The invention also relates to a method of measuring the frequency difference in a radio receiver generating for synchronization purposes a tuning frequency corresponding as accurately as possible to a transmission fre- quency, and in which the strength and phase of a received signal is measured in relation to the tuning frequency.

The invention further relates to a receiver, arranged to generate for synchronization purposes a tuning frequency corresponding as accurately as possible to a transmission frequency, and to generate an estimated impulse response for the channel used by a received signal.

The invention also relates to -a receiver, arranged to generate for synchronization purposes a tuning frequency corresponding as accurately as possible to a transmission frequency, and to measure signal strength and sig- nal phase.

BACKGROUND ART A basic requirement for the operation of a radio system is that the connection between a transmitter and receiver is synchronized. In conven- tional radio technology, a PLL receiver (Phase Lock Loop) is employed to achieve and maintain synchronization. However, this method is not very useful in the CDMA system (Code Division Multiple Access).

In the CDMA method, a narrowband user data signal is multiplied to a relatively broad band by a spreading code having a considerably higher fre- quency than that of the data signal. The aim is to select spreading codes that are mutually substantially orthogonal, i.e. have minimal mutual correlation.

Upon transmission, a broadband signal is multiplied, i.e. modulated, by a con- tinuous carrier having a considerably higher frequency than that of the spreading code. Owing to orthogonal spreading codes, the signals of several users can be transmitted on the same carrier.

In a conventional CDMA receiver, the signal carrier is demodulated

by multiplying the signal by the tuning frequency of the local oscillator of the receiver. The tuning frequency is intended to have the same frequency as the carrier employed in transmission. The data signal again is returned in the re- ceiver to the original band by muitiplying it again by the same spreading code as at the transmission stage.

However, there is usually frequency difference between the tuning frequency of a local oscillator and the carrier employed in transmission. It can be seen as a phase shift between the receiver and the spreading code em- ployed in transmission. Without corrective measures, this phase shift in- creases the more the longer the connection between the transmitter and the receiver lasts. In a fading multi-path environment, the change caused by the difference between the transmission carrier and the receiver tuning frequency in the code phase is called a group shift. In a typical multi-path environment of a radio system this increasing group shift cannot be detected reliably with prior art measurements.

DISCLOSURE OF THE INVENTION It is the object of the present invention to provide a method for re- liably detecting a group shift in a multi-path environment.

This is achieved with a method of the kind described in the pream- ble, characterized by generating at least two temporally successive estimated impulse responses, comparing the mutual temporal shift of said impulse re- sponses in relation to one another, generating a phase shift between the esti- mated impulse responses on the basis of the comparison, and forming a fre- quency difference between the transmission frequency and the tuning fre- quency on the basis of the phase shift.

The method of the invention is also characterized by measuring the phase of a received signal at least two temporally successive times, compar- ing the measured phases with one another, generating a signal phase shift on the basis of the comparison, and forming a frequency difference between the transmission frequency and the tuning frequency on the basis of the phase shift.

The receiver of the invention is characterized in that the receiver comprises means for generating at least two temporally successive estimated impulse responses, means for comparing the temporal shift of said impulse responses in relation to one another, and for generating a phase shift on the

basis of the comparison, and means for forming a frequency difference be- tween the transmission frequency and the tuning frequency on the basis of the phase shift.

The receiver of the invention is further characterized in that the re- ceiver comprises means for measuring the phase of a received signal at least two temporally successive times, means for comparing the phases of the re- ceived signal with one another, and for generating a phase shift on the basis of the comparison, and means for forming a frequency difference between the transmission frequency and the tuning frequency on the basis of the phase shift.

Considerable advantages are achieved with the method of the in- vention. The frequency error between the tuning frequency of the receiver and the carrier frequency employed in transmission can be determined exactly and easily, and the error can be corrected. This makes the connection more reli- able than before.

DESCRIPTION OF THE DRAWINGS In the following the invention will be described in greater detail with reference to examples in accordance with the accompanying drawings, in which Figure lisa flow diagram of the method of the invention, Figure 2 is a flow diagram of the method of the invention, Figure 3 shows a receiver according to the invention, and Figure 4 shows a receiver according to the invention.

DESCRIPTION OF THE PREFERRED EMBODIMENTS The solution of the invention is suited to be used in a radio receiver.

The solution of the invention is particularly suitable for the CDMA radio system without, however, being restricted to it.

Let us have a closer look at the inventive method. The invention is based on the idea that when two different frequency sources have the same frequencies, their phases remain unchanged in relation to time. In other words, if the frequencies of two frequency sources are different, their phasing changes as a function of time. The rate of phase change is proportional to the frequency difference. In the inventive method the phase change is measured by comparing estimated impulse responses with one another at different points of time. As is known, the estimated impulse responses are based on a

signal that has been detected. This is carried out preferably by forming a cor- relation between the estimated impulse responses. In another embodiment the group shift is measured by observing directly the signal phase in relation to the tuning frequency at no less than two different points of time. If the phase changes from one point of time to another, the tuning frequency has to be ad- justed.

Let us now have a closer look at the method of the invention by looking at Figures 1 and 2. Processing starts at point 10. At first, a first esti- mated impulse response is generated at point 11. A predetermined or random measurement interval T is then allowed to pass in block 12, and a second es- timated impulse response is generated for the same signal from the same channel or the same channels at point 13. The generated impulse responses are compared with one another by observing their similarity at point 14. The comparison is preferably carried out by calculating the cross correlation or the like. A temporal shift, or phase shift, or group shift, between the impulse re- sponses and a frequency difference to be utilized when correcting the tuning frequency of the local oscillator are generated on the basis of the comparison at point 15. If frequency difference exists, the aim is to eiiminate it. Processing ends at point 16.

The estimated impulse response is generated in accordance with prior art in a manner obvious to those skilled in the art. At least two estimated impulse responses are generated in such a way that at least the initiation of the measurements takes place at different points of time. It is vital that the measurement is carried out sufficiently often to prevent the generation of ex- cessive phase deviations between the transmission frequency and the tuning frequency.

Figure 2 shows another method according to the invention. Proc- essing starts at point 20. At first the phase of a received signal in relation to the tuning frequency is measured at point 21. Then a predetermined or ran- dom measurement interval is allowed to pass at point 22 and the phase of the received signal is measured again at point 23. The process continues this way until the last necessary measurement, the measurement of the Nth phase, is carried out at point 24. N can be any integer N 2 2. At point 25 the signal phases are compared with one another and at point 26 the signal phase de- viation, or group shift, is generated and then the frequency difference is formed on the basis of the phase deviation. It frequency difference exists, the

aim is to remove it. Processing ends at point 27.

In an embodiment of the invention, groups, totalling C, are prefera- bly formed of the phases measured at point 25 to enable signal phase com- parison, and in each group the phases are grouped chronologically. Thus, each group comprises M measured phases. C is preferably determined by the formula CNNM = M!(N M)!, where M is M < N, N! = 1 2 ,..., N and C, ME (1,2, 3, ...). Successive phases are then subtracted in pairs from one another to generate phase deviations and the phase deviations are proportioned accord- ing to the temporal difference. The measurement is preferably carried out at regular intervals, the proportioning being carried out by dividing the phase de- viation, when required, by integers 2, 3, . . depending on the measurement times to be compared. The phase deviations are used to search for at least one group in which all relative phase deviations are substantially the same, and the frequency difference between the transmission frequency and the tuning frequency is formed on. the basis of the phase deviation of the group found. Let a case with 4 measured phase deviations, X1, X2, X3, and X4, when N is N = 4, serve as an example of this method. In this case four groups 443 = 4 are formed in the following manner: X1, X2, and X3; X1, X2, and X4; X1, X3, and X4; and X2, X3, and X4. Next phase deviations, proportioned by groups according to measurement times, are formed: X2 - X1 and X3 - X2; X2 - X1 and (X4 - X2)/2; (X3 - X1)/2 and X4 - X3; and X3 - X2 and X4 - X3. The generated phase deviations are used further to search for at least one group where the proportioned phase deviations are substantially the same (e.g. X2 - X, = X3 - X2), and the frequency difference between the transmission frequency and the tuning frequency is formed on the basis of the phase deviation of the group found..Hence all groups formed of phases do not have to be included in the comparison, instead it is preferably sufficient to determine the phase deviation on the basis of the first group fulfilling the conditions. Proportioning, e:g. the subtraction of the phase X2 of the second measurement time from the phase X4 of the fourth measurement time, is needed, since the phase slide between two successive measurement times (e.g. X2 - X1) is smaller by at least half than in some other case (e.g. X4 - X2 or in a general case Xp - XQ, where P - Q > 2). In a general case the proportioning is carried out by dividing (Xj - Xj) by the number (i - j), where i and j are phase indices.

The signal phase is preferably measured at the strongest point of the signal in relation to the tuning frequency of the receiver. This is advanta-

geous since at the strongest point of the signal, noise interferes least with the measurement. The phase is preferably measured by calculating the cross cor- relation between the signal spreading code and the receiver spreading code.

Let us now have a closer look at the mathematical basis of the so- lution of the invention. Impulse response comparison and signal phase forma- tion are carried out from a received signal as cross correlation or the like.

Cross correlation C is calculated in its general format from variables x and y as follows where a and b represent the correlation calculation interval. There is a correla- tion between cross correlation C and covariance V Vxy(t,l) = Cxy(t, T) - IJx(t)pyT(T) (2) where boldface symbols refer to a presentation in the form of a matrix or vec- tor, Px is the mean of variable x, and p yT is the mean of variable y. This shows that in the solution of the invention, covariance is an operation substantially similar to cross correlation. Correlation C is calculated from digital samples e.g. in accordance with the following formula for variable vectors X and Y where C(n) = C is given real values within the range C E [-1,...,1]. Instead of calculating correlation C, covariance Vxy or another similar operation measur- ing similarity or dissimilarity can also be used. Covariance Oxy is calculated generally from two vector variables X and Y comprising samples as follows: where x is the mean of the samples of variable X, y is the mean of the sam- ples of variable Y, and N is the total number of samples. When e.g. a RAKE receiver is used, multivariate cross correlation calculation can be employed.

The multivariate correlation C of impulse responses X, for example, can be expressed as a formula in a general form as follows where measurements of the estimated impulse responses x(i) are needed at several different points of time.

In a radio system employing a carrier, the tuning frequency is gen- erally preferably the frequency of the local oscillator of the receiver, and the transmission frequency is the frequency of the carrier used in transmission.

The samples needed in the calculation preferably represent chips of the spreading code, and one or more samples can be taken from one chip.

In order to find the mutual group shift of two impulse responses, the cross correlation of the impulse responses is calculated and a search is made for the highest correlation value representing the group shift and the phase deviation of the code. In searching for the highest value, the values are pref- erably limited by a predetermined threshold value which the highest value must exceed. This way the interference caused by noise is avoided. The maximum value of the correlation is thus at sample n.

In another embodiment of the invention, in which the phase devia- tion between the signal carrier frequency and the tuning frequency is meas- ured directly between the signal spreading code and the receiver spreading code as a correlation, the phase deviation can be detected as the phase shift of the correlation peak to a sample n. If the peak of the correlation or the direct measurement result of the phase shift is at point n = 0, this indicates that the tuning.frequency of the receiver and the transmission carrier frequency are identical. If again the maximum value of the correlation is at n = m, the fre- quency difference Af can be calculated as follows: <BR> <BR> <BR> <BR> <BR> <BR> <BR> <BR> f m fre (6) <BR> <BR> <BR> Af= T (6)<BR> <BR> <BR> <BR> fs' where frc is the frequency of the receiver carrier, i.e. the tuning frequency, fs is the sampling frequency and T is the interval between measurements. Let us assume by way of simulation that the tuning frequency which is close to the

carrier frequency is for example frc = 1.8 GHz, fs = 25 MHz and T = 500 ms. In this case the temporal shift corresponding to one chip (m = 1) represents a 144-Hz frequency difference between the receiver tuning frequency and the transmission carrier. Thus the frequency difference is inversely proportional to the measurement interval T. In a stable environment, the synchronization of the tuning frequency can be improved by increasing the measurement interval T. If again the environmental properties undergo rapid changes, the interval T has to be kept sufficiently short to obtain reliable results. In principle the cor- relation function C is cyclical as is the spreading code. Hence the maximum N possible observable phase group shift is m = # 2 , when the cycle length of the spreading code is N. This limitation can, however, be bypassed, and the shift can be extended to the following spreading code cycles. On the other hand, in a typical situation the spreading code is much longer than the phase N group shift and hence m < # 2.

Figure 3 shows a block diagram of the inventive receiver operating in accordance with the inventive method. The receiver comprises an antenna 100, preprocessing means 110, a mixer 120, a local oscillator 130, post proc- essing means 140, means 150 for generating an impulse response, means 160 for comparing the mutual shift of the impulse responses and for generat- ing a phase shift, and means 170 for forming a frequency difference for con- trolling the local oscillator 130. A signal received by the antenna 100 first propagates to the preprocessing means 110 comprising e.g. a filter. The signal then propagates to the mixer 120 where is it multiplied by the frequency of the local oscillator 130. In the post processing means 140 the signal is filtered and e.g. in.the CDMA system the signal is converted by an A/D converter into digital form. From the post processing means 140 the signal propagates to other radio system processes. In addition, the signal, preferably converted into digital form, propagates to the means 150 and an estimated impulse response is generated of the signal at two different points of time. These impulse re- sponses 200 are compared with one another in the means 160 which are preferably a correlator. On the basis of the comparison, the means 160 gener- ate a phase deviation 210 for the impulse responses and the means 170 form a frequency difference 220 between the tuning frequency of the local oscillator and the carrier of the transmitted signal of the phase deviation 210.

Figure 4 shows a block diagram of the inventive receiver operating

in accordance with the inventive method. The receiver comprises an antenna 100, preprocessing means 110, a mixer 120, a local oscillator 130, post proc- essing means 140, means 180 for measuring signal phase, means 190 for comparing the phases and for generating a phase shift and means 170 for forming a frequency difference for controlling the local oscillator 130. A signal received by the antenna 100 first propagates to the preprocessing means 110 comprising e.g. a filter. The signal then propagates to the mixer 120 where it is multiplied by the frequency of the local oscillator 130. In the post processing means 140 the signal is filtered and e.g. in the CDMA system the signal is converted by an A/D converter into digital form. From the post processing means 140 the signal propagates to other radio system processes. From the post processing means 140 the signal, preferably converted into digital form, propagates to the means 180 in which the signal phase is determined at no less than two different points of time. Phase data 230 propagate to the means 190 which compare the phases and generate a phase shift 210. The means 170 utilize the phase shift to form a frequency difference 220 which is used to control the frequency of the local oscillator 130.

The solution of the invention is particularly suited to a subscriber terminal in a radio system, since it has to synchronize to a frequency given from the base station. The solutions of the invention can be implemented by e.g. ASIC or VLSI circuits (Application-Specific Integrated Circuit, Very Large Scale Integration), particularly as far as digital signal processing is concerned.

The functions to be performed are preferably implemented by software based on microprocessor technique.

Although the invention has been described above with reference to examples according to the accompanying drawings, it is obvious that the in- vention is not limited to them but can be modified in various ways within the scope of the inventive idea disclosed in the attached claims.