KR20060136447A - Method and device for adapting a radio network model to the conditions of a real radio network - Google Patents

Abstract

The present invention relates to a method and apparatus for matching a new wireless network model to an existing wireless network condition having network model parameters related to network location obtained by using measurement data collected from a certain measurement location of a real wireless network. In order to match one wireless network model to the actual conditions of the wireless network in a simple way using the measurement data, the model variables are matched to the measurement data by directly modifying each model variable according to the measurement data through mathematical operations. Let's do it. In this way, the parameters of the wireless network model are not changed according to the measurement data, but model variables are directly modified.

Wireless Networks, Cells, and Regions

Description

TECHNICAL AND DEVICE FOR ADAPTING A RADIO NETWORK MODEL TO THE CONDITIONS OF A REAL RADIO NETWORK}

The present invention relates to a method of matching a new wireless network model to an existing wireless network condition with network model parameters related to network location obtained using measurement data collected from a certain measurement location of a real wireless network.

The invention also relates to an apparatus for carrying out the method.

Wireless networks, such as mobile wireless telephone networks, include transmission stations distributed over the ground. Each transmitting station has one “cell”. Each of these call zones is assigned a "call zone distinguished name". Relevant parameters set for the planning and functioning of the calling area, for example "path loss data", change throughout the calling area. These variables show the characteristics of radio wave attenuation on a physical basis. In the subscriber terminal, for example, the reception power of the transmitting station which is constantly radiated in a constant direction in the mobile telephone is attenuated as the distance from the transmitting station increases. On the one hand, the attenuation phenomenon is caused by the distribution of a wide wave region, which, like light propagation, becomes larger with increasing distance. On the other hand, the degree of attenuation is also determined by local surface uptake or by buildings or terrain.

In order to plan and optimize the wireless network, a wireless network model is prepared. To this end, the wireless call area is divided into a grid network consisting of pieces having a relatively small area. Then, the regional segmented fragments are each given a “model variable”. The model variables are values that are applied to the local segmentation fragment as variables related to the function of the wireless network. Such a variable is mainly a value indicating transmission path loss. In addition, the transmission power of the transmitting station may be received with different strength and propagation time at certain points in the communication area. One pulsed transmit signal will then be received as several delay pulses with different amplitudes. So people say it's an "impulse response." Such model variable values that are assigned to each local segment then form a matrix.

In the planning phase, it is also impossible to measure the relevant physical variables until the telephone network is established. In addition, even after the installation of the wireless network, it is impossible to measure physical variables such as loss of communication path in each individual segment. Most regional fragmentation fragments are never immediately accessible. In addition, measurement of all regional fragments is costly. For that reason, in order to predict wireless communication channels, we develop mathematical models that model various influence variables and their effects, for example, on the loss of communication paths, based on physical laws or relevant facts that have been found to be critical. It was developed. These models contain certain parameters.

The channel models thus obtained are usually only incompletely matched to the actual system. Therefore, the first-obtained channel models need to be matched to the actual system as much as possible using the measurements of the actual variables. For this purpose, in known methods, the parameters of the channel model are changed using measurements. In this case, the channel loss value is newly calculated by repeatedly matching the parameters and conditions of the channel model step by step. This method is very expensive because it does not know which parameter corresponds to the deviation between the measured value and the model variable, that is, how to change the parameter to minimize the deviation between the model and the actual. .

An example of the prior art is the presentation of D.J.Y Lee and W.C.Y Lee's “Fine Tune Lee Model” published on Indoor ans Mobile Radio Communications September 18-21, 2000, IEEE Conference on Personal pages 406-410.

International Publication No. WO 02/073997 A1 discloses information about a base station, that is, a transmission power and a radiation directing diagram, in order to match a channel loss model to actual radio network conditions. In this paper, a method for obtaining a communication channel loss model from radiographic altitude, terrain information, and measurement data, that is, signal strength has been disclosed. The model emerging from the physical conditions is then supported by the actual measurement data. As already mentioned above, the measurement data affects the parameters of the model. Such a channel loss model provides model variables in the form of channel loss values for all points in the observed transmission area. Normally, the channel loss value obtained from such a model deviates from the actual measured value. Such deviation is due to shading. To include this deviation in consideration, the next step is to statistically analyze where the parameters for shading-prediction appear in the observed area. Therefore, a second model for shading is used, and the parameters of the second model obtained from the measurement data are determined by the channel loss value obtained from the first model. The shading values thus obtained are processed prior to the channel loss value obtained from the first model. If the measurements are very reliable, the channel loss value of the measurement location obtained from the above-described model may be replaced with the actual value. For less reliable measurements, the weighted average of the measurements and model variables is used instead.

Therefore, in the method according to International Publication No. WO 02/073997 A1, work is carried out in two steps with two models, and the parameters of the two models are determined using measurements obtained at a predetermined measurement position. As with the prior art described above, the key here is the determination of the parameters of the model. Substitution of model variables by actual measurement data can be made directly at the measurement site itself, if necessary.

An object of the present invention is to match one wireless network model to actual conditions within a wireless network using measurement data in a simpler way than in the prior art.

According to the present invention, the above problem is solved through the following processing steps in the above-mentioned method of the kind. In other words,

-Set up a fine grid network to form small segmentation fragments within one radio call area, and then model variable values for each segmentation segment from the radio network model. Is specified,

In order to match the model parameters to the measurements, a mathematical operation for modifying the model variables of all regional segmentation fragments of the trigrid-like network is performed in advance.

In this case, the mathematical operation for all the local segmented fragments is solved by directly making (dependent upon) the measurement data and the positional conditions of the local segmented fragments.

Therefore, according to the method of the present invention, the parameters of the radio network model are not changed based on the measurement data, but model variables are directly modified. In the model variables of all regional segmentation fragments, the correction step is performed by a certain mathematical operation. The operation then depends directly on the location and the measurement data of the local segmentation fragments.

The method of the invention can preferably be carried out through the following procedural steps:

By dividing the currency area into the above-mentioned fine-grid lattice network, numerous subdivided fragments are formed, and these subdivided fragments each form a relatively large number of tree zones with a plurality of adjacent fragments. Determining a cross-lattice grid that will superimpose these cross-sections together on the fine grid network;

-Acquisition of measurement data at each measurement location; and

Modifying the model variables assigned to the segmentation fragments in each segment by mathematical operations according to the data obtained in the segment.

The mathematical operation may include interpolation.

The interpolation can be carried out according to the following method steps:

-Incorporating the measurement data at each of the tree area measurement sites of the barbed grid network into one value having the characteristics of a general size in the measurement data at that time,

Determining a reference point to which a characteristic value of a measurement is to be assigned in the tree area of the tree grid network;

Determining the radius of influence around the reference well;

Changing the model parameters in all partitions in the radius of influence according to the slope function of the distance of the partition from the reference point.

It is advantageous if the value characterizing the size of the measurement data is an average (algebraic or geometric) of the measurement data. In addition, a weighted average is mentioned here.

A preferred application point of the method according to the present invention is communication path loss data in which the model variables are modeled, and the measurement data are path loss data, and these (path loss data) are the cell of the wireless network. It is determined from the reception power of the reference signals transmitted from the signals.

The model variables may, on the other hand, also be impulse responses modeled according to the aforementioned purpose. These impulse responses are characterized by various physical variables, for example, received power or path loss data and accompanying phase difference or time difference. These physical variables can be integrated into the matrix.

Preferably, as a first step, the following processing steps are planned:

The collection of measurements across the entire calling area,

Determining the value characterizing the magnitude of the measurements,

A calculation step of values characterizing the magnitude of the model variables throughout the currency zone,

Calculating the difference between the values characterizing the magnitudes of the measurements and model data, and

Modifying model variables by the difference.

At this time, it is advantageous that the model variables and the measured data are averaged.

In this way, the first “mismatching” between the entire wireless network model and the measured data is eliminated. Thus, the average of the measured data over the entire wireless coverage area and the average of the model variables for the entire coverage area coincide. Nevertheless, there are usually local deviations. These local deviations can thus be adjusted in the manner described above.

In order to implement the method described above, there is a database in which a virtual wireless network model having model parameters related to position is stored, and a measuring device for calculating a position-related measurement value of a modeled real wireless network obtained at a measurement location. The apparatus of the present invention is characterized in that the data processing apparatus can modify all the model variables directly to match the model variables with the measured values according to the measured data through mathematical operations.

In this case, the measuring device is configured to allow the reference signals radiated from the call region to respond (operate) to the received power. The measuring device is also configured to detect an identification name of the call area. Preferably, the measuring device is movable and mounted on the measuring vehicle so that the roads in the area of the call area can be traversed. The measuring device may be a terminal of a wireless network. It would also be advantageous if the measuring devices were each equipped with means for confirming the actual reception position, ie a satellite navigation receiver. In addition, it may then be arranged to record and output the measurement, together with the identification of the calling area and the location of the measurement point.

Based on the radio network model matched by the above-described method, the optimization of the radio network can be achieved, for example, by changing the antenna angle of the transmitting station.

An embodiment of the present invention will be described in detail with reference to the accompanying drawings.

FIG. 1 is a schematic diagram for explaining the method according to the present invention, which shows the details of the wireless call area and the tree grid network, wherein the grid network is small. Local segmentation fragments are defined, and a broader tree segment is defined in the tidal grid, which includes multiple segmentation fragments.

2 is a schematic diagram showing an apparatus for implementing the method according to the invention.

3 is a block diagram showing the overall implementation of the method of the present invention.

4 is a block diagram showing the preprocessing according to block 48 of FIG. 3, respectively.

5 is a block diagram showing the overall deviation correction according to the block 50 of FIG.

FIG. 6 is a block diagram showing local matching according to block 52 of FIG. 3, respectively.

7 is a schematic showing that the original pathloss matrix has been modified by actual measurements.

8 is a schematic diagram showing the progress of interpolation functions inclined in all directions from the reference point in the form of contour lines.

FIG. 1 shows a single fine-lattice network 10 indicated by reference numeral 10, which is a fine-grained grid 10 divided into a plurality of small regional segmentation fragments 12. It forms a network area to establish a wireless network call area. All local segmentation fragments 12 are each given a model variable value. The model variable consists of currency loss data, for example. The communication path loss data provides the attenuation information transmitted from the transmitting station in the direction of the region segment fragment to the region segment fragment 12 on the radio transmission path. The model variables assigned to the regional segmentation fragment 12 are indicated by points 14 in the figure. The model variables are derived from the channel model. The channel model is first developed as a mathematical model in consideration of various influences such as distance, topography of the earth surface, buildings and cultivation, and related matters empirically obtained. The channel model is usually inconsistent with reality. Therefore, the measurement is to match the actual. The measurement is not made for every individual segment. In FIG. 1, the model 16 or the actual value of the physical variable from which the model variable is to be derived, e.g., the road 16 can be measured along the road 16 to measure the actual value of the received power to obtain the lost data. do. The road 16 may be, for example, a road on which a measurement vehicle can travel. In FIG. 1 the road 16 can be opened to obtain measurement data relating to the location, which are represented by points (data display points) 18.

Using the measurement data 18, which do not include the entire regional segmentation fragments 12, a direct correction to the model variable 14 is performed by interpolation. This modification is done in the following way.

A superficial grid network (20) is superimposed on the fine grid network (10). The tree grid network 20 is divided into a plurality of tree zones 22 in the call area. Each of the trough zones 22 again contains more small regional segmentation fragments 12. In each of the tide zones 22, one reference point 24 is set. For convenience these reference points 24 basically form the center point of the right-angled tide zone 22.

To modify the wireless network model, proceed as follows:

First, a global correction is made for the mismatch of model variables within the entire monetary zone. For this purpose, the mean value of the measurements on the one hand and the mean value of the model variables on the other hand are calculated over the entire currency zone. The model variables are overall corrected by the difference in their mean values.

However, this step still allows local deviations between the modified model variables and the measurements as described above. Therefore, additional local matching of model variables is performed by interpolation. For this purpose, an average value of the measurement data is made for all the zones 22 of the barbed grid network 20 from which the measurements were obtained. The model variable at the center point 24 of the trough zone 22 is placed at a value that depends on the mean. The model parameters of the other regional segmentation fragments 12 of this light zone 22 are modified as a function of decreasing distance away from this center point 24 from this point.

An apparatus for the implementation of the above method is shown in FIG. In Fig. 2, reference numeral 26 denotes a wireless network. One measuring device 28 receives measurement data from the wireless network 26. The measuring device 28 is mobile and moves along the road 16 of FIG. 1 in the embodiment described above. The measuring device 28 may be mounted on the measuring vehicle. Also important here is the terminal of the wireless network. The measuring device 28 comprises a device for determining its instantaneous position. The device may be a satellite navigation receiver (e.g. a GPS) or a direction finder that locates the measuring device by detecting the location of multiple transmitting stations.

The wireless network includes a plurality of call areas or base station areas that may overlap each other. Each currency zone has a unique identifier. The measuring device is adapted to respond (response) to the identification code. By doing so, the measurement data obtained when the coverage area is overrun can then be assigned to the individual coverage areas.

The measuring device 28 may retrieve, record, and output the obtained measurement data.

Data of the wireless network 26 is stored in the database 32. Among them are the model variables of the "original" unmodified wireless network model. In FIG. 2 such a relationship is represented by a bidirectional interface 30 between the database 32 and the wireless network 26. The database 32 exchanges data with the computer device 36 via another bidirectional interface 34.

In the computer device 36, a wireless network model is first stored in the storage device 38 in the form of wireless network planning data together with model variables. Measurement data is supplied from the measuring device 28 to the computer device 40. This process is indicated by arrow 42. The computer device 40 exchanges bidirectional data with the storage device 38. The computer device 40 thus receives model parameters and measurement data of the radio network model. The computer device 36 corrects the mathematical operations described above, ie, the inconsistency of the model variable with respect to the measurement, from the value, and then performs a local correction. The modified model variables are stored back into storage and stored in the database 32 via the interface 34. The computer device 40 is operated by a person in charge of wireless network planning and optimization.

The database 32 receives information from the wireless network 26 via the interface 30 about changes in the wireless network, for example various connections not yet installed. Such matters are taken into account in the optimization of the wireless network.

When the method of the present invention is specifically implemented in accordance with FIG. 3, the radio network planning data (44) for metropolitan areas having 66 transmitting stations of a universal mobile telecommunication system network in an area of about 53 km ² . ) Is presented. Each transmitting station is arranged to manage one to three calling areas (base station areas) each having an independent antenna.

The radio network planning data includes a topographic relief map, and a transmission path loss matrix for all call areas according to a very radio wave model using this relief map. (傳送 經 路 傳送 經-) is calculated. In particular, in the radio propagation model, since there is no data on the building of the observation area, the influence of the building structure is not considered. In that regard, some apparent deviations were expected between the propagation loss prediction data and the loss measurements recorded in this area. The propagation path matrix was decomposed into a grid area of 25m x 25m, whereby a small regional segmentation of the three-grid network as described in claim 1 was established.

In addition, the radio network planning data also included data on antennas and entire installations as well as other additional attenuation factors in the call area. Antennas are represented by three-dimensionally representing the antenna directivity. In addition, reference to pilot signals emitted from individual call areas and similarly radiated cell identifiers (call area identifiers) were stored in the radio network planning data in the form of scrambling codes. Using this data, the planar distribution of the expected reception power of the test signals could be calculated in a matrix form from the path loss data presented in the matrix described above. The received power matrix is divided into the same lattice network as that of the path loss matrix on which it is based, that is, a mesh of 25m x 25m, for example.

The measurement was performed by using a wireless direction measuring device in a real area of the universal mobile communication wireless network reflecting the aforementioned wireless network planning data. In addition to measuring the received power of the test signal, the measuring device was able to detect from which call area the test signal at that time was radiated using the corresponding scrambling code. In addition, for each measurement value recorded, the reception position received by the GPS receiver is calculated and stored together. By driving a series of roads in a vehicle with a mobile radiometer during the measurement, it was possible to obtain / use a sufficient amount of measurements 46 from the observation area as input data according to FIG. 3. In order to compare the measured values with the received power predictions of the test signal, both values were appropriately averaged locally before matching, and then compared for all measurement locations. In the above examples, it was confirmed that the average deviation of 13.5 dB or more was shown in the standard deviation of 11 dB or more between them.

The method according to the invention is implemented with a computer device so that matching of the pathloss matrix can be made automatically using the measurements. First, the input data was preprocessed in step 48. At this time, as shown in FIG. 4, first, in step 56, a tree lattice network of 250m x 250m is formed, and the spacing of two adjacent segmented fragments in the tree lattice network is approximately It corresponds to the average distance of two roads and their measurements are presented. By parameterizing in this way, measurements were assigned to each of the regional fragments of the rough grid network in step 58, and the average value was calculated for each call area (cell) in step 60. In addition, in step 62, the total measurements assigned to one particular call area are counted for each fragment as well as the total sum of the grid-like network. The call zone designation was then made by the scrambling code of the corresponding call zone. The sum of the measurements per call area was determined in step 64 whether matching of the transmission path loss matrix would be made for the call area in accordance with the purpose of the present invention. If there are too few measures for the currency zone, the measures presented are excluded from consideration because they are not reliable enough. For example, some areas of communication did not transmit because the switch was switched off during the measurement process and there was no measurement. As a result, these areas were automatically excluded from matching. In addition, the segmentation of each call zone and tide grid network determined at step 66 whether there are enough measurements for local matching later. If the number of measurements in one segmentation fragment for a currency region is less than or equal to a predetermined minimum, then that segmentation segment for that currency region excludes local matching.

Substantial matching of the pathloss matrix using the measurements was made in steps 50 and 52 according to FIG.

First of all, global matching was performed for all monetary zones where enough measures were presented. Then, the available total measurements, designated by scrambling code throughout the region by currency zone, were averaged in step 68. In step 70, the average value of the predicted received power values of the test body was calculated at each measurement site for each call area, and both deviation values were compared with each other. In step 72, the ratio of both means or logarithm deciBel-difference was calculated as the total deviation of the predictions for the measurements. In step 74, the corresponding matrix values of the total regional fragments of the fine grating network were corrected by the total deviation.

In the next step, local matching of the pathloss matrix was performed according to the flowchart of FIG. In step 76, each center point of each tree zone of the tree grid is defined as the reference point 24. Next, in step 78, the deviation between the actual measured values of the test signal and the estimated received power values, which were previously averaged, was calculated for each call area and segmented grid network segment. In other words, the measurement deviation values) are assigned to corresponding reference points for each call area. Using the measured deviation value as the reference point, an interpolation function was calculated in step 82, wherein the spaces between the reference points were interpolated according to the gradient function of the interval. The interpolation function 88 illustrated in FIG. 7 is shown for any call zone, where the measured deviation values of the reference points are indicated by points 86. For the interpolation function, an influence radius surrounding a reference point may be defined, and the influence radius represents the influence of the gradient function over the separation distance (거리). The radius of influence may, for example, coincide with the net width of the barbed lattice network (250 m). 8 shows the contour of the interpolation function 88 illustrated by way of example. It can be seen that the ellipses drawn around the reference point are inclined as the distance increases gradually. After calculating the interpolation function, in step 84 the interpolation function is applied to the small segmented fragments of the original pathloss matrix 90 in each call area. As long as both matrices are logarithmic, the point of the calculation is addition. The calculation result is a matched path loss matrix 92 as shown on the right side of FIG. 7, which contains the test signal reception power value measured in the corresponding call area. Compared with the original pathloss matrix 90, the effect of the interpolation function can be clearly seen. In doing so, the three reference values 86 selected as the above example reappear as distinct elevations 94 in the newly lost path matrix as a result of the calculation.

This two-step method can be fully implemented in seconds thanks to computer-assisted automated processing for all pathloss matrices. After all of the pathloss matrices of the entire coverage areas were matched in the manner described above, again a new comparison was made between the measurements and the predicted received power matrices. At this time, in the standard deviation of 9 dB or less, the average deviation could be reduced to 1.4 dB or less.

Claims (16)

Using the measurement data about the model parameters of the actual wireless network obtained from the measurement sites, one wireless network model is matched to the conditions of the actual wireless network providing the model variables related (dependent) to the measurement location. In the method, Splitting the call area (cell) into small fragments (12), defining a fine grid (10) for assigning one model variable value to each segment (12) by the radio network model. (Estimating), Performing mathematical operations for modifying model variables of all the divided fragments 12 of the fine grid 10 in order to match the model variables with the measurement data, and the like; At this time, the mathematical operation for each divided fragment 12 is performed by the processing steps that are directly influenced by the measurement data at that time and the position data of the corresponding divided fragment 12. A wireless network model matching method characterized by the above-mentioned. According to claim 1 -To determine the cross-grid network (20) for forming overlapping zones (22) each comprising a plurality of segmented sections 12 of the fine grid network (10) to superimpose on the fine grid network (10) To do that, Obtaining measurement data 18 at the measurement position and Modifying the model parameters assigned to the area segmentation fragments 12 in each tree segment 22 by mathematical calculations of measurements of the tree segment 22 of the tree grid network 20; A method characterized by including processing steps. The method according to claim 1 or 2, And said mathematical operation comprises an interpolation method. The method of claim 3, The mathematical function has the following processing steps, i.e. Incorporating measurements (18) obtained at each measurement location in each of the tide zones (22) of the tide grid network (20) into measurements having a general size with characteristic values; , Defining one reference point (24) in each trough zone (22) of the trough grid (20); Defining an influence radius around the reference point 24; Changing the model parameters of all the divided fragments 12 within the influence radius according to the slope function of the distance of the divided fragments from the reference point 24. Method comprising a. The method of claim 4, wherein Wherein the value characterizing the magnitude of the measurement data is an average of the measurement data. The method according to any one of claims 1 to 5, The model variables are modeled transmission path loss data, and the measurement data are path always data, wherein the path loss data is determined by the reception power of reference signals transmitted from call areas of a wireless network. How to. 7. The method of claim 1, wherein The model variables are modeled impulse responses. The method of claim 1, wherein The first processing steps are as follows: The collection of measurement data across the call area; Determining a value characterizing the magnitude of the measurement data, A step of calculating a value characterizing the magnitude of the model variables over the entire currency zone, Calculating a difference between values characterizing the magnitude of the measurement data and the model data, and Modifying model variables by the difference Characterized in that the method. The method of claim 8, The values characterizing the model variables and the magnitude of the measurement data are means. In order to carry out the method according to any one of the preceding claims, One database 38 to store the virtual wireless network model together with the location-related (location dependent) measurement data 18 and for calculating the location-related measurement data 18 of the actual wireless network modeled at each measurement location. In the apparatus provided with the measuring device 28, A facility for implementing a wireless network model matching method, characterized in that each model variable can be directly modified through a mathematical operation based on the measurement data (18) in order to match the model variables with the measurement data. The method of claim 10, The measuring device (28) is a facility for implementing a wireless network model matching method, characterized in that for responding (operating) to the received power of the reference signals transmitted from the communication zones. The method of claim 11, The measuring device (28) is a facility for implementing a wireless network model matching method, characterized in that it is installed to retrieve the identification name of the call area. The method of claim 10, wherein The measuring device 28 is a facility for implementing a wireless network model matching method, characterized in that the mobile. The method of claim 10, wherein The measuring device (28) is a facility for implementing a wireless network model matching method, characterized in that the terminal of the wireless network. The method of claim 10, wherein The measuring device (28) is a facility for implementing a wireless network model matching method, characterized in that it comprises a device for checking the actual receiving position at that time. The method of claim 12, wherein A facility for implementing a wireless network model matching method, characterized in that it comprises a device for recording and outputting the measured value together with the identification name of the calling area and the location of the measurement point.

Images