METHOD OF CELL SITE LOCATION IN A CELLULAR MOBILE NETWORK
This invention relates to a method for identifying the location of cell sites in a cellular mobile telephone network.
More particularly, the invention provides a method of more accurately locating one or more cell sites in a competitor' s network involving reverse engineering and using only standard data reported by a mobile test telephone, operated on a selected drive route in town or country, without requirement for any modifications to the telephone. This is in contrast to location-finding methods of the prior art in that it applies to the case of finding the locations of a number of fixed transmit cell sites from a single mobile telephone receiver.
According to the present invention there is provided a method for identifying the location of a fixed transmit cell site in a cellular mobile telephone network, comprising the steps of:
deriving survey data measurements of the network by means of a mobile test telephone, the survey data measurements comprising: geographical location data of
a receiver of the test telephone; timing advance data providing distance between the fixed transmit cell site and the receiver of the test telephone; and cell identity field data;
grouping into individual subsets the survey data measurements derived on a drive route, wherein each subset corresponds to measurements belonging to the same fixed transmit cell site;
identifying first site cluster data and second site cluster data for the transmit cell site location by means of multiple triangulation computations in which each individual point in respect of the geographical location data of the receiver of the test telephone is paired with each other individual point to provide in each case a pair of locations whose distance from the transmit cell site location is determined from the corresponding timing advance data; and
determining which of the first site and second site cluster data has the greatest cluster density and determining centre of gravity of that site as identifying the location of the fixed transmit cell site.
The survey data measurements may include mode data for establishing that the test telephone is switched into a dedicated mode, as when engaged in a call.
The geographical location data of the receiver of the mobile test telephone may comprise longitude/latitude data or Eastings/Northings data.
The method may include the step of checking the derived survey data for completeness and either discarding incomplete data or entering a predetermined value for any parameter where no data value has been derived.
The cell identity field data may be used to group the survey data measurements into individual subsets.
Adjacent points on the drive route having the same timing advance data values may be grouped together. Such grouping together of adjacent points on the drive route having the same timing advance data values would be equivalent to spatial filtering. In this respect, the median location for any particular timing advance may be used as the representative receiver location for the respective group.
A likely transmit cell site location may lie at one or other of intersections between a pair of non-concentric circles
each having a geographical location of the receiver of the test telephone at a centre thereof and a radius given by the appropriate timing advance data value.
The multiple triangulation computations used to obtain the first and second site cluster data may be arranged to provide a matrix of elements of intersection points. Gatherings of intersection points with highest density of points may be determined. Such determination may be effected by searching the matrix to establish a row therein having a maximum number of elements falling within a distance threshold. The distance threshold may be of the order of 500 metres and may correspond to the resolution of the method allowed for by the use of the timing advance procedure.
The first cluster site data may comprise all points within the selected row that satisfy the distance threshold condition. Points belonging to the first cluster site data may then be removed from an overall pool of data and, after building a new matrix of elements with intersection points, the second cluster site may be identified in like manner to the first.
Determination of which of the first site and second site cluster data has the greatest cluster density may be
effected by determining in each case a ratio between mean distribution of distances in each cluster site and the number of points forming each cluster site and selecting that cluster site having the smallest value for such ratio.
Alternatively, the determination may be effected by estimating the moment of inertia (second moment) of the cluster around the mean and selecting that cluster site having the smallest value for such moment of inertia.
The method of the invention may be repeated whereby the locations of a plurality of fixed transmit cell sites in a cellular mobile telephone network may be identified.
For a better understanding of the present invention and to show more clearly how it may be carried into effect, reference will now be made, by way of example, to the accompanying diagrammatic drawings in which:
Figure 1 is a flow chart showing steps in the method of the present invention;
Figure 2 is a graph showing a section of a drive route in respect of a fixed transmit cell site and showing two cluster sites as candidates for the site location;
Figure 3 is a flow chart showing steps in identifying a correct cluster site and the locating therein of a fixed transmit cell site; and
Figure 4 is a graph illustrating accuracy of identifying a fixed transmit cell site by the method of the present invention.
A reverse engineering method is illustrated in the flow chart of Figure 1 which is intended for accurately identifying the location of one or more fixed transmit cell sites in a cellular mobile telephone network, such as a network operated by a competitor.
The method is described in the following steps.
1. Survey Data
Survey measurements of a competitor' s mobile telephone network are performed on a drive route using a mobile test telephone operating in dedicated mode (that is, while engaged in a call) . The survey data is arranged to contain the following measured fields:
a) Receiver geographical location. Such geographical location is provided in the form of longitude/latitude data, or Eastings/Northings data.
b) Timing advance. Timing advance data provides a measure of the distance between the fixed transmit cell site and the receiver of the mobile test telephone. The timing advance measures the distance from the fixed transmit cell site in time bins equivalent to about 500 metres in distance. A timing advance of zero would mean that the receiver is at a distance of between 0 and 500 metres from the fixed transmit cell site and a timing advance of 1 would mean that the receiver is at a distance of between 500 and 1000 metres from the fixed transmit cell site; and so on.
c) Mode. It is necessary to ensure that the receiver of the mobile test telephone is operated in a dedicated, rather than idle, mode.
d) Cell identity (ID)
It is to be noted that the parameters referred to above follow the terminology of the GSM cellular standard. However, equivalent parameters are available in other
digital cellular standards, to which the method of the present invention is equally applicable.
2. Pre-processing
After acquiring the survey data, the relevant fields are extracted, such fields being the geographical location of the receiver (longitude/latitude data, or Eastings/Northings data) , timing advance, mode, and cell identity.
The mode in which the test telephone was set is then inspected to ensure that the survey data corresponds to measurements made in dedicated mode. This is important since hand-over failures would switch the mode to an idle mode setting, resulting in loss of the timing advance field. Any data corresponding to periods where the telephone was switched to idle mode is discarded.
3. Data Partitioning
The next step is to separate into subsets the drive route employed in obtaining the survey data, with each subset corresponding to measurements belonging to the same transmit cell site. The cell identity field is checked to ensure that individual subsets contain all the selected data relating to a given site.
The cell location process described hereinafter relates to individual subsets.
4. Data Filtering
The transmit cell site location procedure is enhanced with regard to speed and efficiency by grouping together adjacent points in the drive route which have the same timing advance values. This is equivalent to applying a spatial filter and results in a reduction in the number of points to be processed, with a significant reduction in processing time. It also results in increased accuracy of the site location, since any ambiguity of the estimated locations is minimised. Any representative location could be used, but the median location of any group is used as the representative receiver location for the particular timing advance to which the group, refers. This approach has been found to maximise accuracy, since it ensures that the representative location is on an actual location visited by the mobile test telephone on the drive route.
5. Initial Site Location Process
Computation of the likely site location is achieved by means of multiple triangulations. Each point in respect of the geographical location data of the receiver of the test
telephone is paired with each remaining point from the transmit cell site subset, the distance between these receiver locations and the corresponding site locations being determined from the timing advance data. The likely site location would consequently lie at the intersection between a pair of circles centred at the two receiver locations and each having a radius given by its associated timing advance data value. However, since two non- concentric circles of comparable radii would normally intersect at two locations, a second intersection point, hereinafter referred to as the image point, will also appear as a result of solving a second order equation. In addition, the inaccuracy in determining the distance from the timing advance data will also lead to a spread of the intersection points.
For a particular drive route involving a subset of M receiver locations, if all the pairs of receiver locations produce two intersection points, a set of M x (M-l) intersection points will be produced by the triangulation.
6. Cell Site Location
Figure 2 shows a section of a drive route executed in relation to a particular fixed transmit cell site having a cell identity number 3862. The drive route is shown by the
dashed line and the circle symbols denote the various receiver locations of the mobile test telephone used to obtain the survey data. The clusters of cross symbols represent two candidate locations for the fixed transmit cell site as yielded by a cluster identification method hereinafter described.
Cluster identification is performed according to the flow chart diagram of Figure 3. Firstly, each point produced by the triangulation process has its distance to all the points calculated, leading to a matrix of N x N elements, where N is the total number of intersection points. The matrix has, however, a very singular shape, as only the elements of the upper and lower triangles in the triangulation process will have non-zero values. In addition, the elements of the lower triangle will be the reciprocals of the elements of the upper triangle, while the diagonal will be zero. This is because the diagonal represents the distance between each point and itself, giving an effective number of significant elements of N x (N-l) / 2.
The next step is to find the gatherings of intersection points with the highest density of points. This is effected by searching the distance matrix to find the row with the largest number of elements falling within a distance threshold. A particular selected threshold is 500 metres
and corresponds to the resolution of the method allowed for by the use of timing advance. Of the two clusters identified by the cross symbols in Figure 2, a first cluster will be made up of all the points within the selected row that satisfy the distance threshold condition. After removing the points belonging to the first cluster from the data pool, a new distance matrix is built and the same operation is carried out to identify the second cluster.
Having identified the two clusters with regard to the number of points they comprise, an analysis of the cluster density is effected to reveal which of the two clusters effectively includes the fixed transmit cell site.
A parameter p is estimated for each cluster and is given by:
p = mean (dist) divided by size (cluster) ,
where mean (dist) represents the mean distribution of distances in the cluster and size (cluster) represents the number of points making up the cluster.
The centre of gravity of the cluster with the smallest value of p is identified as the location of the fixed transmit cell site.
Alternative parameters may be considered for selecting the correct cluster. Such parameters may include the (Eastings and/or Northings) spread of the cluster and the second moment of the distance distribution if the clusters are of comparable density. In either case, the cluster with the minimal value of the parameter under consideration should be selected.
Referring again to Figure 2, there is a clear distinction between the true site cluster, shown at the upper right corner of the graph, and the image site cluster, shown at the lower left corner of the graph. Points in the true site cluster are more concentrated around the actual site, which is denoted by a triangle symbol 1, while points in the image site cluster are dispersed over a larger area. The location of the site, as computed by the method of the invention, is indicated by the square symbol 2.
The accuracy of the method may be established by a statistical approach. A significant number of cell sites of known locations may be processed using the method described above, and the distance between the location derived from this method and the actual site location analysed against a number of parameters. These parameters may include:
The size of the cluster from which the site location is derived.
The spread of the timing advance values from which the location of the site location is derived. The spread could be the difference between the maximum and the minimum timing advance values, the standard deviation of the distribution of the timing advance, or any other statistical variable that describes the spread of data.
The average of the timing advance values used for deriving the site location.
One model specifying the average error distance (distance between true location and the location derived from this method) in units of metres may be given by:
Error = 452.6 + 29.6 x mean (timing advance)
An extensive statistical analysis of the error distance has been performed using three test measurement areas. The true cell site location was accurately known in all three tests so that the distance between the site location computed from the method and the true site location could be estimated. The results of this analysis are shown in Figure 4.
More than 60% of the estimated locations are within 500m of the true locations. The accuracy of the results is found to depend on the outline of the routes driven with respect to the site and the number of measurement points available. The more uniform the route is distributed around the sites, the better the accuracy of the method.