printout is now-25 moments and the printout reliability is calculated by the filter 2.

FILTER 3 is an error correction filter.

The geometry of the feed is a point and the reliability of the feed is calculated by the filter 2.

The filter generates a statistical distribution of distances, where the distances are the distances between the inputs of this filter and the associated GPS points, e.g., errors in the inputs relative to the GPS locations.

»·. · '· During the tracker optimization step, the filter receives GPS points as input ♦» I · i3 113141. Each distribution distance value contains the number of input and GPS point pairs where the distance is the same as that distance value.

The filter is a constant describing the statistical distribution of input point errors. This constant distribution is created during the optimization phase. The timestamp of the last 5 feeds has been stored.

The filter maintains a sum distribution having the same scale as the constant distribution. The sum distribution contains the number of coordinate values. The standard distribution is added to the sum distribution when a new feed is received. The constant distribution is added to the sum distribution such that the number of zero error 10 of the sum distribution is added to the number of input coordinates and to the other numbers, respectively.

The old values of the sum distribution are multiplied by the dilution factor before the standard distribution is added to the new feed. The dilution factor is Jh ', where J is constant and At is the time difference between the input and the previous input. 15 The result point is the weighted average of the sum of the coordinate points distribution P result h Σ / (Λ) i = 1 where n is the sum of the distribution, which is a constant,:; Pi are the coordinate points of the sum distribution and 20 f (Pi) are the sum of the distribution of the coordinate values Pi.

The geometry of the output is a point. Printout time is current and printout reliability is calculated by filter 2.

: '* FILTER 4 is a motion adapter filter.

25 The feed geometry is a point and the reliability of the output is calculated in filter 2.

The tracker maintains a buffer for feeds. Maximum buffer size. >. · Is a predefined constant N3. The feed is not added to the buffer before counting.

t ·

The filter calculates the new position as follows: Initially, it calculates no- :. 30 sum of loop vectors from buffer columns to input point:: ·: · v = fv = yh ^ iZL · * »* sum / i input, / / j *,: i = l / = 1 * *» t 14 113141 where Ar, is the time difference between the coordinate points pjnpill and pt and n is the number of entries in the buffer, n <N3.

The filter then divides the sum of the velocity vector over the buffer length, yielding a summer velocity v of 5 v _ _ «« «

The result point is the sum of the input points and the mean velocity multiplied by a predetermined constant K

P result = P * K

j. rcMiit * · inpuI {n> i!

The geometry of the output is a point. The time of printout is current and the reliability of the input-10 tea is calculated by the filter 2.

FILTER 5 is a hopping reduction filter.

The geometry of the feed is a point and the reliability of the feed is calculated by the filter 2.

The filter maintains a buffer for feeds. It adds an entry point to the buffer 15 after counting. The maximum buffer size is N4

Initially, the filter drops the average speed to the buffer point. The velocity is calculated from two consecutive points piA and pj ft ft n - n

V .Tv-LV P '-PtL

sum / 41 /, / - 11 / j.

i ~ 2 i ~ 2 Δί, • where Δ /, οη is the time difference n between the coordinate points pi and j5, _; \; 20 is the number of inputs in the buffer, n <N4.

• ♦ '* ·' 'Then the average speed is calculated: v = n - 1

For the final step, At is needed, which is the time difference between the input point and the previous input point

'· I * 1 Pinput Pinput-X

V = v = —-- 1 1 At 25 If v> L * vOT, g where L is the constant associated with the filter, then ll * v. V If v> M where M is a constant associated with a filter, then - -. [m

y .; v = v * iT

30 ♦ · · is 113141

The conversion point is given by the formula:

Prc.su/, = P input- \ + V * M

The geometry of the output is a point. Printout time is current and printout reliability is calculated by filter 2.

5 The trust table and the trust function are alternatives to each other. Both are used to determine if a new column is needed to meet customer requirements.

Figure 7A illustrates a reliability table filled with confidence values. The trust table is static, meaning the tracker uses it but does not change its contents. The accuracy and reliability parameters control the operation of the tracker. For example, the value of the accuracy parameter could be at least 250 m and the value of the reliability parameter could be at least 0.7. In this case, the tracker would use a line in the trust table labeled "250 m". As can be seen, the column of the reliability table labeled "> 1.2" contains a confidence value of 0.71. Therefore, the tracker concludes that the confidence sum must be greater than 1.2 in order to satisfy customer requirements obtained as parameter values.

Figure 7B illustrates a hit-filled hit table. A hit table is needed when creating a reliability table. Later, it will be described how the availability table and hit table are constructed.

8A and 8B illustrate how a tracker performs trace tasks.

; More specifically, the figures illustrate how the tracking process performs a tracking task for a particular subscriber. Several tracing processes are running at the same time. The tracking task consists of subtasks marked with numbers 25,801-818.

. , (801) Initially, the tracker checks if the column poll flag is on. This flag is associated with a specific post. If a trace process has already requested a column, the next subtask is 808, ie the trace skips.

(802) The tracker checks to see if the track is out of range until f00 30 and that it is not yet t00. An offline area is an area that does not:, * ·· interest a particular subscriber. The tracker can estimate the time t00, which determines when: "" the subscriber is coming to an area of interest.

(803) The tracker retrieves new columns in a column set if there are any new columns in the database. The trace processes can take advantage of the same columns, so each trace process first checks to see if another trace process has already requested, received, and stored a new column in the database.

16 113141 (804) The tracker uses a set of columns, calculates a fitted estimate for the subscriber location, and checks the reliability of that estimate. The fitted estimate contains the confidence sum.

(805) The tracker uses the reliability estimate of the fitted estimate and the accuracy parameter set by the service application. In this way, the tracer receives a specific trust table cell. The tracker checks the cell to see if the reliability is below the requested reliability. More specifically, the tracker compares that reliability to the value of the reliability parameter set by the service application.

10 (806) The tracker checks to see if the last column is less than the "minimum interval" old. The minimum time interval is a parameter value set by the service application.

(807) The tracker checks to see if the finisher column is more than the "maximum interval" old. The maximum time interval is a parameter value set by the service application.

(808) Partial tasks 801, 802, 805, 806, or 807 may cause you to skip updating the column set.

(809) Otherwise, the tracker sets the column poll flag.

(810) The tracker requests a new column from the GMLC, obtains a new 20 column and stores it in a database, and adds a new column to the column set.

•; * However, it is possible that the trace or trace process is interrupted if its timer is triggered. Sometimes it can take so long to get a new column to trigger that timer.

(811) If the tracker was not interrupted in subtask 801, it will disable 25 column poll flags.

(812) The column set is now updated and the tracker calculates a new matched estimate for the subscriber location and checks the reliability of that estimate.

(813) The tracker checks if the history parameter is on. The value of the historypa-30 rameter is set by the service application.

(814) If the history parameter is on, the tracker stores a matched estimate.

. (815) The tracker checks whether the user is outside the area of interest. The area of interest is an optional: ··· 35 parameter set by the service application. The location of the subscriber is the coordinate of the matched estimate. Thus, the tracker * 113 171 141 checks if the coordinate point is outside the rectangle of interest.

(816) If it is located outside the area of interest, the tracker will set the Out of Range flag.

5 (817) The trace calculates the time too, when the trace may be over the area of interest and records the time too This calculation is based on the third formula described above.

(818) The Tracker sets the Out of Range flag.

(819) The tracker reports the subscriber's location, if necessary. However, if 10 out of area flags are on, reporting is not required.

The predefined constants are obtained during the tracker optimization step. The following description applies to the predefined constants A-D and N. However, the same principles can be used to find the best values for the predefined constants F-M and N1 to N4. The values of the predefined constants depend on the source system and its characteristics. A source system, such as a cellular network, provides location data.

Figure 9 shows the steps of the initialization step of the tracker. The Initialisoin step produces the required pre-defined constants and a reliability table.

In step 1, a few auxiliary subscribers are required to carry 20 source system terminals and a GPS system terminal. These subscribers move within a certain geographical area and their location information is simultaneously collected from the source system and the GPS ..... system. The first step takes a certain amount of time, for example one hour.

In the second step, the columns collected from the source system and the coordinate points collected from the GPS system are converted to a coordinate system, such as the Uniform Coordinate System (UCS). Pairs of columns and coordinate points are paired so that each column and coordinate point contains a user ID. For example, the user-30 identifier is MSISDN (Mobile Subscriber Integrated Services Digital or Network). Thus, each pair contains the following information: '' - User ID, Time, UCS Coordinates, V - User ID, Time, UCS Coordinates of the coordinate point.

'; '; The first "time" is the time of measurement of the column and the second "... 35" time "is the time of measurement of the coordinate point. The resultant pairs in step 2 are stored in a database.

»» 18 113141

In step 3, the stored pairs in the database are viewed as a time series. Each time series is processed such that times are replaced by discrete time intervals. The discrete time interval can be, for example, 15 seconds. For example, if step 1 lasted from 9.00am to 10.00am, then the time intervals from 9.00: 00-5: 009: 15 is replaced by a time interval of 0-15 and the time between 9.00: 16 and 9.00: 30 is replaced by a time interval of 16-30, etc.

Paired time series are replaced by time series that are made up of time intervals. Each time interval contains the following information: 10 - User ID (which indicates the time series) - Time interval value - UCS coordinates of the column - UCS coordinates of the coordinate point.

Step 4 looks at time-interval sets of time intervals. Time series that meet the following two criteria are selected for the next step. The first criterion is that the time series should be long enough, ie each time series should contain a sufficient number of pairs. Another criterion is that each time series should be sufficiently homogeneous, ie the time interval between two pairs should not exceed a certain time.

Step 5 looks for the best values for the predefined constants. The search may be based on a known genetic algorithm. Genetic. \; the operating principles of the algorithm are illustrated by presenting a simple brutal algorithm. This algorithm checks all combinations of predefined constants A-D and N. Constants A-D are used in the counting diagrams of the tracker * 25, and constants N determine the default size of the column set.

The values for constants A and B could be, for example: 0.90, 0.91, 0.92,: ... · 0.93 ..... 0.99. Thus, both have ten possible values. Also, constants C and D could have ten possible values. In addition, the constant N could have ten possible values from 2 to 11. Thus, the constants A-30 D and N could have 105 or 100000 value combinations.

A genetic algorithm, or here a brutal algorithm, calculates a '' trace error ''. Tracker error is the difference between the adjusted estimate and the GPS position. The fitted estimates are calculated using the above described calculation diagrams. The brutal * · algorithm uses the variable "SSTE" to sum up squares of tracker errors.

35 The brutal computation algorithm contains the following FOR loops.

• · I

1β 113141

FOR each value of N

{

Set 0 to SSTE.

FOR each constant A-D values of 5 {FOR each time series {

If the time series does not include at least N time intervals, skip the time series.

10 FOR for each time series {

Calculate the fitted estimate using constant A-D values and a set of N columns.

15 Calculate the Tracker Error.

Add the trace error square to SSTE.

}}

Write down the values for constants A-D and N, as well as the value for SSTE for 20 in the result file.

f>>

The above result file is formed from records containing values for constants A-D and N and the value of variable SSTE. The record with the lowest SSTE value of the variable is selected and the predefined Constants A_D and 'N are obtained from that record.

In step 6, time series that are sufficiently long and are sufficiently uniform and predefined The constants A-D and N are used in the algorithm 30 ΉΙΤ ”(see Figure 9). A hit table can be viewed as a data structure that. . . * contains a counter for each cell. Initially, the counters are set to zero, meaning there are no hits. The algorithm goes through time series and increments the counter value '·' · and calls a function called ΉΙΤ '.

• * 20 113141 FOR each time series {

If the time series does not include at least N time intervals, skip the time series.

5 FOR for each time series {

Calculate the fitted estimate using a set of N columns and constant A-D values.

Calculate tracker error.

10 Call the HIT function using the matched barrier's reliability sum and the tracer error as parameters.

}} 15 The HIT function deduces a hit table column based on that reliability sum parameter, and the HIT function deduces a hit table row based on the tracker error parameter. Thus, the parameters of the HIT function determine which cell, i.e. which counter, the HIT function should increment by one.

For example, the reliability sum parameter could have a value of 1.2 and the trace error parameter could have a value of 75. In this case, the HIT function selects a column labeled "> 1.2" and a row labeled "100 m" and. . : "250 m" (see Figure 7B). The HIT function concludes that the counters for the cells at the intersection of this column and ιι = those rows need to be incremented by one. In Figure 7B, those counters have values of 159 and 710.

'25 In step 7, the reliability table is formed using a hit-filled hit table. The cell in the lower right corner of the hit table contains the highest hit value. In Figure 7B, the maximum hit value is 710. The number of time series for each column in the hit table is known. Each confidence value is obtained by dividing the number of hits by that number of time series. For example, in Figure 7B, the last column of the hit table has a thousand time series. Thus, the confidence values shown in Figure 7A are obtained by dividing the hits shown in Figure 7B by 1000.

> I * »t *» »* *« · · I ·

Claims (44)

1 At where At is the time difference between the input point pinpul and the previous input point Pinputs and 5 when v> L * vmg, where L is a predefined constant, correcting the rate with JL * v arg - V, and when v> M, where M is a predefined constant correcting the velocity with the formula "A _ _ [M 10 v = v * / -, V v to obtain the score point pKSUII with the formula P result = P", put-1 + V * At, add the presuit to the second set of points and print the presul. 1. A method for locating objects in a system, the method comprising: receiving an object position determining column, providing location estimates, determining each location estimate where the source system object is at a particular time, the method comprising: maintaining a plurality of objects for each object to be located; - 10, calculating the position estimate using the set of columns, examining whether the position estimate satisfies the requirements and only if the position estimate fails the requirements, requesting a new column from the source system, 15 receiving a new column, adding the received column to Method according to claim 1, characterized in that a position estimate is transmitted. The method according to claim 1, characterized in that: That the requirements include a reliability value defining the probability of how reliable the position estimate must be. «* * · A method according to claim 1, characterized in that the requirements include a precision value determining the distance how:. the exact position estimate must be. A method according to claim 1, characterized in that the requirements include a time interval defining how often at least the method must ask for a column. The method of claim 1, characterized in that the requirements include a second time interval defining how often the method may at most request a column. ♦> A method according to claim 1, characterized in that, when calculating a location estimate, the method uses calculation formulas including predefined constants and, during the optimization step preceding the implementation of the method, looks for the values of the predefined constants in the method adapt. to the source system. 8. A method for locating source system objects, the method comprising the steps of: receiving a position determining column, providing location estimates, determining each location estimate of where the source system object is at a given time, 10 with predefined constants, and during the optimization step preceding the implementation of the method, the values of the predefined constants are searched to adapt the method to the source system. Method according to claim 1 or 8, characterized in that the source system is a telecommunications network and the object is a subscriber. Method according to claim 6 or 8, characterized in that, in order to find predefined constants values:, · using moving objects in the source system, computing position estimates for said objects by changing the predefined constants values to relatively accurate position data for said objects, reference system, defines errors between location estimates and location data to get · '. , resulting in values for predefined constants that reach a minimum of errors. The method according to claim 6, characterized in that the process optimization step comprises the steps:. collecting location data from the source system and the refe, ··· licensing system simultaneously, with both location data containing user IDs'! . and time stamps, »i v. time series are formed from source system location data and ref1,,,, 35, the time system-measured location data by replacing time stamps, ··· 1 at time intervals, and pre-combining i» 1 i 1 «» 23 113141 the possible values of the defined constants, and for each value combination, an algorithm that calculates location estimates using the formulas and calculates errors between the location estimates and the locations measured by the reference system 5, and during the execution, and the value of that variable in memory, and after execution, 10 reads the memory, and looks for the value combination that has the smallest value of the variable, thus obtaining predefined constants values. Method according to claim 12, characterized in that the algorithm is a genetic algorithm. 13. A method according to claim 10 or 11, characterized in that the method optimization step includes further steps: obtaining reference points from the location data measured by the reference system, and storing the reference points in a database. A method according to claim 10 or 11, known; . it is understood that the reference system is a GPS system. Method according to claim 14, characterized in that the reference points are locations measured by the GPS system. A method according to claim 11, characterized in that the method optimization step includes further steps: using a cell hit table such that each \. the cell targets a given table value for a given sum of values and a given table value for accuracy, and calculating position estimates for each time series using the formulas with predefined constants values obtained, and calculating an error between each computed position estimate and the reference position, ·· * · Obtains a certain table value for the accuracy of the error in question, v. Obtains a certain table value for the confidence sum from that computational »I · 35 ta position estimate, and •». * ·· obtains the specific cell to which these table values refer, * · »♦ *» * * »· 113,141 adds a counter for that cell by one, and adds another counter by one, where the second counter is associated with a table value of the confidence sum. The method of claim 16, characterized in that the optimization step includes further steps: dividing each hit table cell by a value stored in the cell counter with a counter value associated with the same confidence sum table value as each cell of the division table, has the same logical structure as the hit table and each cell in the trust table is intended to store the trust value. 18. The method of claim 17, wherein the step of optimizing the method includes further steps: using a trusted table filled with tabular values and cells filled with trusted values, constructing a function by considering a tabular value associated with a given accuracy and a tabular value associated with a given confidence sum as the output of the 20 value of the reliability value. . . The method of claim 17, characterized by *; '; whether 'Location Estimates Meets Requirements' is used to determine; reliability table as follows: * 'retrieves table value for accuracy from precision value that defines': · 25 distance how accurate position estimate should be, '' * · retrieves table value for reliability sum from location estimate reliability sum, retrieves specific cell of the reliability table reliability value to a confidence value of 30, which determines the likelihood of how reliable the location barrier is. must be, thus, resulting in a position estimate of * *!! .. 'meeting the accuracy and reliability requirements when the former reliability value reaches the latter reliability value. : 20. The method of claim 18, characterized by: "'': 35 that is used to determine if the location estimate is satisfied. / function as follows: 25 113141 obtains the precision input value from the precision value that defines the distance how accurate the position estimate should be, obtains the reliability sum input value from the position estimate reliability sum, and uses a function with those input values, 5 gives a given reliability value , how reliable the position estimate must be, resulting in the position estimate meeting the requirements for accuracy and reliability when the former reliability value reaches the last-mentioned reliability value. A method according to claim 15, characterized in that, when calculating a location estimate, the method performs an estimate calculation comprising the steps of: using a set of columns of size Λ / ι, where A / ι is a predetermined constant; within a column set, and when those reference points are obtained, a circle is created with the center of the reference points averaging 20 points and the radius of the circle such that F percent of the reference points remain. . inside a circle where F is a predefined constant and *; ; prints the item with a timestamp, the item being a circle. 21 113141 The method of claim 21, characterized in that the estimate calculation includes the additional step of: printing an element with a timestamp, the element being a circle, when it is possible to retrieve said reference points. A method according to claim 21 or 22, characterized in that the estimate calculation includes an additional step: feeding an element, yielding a set of elements of size N2, wherein N2 is predefined. calculated constant, add the element to the set of elements, calculate the reliability sums for the elements of the set so that the initial value for each element is given by the formula * * * · C * · · · ^ preliminary ~] (f 'range »» · • »26 113141 where c is the constant reliability, G is the predefined constant, the region is the region of the item, and when the area <1, credit scale = c, the reliability at this time is given by formula 5 to = trust ^ vc, _ * d 'where H and / and d is the distance between the midpoint of the element and the midpoint of the columns of the array, calculate a point called the midpoint p of the element of the array p using the formula: Y p you created, * p, p = ± 1 = 1 where the bullet is the confidence sum of the elements in the set, pi is the center of the elements and n is the number of elements in the set, where n <N2 and 15 print the average center point p with a timestamp. A method according to claim 21, 22 or 23, characterized in that the estimate calculation includes the additional steps::: '. let's enter a point, i f i. '. ('obtains a statistical constant distribution related to point errors, 20 obtains a sum distribution including the number of each coordinate point in the sum distribution,. multiplies each coordinate point of the sum distribution by the dilution factor. between points, 25 is added a constant distribution to the sum distribution such that the number of 0 errors of the constant distribution is added to the coordinate numbers and to the other numbers, v is the point /? rew; which is the sum of the distribution: '*: man coordinate points / (Α) * Ρ;.:: 30 Presul, = - ΣΑΡ,). · '' '= 1 i * 113141 where n is the size of the sum distribution, the coordinate points of the sum distribution and f (pt) are the coordinates of the /?,. number values are printed on the presul. A method according to claim 21, 22, 23 or 24, characterized in that the estimation includes an additional step: supplying a point bundle, obtaining a first set of points of size Λ / 3, where Λ / 3 is a predefined constant, calculating the velocity vector sum vmm the first set of pis-10 of you to phipul using the formula: - V- '- _ P input _ P1 ^ sum / 1 ^ input, i /. . '; = 1 (= 1 where At, · is the time difference between pinpul and pt, where p (is the point of the first set of points and where n is the number of elements in the set of elements such that n <Nit where A / 3 determines the size of the first set of elements average velocity vmj, from the formula:%, = - and, and obtaining the point pnsuhl from the formula B = P + v * K, where K is a predetermined constant, Γ riSUIH i input tn'g .. add a point to the first set of pnsuln and print pKSuUx. A method according to claim 21, 22, 23, 24 or 25, characterized in that the estimate calculation includes the further steps of: 'entering a point, obtaining a second set of points of size N4, where Λ / 4 is a predetermined constant, calculate the sum of two consecutive points and p, by the formula: n λ n - nv = Yv -, - sum / a | i, i ~ 11 / j a. . \: / = 2 / = 2 r · where Att is the time difference between the points pj and n is the second pis-'1. The number of points in the 30 sets and n <NA, v. Calculate the average velocity Varg by formula l V, ► y - —1 ^ n- \ t · 1 »» · 28 113141 calculate the velocity by formula I— | P input Pinput-1 V = - --—, A method according to claim 10 or 11, characterized in that, for computing each location estimate, the method performs an estimate calculation comprising the steps of: generating coordinate points such that they form a set of points> 1. *. T covering the area of the column set; within a set of 20 columns, calculate the reliability value such that the reliability value = original reliability * Accumulated time, »II» where the original reliability is the reliability constant divided by the size of the columns · · tan area, where A is a predefined constant and the time since the measurement. A method according to claim 27, characterized in that, when a point of a set of points lies within a column of a set of cells, • '·. 'calculate the confidence value using the formula •;' the confidence value = original reliability *, Y 30 yrs time ^ (q / distance2) »* 'γ Where is the original confidence value the confidence constant divided by the column:« Y tan size and B and D are predefined constants where elapsed time is »ri *>» t »» i »*. I I 29 113141 Time elapsed since the time of measurement of the column and where the distance is the shortest distance from the generated point to the outer limit of the column. A method according to claim 27 or 28, characterized in that the estimate calculation includes the following additional steps: a preliminary estimate is obtained for each point in the set and for each column in the set, wherein each preliminary estimate is a confidence value calculated from the generated point; to the associated estate sum, resulting in an estimate sum for each point, 10 is searched for the point with the largest estimate sum, resulting in an unmatched estimate consisting of the coordinate point and the largest estimate sum. A method according to claim 29, characterized in that the estimation computation includes further steps: obtaining a direction by constructing a vector from the previous matched coordinate point of the estimate (xp, yp) to the coordinate point (xn, yn) of the unmatched estimate; U is the time from the previous fitted estimate to the unmatched: the time elapsed to the estimate and t2 is the time elapsed from the previous fitted estimate to the preceding · · · estimate (xp_i, yp_i), | J is calculated by coordinate point (xa, ya) of the fitted estimate with the formula Xa = Xn + C * (Xp-Xp-1) * dt, 25 ya = yn + C * (yp-yp.i) * df, where C is a predefined constant . 31. The method of claim 27, wherein the generating of coordinate points is performed according to the steps of forming a region covering a plurality of columns: a map covering the geographical area of the plurality of columns is used; the object is located on a particular type of vehicle: '· moving on the map in an area intended for that type of vehicle 35 the coordinate points located on that particular vehicle are deleted. · ·. outside the map area. 30, 113141 Method according to claim 24, 25 or 26, characterized in that additional steps are performed before printing the presull point: using a map of the geographical area of the set of columns, calculating the speed of the object and when speed exceeds a certain threshold value in the area intended for that type of vehicle, a point is located within the area intended for that type of vehicle and the distance to prmill is the shortest and 10 is set to the same coordinates as that point. A method according to claim 23, characterized in that the following additional steps are performed before printing the mean midpoint p: 15 using a map covering the geographical area of the set of columns, calculating object velocity when velocity exceeds a certain threshold, concluding that object in the area reserved for that type of vehicle, a point shall be inferred that is situated in the area provided for that type of vehicle and at a distance from the mean center point p of:. shortest and. place the average center point ^ the same coordinate points as that point. 34. The method of claim 8, characterized in that the method optimization step includes the steps of: simultaneously collecting location data from the source system and the reference system, and generating time series of source system data and reference system data based on user identifiers and time stamps in the location data. such that time stamps are replaced by time stamps. *. : at intervals, and to determine predefined constants,. · · Combines the possible values of the predefined constants and performs an algorithm for each value combination that computes the location bar - ::. calculates the errors between the location estimates and the positions measured by the reference system, and during execution, · · »31 113141 adds a numerical value for that error to each interval in the time series, and writes the value combination and the error value after each run, reads the memory and looks for the value combination with the smallest variable value, thus obtaining the values of the predefined constants. 35. An apparatus for locating source system objects, the apparatus including 10 means for receiving a column, the column defining an object's location area, means for providing location estimates, determining where the source system object is located at a given time, the apparatus being characterized by: means for receiving requirements from the unit regarding location estimates, and means for calculating a location estimate using a plurality of columns, said location estimate including a confidence sum whose value determines whether the location estimate satisfies the requirements. Apparatus according to claim 35, characterized by: further, the apparatus further includes means for requesting a new column from the source system. The apparatus of claim 35, characterized in that the requirements include a reliability value defining a probability of how reliable the location estimates must be. The apparatus of claim 35, characterized in that the requirements include an accuracy value defining a distance for how accurate the location estimates must be. ·. Apparatus according to claim 35, characterized in that the requirements include a time interval defining how often the least. now the hardware must ask for a column. v. The apparatus of claim 35, characterized in that the requirements include a second interval determining how often. · * * Hardware may not ask more for a column. 32 113141 The apparatus of claim 35, characterized in that the apparatus further comprises: a reliability table consisting of cells each cell containing a reliability value and a reliability table including table values as table rows and 5 column headers each determining whether the location estimate meets the precision and accuracy requirements. , which determines the distance how accurate the location estimate must be, to receive the table value for the reliability sum from the location estimate-10 country reliability sum, to receive the specific cell of the trust table that is associated with those table values, to compare the reliability value stored in that cell with the resulting in a location estimate that meets the accuracy and reliability requirements, k and the former reliability value reaches the latter reliability value. An apparatus according to claim 35, characterized in that the apparatus further includes 20 reliability functions that output the reliability value in response to the precision input value and the reliability sum input value and to determine if '·' meets the accuracy and reliability requirements, the hardware is (* matched) : receive the precision input value from the precision value that defines the distance how accurate the position estimate must be to receive the reliability sum input value from the position estimate *. resulting in a location test that meets the accuracy and reliability requirements ··· when the first reliability value reaches a the latter reliability value v. 43. Apparatus for locating source system objects, the apparatus 35 comprising 33 113141 means for receiving a column, wherein the column defines an object's location area, means for providing location estimates, determining each location estimate where the source system object is located at a particular time. means for maintaining a plurality of columns for each object to be located, means for calculating a location estimate using a plurality of plots and 10 computation formulas with predefined constants, and means for finding predefined constants values during the pre-hardware optimization step; Apparatus according to claim 35 or 43, characterized in that the source system is a communication system and the object is a subscriber. I I »t f t * * *» t »* *» * {»» 113141