TITLE OF INVENTION: METHOD OF MOBILE TERMINAL TRACKING Field of the invention The present invention relates to a method of mobile terminal tracking, and more particularly, a method of keeping track of mobile terminals in a telecommu- nication system using narrow lobes for communication between a base station and mobile stations. In narrow lobe telecommunication systems, two signals may appear in different narrow lobes but emanating from the same signal source, either due to movements of a mobile station or multiple path propagation. The present invention provides a method of determining if the two signals emanate from the same signal source. If the determination is positive, i. e. the two signals in fact emanate from the same signal source, a switching may be made to a new narrow lobe for communication. The switching itself does not form any part of the present invention. The present invention employs a cross correlation method performed directly on the communication signals, especially speech signals. The method may be complemented by supervisory (+-tone) signals.

Thus, the invention comprises a method of identification of radio signals with different angular origin through the use of cross correlation between the unknown signal and a known signal, using the speech content of the signals only, without having to inject any additional unique information in the signal.

State of the art Today the method of identification of mobile analog radio stations as used in e. g. NMT (Nordic Mobile Telephone) cellular networks relies on sending a signal called a supervisory signal together with the speech signal. This supervisory signal, also called +-tone signal, has a limited number of states. In the case of the original NMT supervisory signal, there are 4 different identifications available, and in the case of the later developed digital supervisory signal there are 35 identifications available.

The 4 signals in the original version are absolutely insufficient to resolve possible identification problems when a base station may be surrounded by 6 nearest neighbour stations, and the 3 5 identification signals available with the

digital supervisory signal also gives strong limitations, if multiple reuse of one frequency on the base site is contemplated.

Additionally, the supervisory signals are kept within a very narrow band of the full signal bandwidth, and therefore require proportionally long time to resolve.

The time to resolve the digital supervisory signal becomes even longer than for the analog version, as each bit in the signal has to be identified. The time required is on the order of 30 ms for the analog supervisory signal and more than 100 ms for the digital signal.

From EP-A1-0 343 570 is known a method of authentification of a mobile in a telecommunication system. The method uses cross correlation on a unique special identity signal. The cross correlation is used in order not to have to synchronize the identity signals. Thus, this method introduces a special identity signal and moreover, does not solve the same problem as the present invention.

The present invention solves the above problems by using cross correlation of the known communication signal, especially the speech signal, with the unknown signal. This overcomes the problem of the limited number of identities and the in this connection long times needed for the supervisory signal identification. The cross correlation is done on a signal with about 3,3 kHz bandwidth and provided that the communication signal is not silent a proper identification should be possible in about 10 ms or less. This is achieved without introducing any additional signalling, such as a special identity signal.

Summary of the invention Thus, the present invention provides a method of mobile terminal tracking in a telecommunication system using narrow lobes for communication between a base station and mobile stations. A first signal is received from one mobile station in a main communication lobe while the base station receives a second signal in an- other lobe.

According to the invention the two signals are compared by means of cross correlation to identify the second signal as emanating from the same mobile station. Preferably, the signals include speech signals upon which the cross correlation is done.

The invention as well as preferred embodiments are defined in the accompanying claims.

Brief description of the drawings The invention will be described below with reference to the accompanying drawings, in which:

Figur 1 is a schematic illustration of two different signals from one mobile, and Figur 2 is a diagram of a cross correlation curve.

Detailed description of the preferred embodiments In a narrow lobe telecommunication system array antennas are used to obtain spatial division, i. e. by using narrow lobes in different directions the same fre- quency may be reused in the same site, thereby increasing the capacity of a system.

However, there are reasons that a new signal may appear in a lobe different from the main lobe in which the communication is proceeding between a base station and a mobile station. If, as will be explained later, it is the same signal source that generates both signals, it may be advantageous to switch the communication from the main lobe to the other, new lobe. The reason may be that the second new signal is stronger than the first one or the mobile station is moving in a predicted direction.

Thus, it is vital to track the movement of the mobile as apparent to the base station during narrow lobe radio communication. This apparent movement may be caused either by actual physical movement on the part of the mobile station, or virtual movement caused by changing reflections of the radio waves in the en- vironment, e. g. surrounding buildings. In the case of virtual movements, these are not necessary continuous and it becomes important to be able to identify if a newly appearing signal on a given frequency and bearing belongs to certain mobile station that is already active on that frequency with another bearing or if it belongs to a noise source.

Noise in this context includes white, pink as well as otherwise distributed noise, and signals from other radio transmitters on the same or neighbouring frequencies.

Tracking of continuous movement in its simplest implementation would be to accept limited changes in angular position in subsequent Direction of Arrival (DoA) measurements, so that a signal source found with one angle and frequency during one DoA measurement is assumed to be the same as a source on the same frequency and nearly the same angle on the previous DoA measurement.

This method can be enhanced by introducing first and higher order analytical predictions, based on previous angular positions. (The simple case above is in effect a zero-order prediction). If a mobile during the previous two DoA measure- ments moved with a certain (angular) speed, we may predict that it will continue to move in the same manner, and thereby accept a new position as valid for the mobile of interest even if it is some distance from the previous position. Addi-

tionally, we can use this information to adjust the lobes to follow the mobile optimally.

Further, empirical predictions can be entered into this too: Let's say that we have a heavily trafficked road with a traffic light controlled intersection close to the base. We may then, based on movement history, say that sources moving with certain speeds through certain angles will probably slow down and halt, possibly with a probability that changes periodically (with the rate of the traffic light).

However, all of the above cases cover situations where we predict a new position, and on finding a mobile with the correct frequency at that position, we are satisfied that it is the same one we communicated with previously. This is however not the only case we have to consider.

In environments rich with reflections, we will probably have multiple paths from one mobile source to the base antenna. We will effectively see signals from the same source coming from different direction with different intensity. Further- more, reflections may appear and disappear, and change in intensity.

In figure 1, a base station BS is receiving two signals a and b in different directions from position a and position P, respectively.

Suppose we are following the signal'a'from a mobile source A on a fre- <BR> <BR> <BR> quency fa with position a, and there suddenly appears another signal'b'with the same frequency fa in position ß. The new signal has better signal to noise ratio SNR, so if it actually comes from the same source A, then we would like to change to following A in position ß instead of the old position a. But how do we know that this is indeed a signal from A, and not a signal from mobile X belonging to a neighbour site? In another case, we may have mobile A, with signals'a'and'b'in positions a and ß. Signal'a'is the stronger one, so we follow it, but A suddenly turns a corner, and'a'becomes very weak or disappears altogether. We would then like to jump to using signal'b'in position P, but again we need to know that this is indeed a signal from A and not from a neighbour site mobile X before we jump.

So we need to be able to identify several diverse signals coming in from diverse directions as to which source they are coming from. We need to keep talking to the mobile we are handling on the given freqency; we may want to change frequency if our mobile comes to close (in angle) to another mobile using the same frequency; and we may want to measure data from other mobiles for relay to neighbour sites.

Let it also be mentioned that identifying the mobile is only a problem for the analog mobiles; the digital mobiles will be able to identify themselves through signatures in their data stream.

However, the present invention is applicable in any system in which the problem of identifying two signals may arise.

The invention proposes to use the cross correlation method to identify the new signal. The cross correlation method as such is a known mathematical method and will not be discussed in all mathematical details here. The idea is that each time a new signal appears, a cross correlation is done between the new signal and the already identified signal. The result of this will be either a positive identi- fication of the new signal as coming from the known source; a conclusion that there is insufficient signal contents to make any identification; or an identification of the new signal as belonging to a new source. The second case, where no identification is possible may occur for instance if there is silence on the channel when the correlation is done, in which case new attempts at identification will have to be done.

Before the cross correlation is performed, the signals in question undergo demodulation, sampling, and normalisation, in this or any other suitable order. The two signals are multiplied with each other, in a number of points and with various time displacements. The product of the two signals may appear as shown in Figure 2, where the product is on the ordinate and the time displacement is on the ab- scissa. Due to the different lengths of the signal paths, the product curve will have a peak at a certain time displacement, if the signals are correlated to each other, i. e. belong to the same signal source.

If the curve has a peak, it will have to be evaluated to make a determination.

As the two signals are normalised, it is possible to use the height of the peak as a measure. Also the width, e. g. the width at half the height, may be used. It will be appreciated that other criteria on the shape of the peak may be applied.

The criteria may be fixed or adaptive. For instance, if the signals are in a noisy environment with much interference, the criteria might be adjusted so that the levels are not set too high for a positive identification.

If direction of arrival determination (DoA) is done through digital signal processor treatment, FFT analysis or other methods, it may be possible to combine the DoA determination with the correlation, through using higher-order spectra analysis. Otherwise, if DoA is done through scanning receivers, cross correlation will have to be done as a separate operation, to be done each time a new unknown signal appears.

The cross correlation method has the advantage that it potentially gives much better identification than the supervisory signal (+-tone) method. However, supervisory tone signals may be used as a complement to the present invention. If the supervisory tone signal identification yields a negative result, i. e. the two signals may not emanate from the same source, the cross correlation step does not have to be performed. At any time a new signal appears, a new path from an existing source or a new source, an identification attempt by cross correlation may be done. If the +-tone is retained in the spectrum of the two signals which are

cross-correlated, the cross correlation match will also depend on matching +-tontes, so additional ¢-tone identification does not add information here. For new sources which appear, +-tone identification may indicate to which base sites a source could belong.

Thus, the present invention provides a method giving fast identification of a new signal without introducing any signalling and without requiring additional equipment. The implementation of the method in hardware is straightforward and does not pose any problem for a person skilled in the art. The scope of the invention is only limited by the claims below.