WO2007003680A1

WO2007003680A1 - A method and device for pilot power measurement in a receiver

Info

Publication number: WO2007003680A1
Application number: PCT/FI2005/000306
Authority: WO
Inventors: Mika Ventola
Original assignee: Nokia Corporation
Priority date: 2005-06-30
Filing date: 2005-06-30
Publication date: 2007-01-11

Abstract

The present invention discloses a method for estimating the power ratio of a pilot channel power to a total transmit power of a base station in a receiver. In the method the covariance matrix of interference is estimated with the help of measured signal samples and measured pilots. The interference covariance matrix contains error, which is minimized when substantially correct power ratio value is used in the estimation process. Thus, the error is minimized by estimating the positive definiteness of the interference covariance matrix. This can be performed by calculating eigenvalues or by making the Cholesky decomposition. The power ratio value used with the error minimum is the measured power ratio. The achieved pilot power data can be used in a rake receiver, IRC receiver or with SHO procedure. The present invention can be used in mobile terminals using WCDMA technology, for example.

Description

A METHOD AND DEVICE FOR PILOT POWER MEASUREMENT IN A RECEIVER

FIELD OF THE INVENTION

The invention relates to a Code Division MuI- tiple Access (CDMA) or Wideband Code Division Multiple Access (WCDMA) technology and particularly to the pilot channel power measurement in the (W)CDMA receiver.

BACKGROUND OF THE INVENTION Spread spectrum technique means that a data signal is spread to a considerably wider frequency band before the signal is transmitted. Spreading to the wider band is achieved by means of a spreading code. Typically, the spreading codes are chosen in such a way that they are orthogonal with respect to each other. Orthogonal spreading codes do not correlate with each other. Moreover, each traffic channel is spread by its own spreading code. By allocating each traffic channel its own spreading code, it is possible to distinguish the traffic channels even though the spread traffic channels are transmitted simultaneously on the same frequency band.

The spreading codes are used to spread the spectrum of the transmitted signal by multiplying the data bits by the higher bit-rate spreading code, thus increasing the data rate of the signal .

Once the spreading has been done, scrambling is used in the network to separate cells and sectors. Scrambling is done by using scrambling codes, which are pseudorandom sequences that are significantly longer than the spreading codes. Thus, the spread data is also multiplied by the scrambling code. However, unlike spreading codes, the scrambling codes do not change the data rate nor the bandwidth of the signal . Instead, the codes are overlaid on top of the spread data as a mask to identify the source of the data.

There are two types of scrambling codes, pri- mary and secondary. The typical primary scrambling codes are formed by truncating Gold codes to a length of 38400 chips (e.g., a frame length). There are 512 different primary scrambling codes in an UMTS FDD network, and each of them is allocated on the different cell. A secondary scrambling code typically includes a set of 15 codes per one primary scrambling code. The primary scrambling codes are first used for separating base stations and their sectors from each other. If the number of primary codes is not sufficient to ful- fil the interference specification between any two connections, the secondary codes are additionally taken into use .

In a (W) CDMA system three separate channel concepts are used for signal transmission. Logical channels define what type of data is transferred. Transport channels define how the data is transferred by the physical layer and the type of characteristics of the data.

In the downlink direction, there are two dif- ferent types of physical channels, namely dedicated channels and common channels. Dedicated channels are reserved for a single user only whereas common channels are available for all users. There are also physical channels for signaling purposes. These chan- nels carry information between the network and terminals .

One of the physical signaling channels in the WCDMA downlink is the common pilot channel (CPICH) . The CPICH has a fixed data rate of 30 Kbps and a spreading factor of 256. There are two types of common pilot channel, primary and secondary. The difference is that Primary CPICH is always under the primary scrambling code, it has a fixed channelisation code and there is only one Primary CPICH for a cell or sector. The Secondary CPICH may have any channelization code of length 256 and may be under a secondary scrambling code as well. CPICH can be used, for example, in channel estimation. Moreover, as will be explained later, the power of CPICH can be controlled by the network operator. Moreover, the power level of the CPICH is needed in various WCDMA network and receiver functions like in hand over (HO) and in an estimation of interference characteristics. Because the power level of CPICH is not signaled to the terminal, it has to be estimated as will be explained later in the text.

There is another type of channel for carrying the physical layer control information. This task is performed by the Dedicated Physical Control Channel (DPCCH) with a fixed spreading factor of 256. The DPCCH is a part of a Downlink Dedicated Physical Channel (Downlink DPCH) . Especially power control information is transferred in DPCCH. The DPCCH uses a slot structure where each slot includes one field to be used for pilot bits. These pilot bits can also used for the channel estimation in the receiver. Later these pilot bits are referred as dedicated pilots.

In (W) CDMA several physical channels are transmitted simultaneously on the same frequency band. Therefore, when the channel coefficients are estimated from the pilot channel, they are scaled by the transmit power of the pilot channel. The estimated channel coefficients (i.e. the channel estimates) in WCDMA can be expressed by:

(D

where h is a vector that contains the coefficients of the propagation channel, n is a vector containing interference and noise and E_cjl_or is the power of the pilot channel (from which the channel estima- tion has been done) divided by the total transmit power of the base station. The power of the estimated channel coefficient is practically speaking always lower than the power of the channel coefficient. The only situation when the power of the estimated channel coefficient would be of the same order as the power of the channel coefficient is a theoretical situation wherein the pilot channel is the only physical channel of the cell and the interference and noise level in the system is negligible small. One characteristic of the (W)CDMA technique is that the multipath propagation can be exploited in the (W)CDMA signal reception. A conventional CDMA receiver is e.g. a rake receiver, which consists of one or more correlators that are called rake fingers. Each rake finger acts as a separate receiver unit. The task of the rake finger is to despread one received multipath propagated signal component in order to produce a decision variable, which is an estimate of the received symbol. In order to compensate the amplitude and phase error caused by the radio channel, signal of each finger is phase-rotated and scaled. This is done by weighting the decision variable by a combiner weight w_lm . The weighting can be performed in the despreader or in a combiner. A task of the combiner is to combine the decision variables of the fingers in order to produce one combined decision variable i.e. estimate of the received symbol. The combined decision variable can be expressed by:

where d_lm is the decision variable, 1 is the finger index, L is the number of fingers, m is the antenna branch index (equal to the antenna index of a diversity antenna system) , M is the number of antennas and

(■)^* refers to complex conjugate. When the combined de- cision variables have been produced, they are detected, for example, by so-called Viterbi decoder.

In equation (2) the number of antennas M can be greater than one. If M is greater than one, antenna diversity is used. Antenna diversity is a well known method to improve quality of a received signal. In antenna diversity reception all antennas of a diversity antenna array receive a signal transmitted by a communication system. If the antennas are placed far enough from each other in the array, the propagation paths from a transmitting antenna to the receiving antennas have slightly different lengths. Therefore, the received signal components of the transmitted signal have different amplitudes and phases. Compared to a case where only one of the antennas is used, the qual- ity of the signal combined over the antennas has better quality, because the energy of the combined signal is collected from more than one detector. Combining over the antennas improves the quality of the received signal especially in a single-path channel, but also in a severe multipath channel.

The prior art presents a method for calculat- ing complex covariance matrices. First, it is assumed that the channel estimates are stored to the vector represented by h :

Next, a covariance matrix of the channel estimates is calculated according to the prior art means by:

6» =hh\ (4)

Equation (4) is a well-known way to calculate a covariance matrix of a vector. Therefore, equation (4) can be used to calculate covariance matrix of the received samples and received symbols . Assuming vector u contains the received samples:

r(i+τ₂) ... Kϊ+τ,) ... r(i+τ_L))^τ , (5)

where r is the received sample, i is the sam- pie index and Ti is the delay of the propagation path (for example the delay of an allocated rake receiver finger) . The samples can be either received chips or despread symbols. A covariance matrix of the samples can then be calculated by:

As shown in equation (2) the combiner weights are required in combining. In a typical (W)CDMA receiver, channel estimation is required for determination of combining weights. Channel estimation means approximating the channel characteristics for example with the help of known transmission pattern. Typically, the channel estimation is performed data aided. In the data aided channel estimation the coefficients of the radio channel can be estimated by multiplying the received pilot symbol by the complex conjugate of the transmitted pilot symbol. Typically, this is repeated over the transmitted pilot sequence and for all the detected paths of the radio channel. Furthermore, it is often assumed that the channel remains constant over several pilot symbols, and therefore it is possible to average consecutive estimated channel coefficients in order to improve their quality.

A well-known combining method is so-called Maximal Ratio Combining (MRC) . In case of MRC the com- biner weights are the channel estimates. Another well known combining method is so-called Optimum Combining (OC) or, alternatively, Interference Rejection Combining (IRC) . IRC requires knowledge of the channel and the covariance matrix of interference plus noise that can be estimated from, for example, the received wideband samples by using means known to a person skilled in the art. The combiner weights of the IRC can be calculated by:

w = c: UiUh_f (7)

where C_uu is an estimate of interference covariance matrix (interference containing intra- and intercell interference plus noise) , (-)^"1 denotes matrix inversion and h is a column vector that contains the channel estimates.

There are a few ways to estimate the inter- ference covariance matrix C_uu by using the received sample covariance matrix C_n.. It is possible to use either :

C =C ^• (8)

f* —c 1 /-* (9)

■C,. / I_n,.

where C_w is the symbol covariance matrix of the channel coefficients (see equation 4) and E_c/I_ox is the power of the pilot channel from which the channel estimation has been done divided by the total transmit power of the base station.

The first equation (8) is an approximation that the interference covariance matrix can be esti- mated directly from the received sample covariance matrix. The received sample covariance matrix contains information of interference and the desired signal. Therefore the equation (8) suffers from the error the desired signal causes to the actual interference co- variance matrix. In the latter equation (9) the symbol covariance matrix of the channel coefficients is subtracted from the sample covariance matrix. The equation (9) can be used to mitigate the error the desired signal causes to the interference covariance matrix. The problem with the equation (9) is that the channel coefficients are not known at the receiver. Moreover, when the channel coefficients are estimated they are scaled by the transmit power of the pilot channel .

Therefore, C₁₁₁₁ must be scaled by multiplying it by the inverse of the transmit power of the pilot channel . One common function to improve the performance of the (W)CDMA is to use handover. The coverage area of the network is divided into cells and they are further divided into sectors . Often one cell includes three sectors . The soft handover occurs when the con- nection has to be passed from one sector to another sector within one cell or between two cells as the user moves. The soft handover (SHO) means that the connection is not broken at any stage but rather the second connection to the second sector is formed be- fore the first connection to the first sector is disconnected. During the soft handover a mobile terminal is situated in the overlapping area of two sectors belonging either to the same cell or two different cells. The communication between the mobile terminal and the base station (s) takes place concurrently via two radio links. The signals of both radio links are received at the mobile terminal and combined in the receiver.

Also SHO requires the ratio of pilot channel power and total transmit power, E_c/I_or. When the measurement method of the E_c/I_or of the pilot channel is made more accurate, the SHO will work more accurately. With good SHO functionality it is possible to avoid a situation where a mobile terminal penetrates deeply into an adjacent sector without being power-controlled by the latter. An accurate soft handover increases the capacity of the (W)CDMA system.

The power level of the pilot channel depends on the network configurations. For example, in WCDMA network the cell load can be balanced between the different cells by adjusting the common pilot power. Reducing the common pilot power causes the terminals to hand over to another cells, while increasing it will invite more terminals to hand over to the cell, as well as to make initial access to the network in that cell. Additionally, a load of one cell varies according to the number of served users and their services. Even if the common pilot power would be kept constant, its relative ratio compared to the total transmit power of a base station varies over time due to variations in the cell load. The problem is that in (W)CDMA system the transmitting power of the pilot channel is neither specified beforehand nor signaled to the ter- minal . Therefore, the ratio has to be estimated.

According to prior art, one possible method for estimating the E_c/I_or is to calculate a sum of powers of received pilot symbols (or, alternatively, channel estimates) and to divide the sum by the total received power of the received wideband samples. This method is considered a reference method hereafter. The reference method works for an AWGN (Additional White Gaussian Noise) channel and for a 1-path Rayleigh fading channel, but the method is significantly inaccu- rate in multipath fading conditions. Furthermore, the reference method suffers from bias, which is caused by noise and interference to the estimates of the power ratio. This happens especially on the edges of a cell or sector where the interference can be high. Cholesky decomposition is a well-known method in matrix theory. With Cholesky decomposition the given matrix A can be decomposed into two new matrices if certain conditions are valid. The Cholesky decomposition is mentioned for example in publication EP 1376380. In Cholesky decomposition the matrix is divided into an upper and a lower triangular matrix so that:

A = -LiI , (10)

where L is the lower triangular and L^τ is the upper triangular matrix. The lower triangular matrix means that all the elements above the diagonal of the matrix are zero. Respectively, all the elements below the diagonal of the upper triangular matrix are zero. The prior art includes means to test whether the decomposition is possible or impossible. The Cholesky decomposition can only be applied to a certain matrix if the matrix is square, symmetric, positive definite and with real entries.

The formulae for determining the elements for the decomposed matrix L are :

If the matrix A is not positive definite, the formulae will produce complex results because at some point of the calculation of the elements, the value under the square root in (11) will become negative.

According to matrix theory, if a matrix A is positive definite, the matrix A equivalently has eigenvalues, which are all positive. Generally, the ei- genvalues λ of the matrix A are defined by equation: det|A -λl| = 0 , (13 )

where I is a unit matrix.

The prior art methods for measuring the power ratio E_c/I_or give poor estimates, which lead to problems when these power values are used in the other functionalities of the receiver.

SUMMARY OF THE INVENTION The present invention discloses a method for measuring the ratio of a pilot channel power to a total transmit power of a transmitter, in a receiver. In a preferable embodiment the transmitter is a base station and the receiver is a mobile terminal device in a CDMA-based system. As explained before, the power ratio B_c/I_or is utilized by the receiver in the handover procedure. Also the IRC receiver can utilize the pilot power data. The pilots can be common pilots, dedicated pilots or channel estimates. The method for power ratio estimation uses mostly matrix operations. At a first stage of this embodiment according to the invention, wideband signal samples are received in a receiver. A sample covari- ance matrix of received wideband signal samples is calculated. At a second stage, the pilot symbols are detected and a symbol covariance matrix is calculated based on the pilot symbols.

At a third stage of this embodiment, an estimate of the power ratio, which is the quantity to be measured, is set. After that the signal covariance matrix is estimated by dividing the symbol covariance matrix with the power ratio estimate at a fourth stage. Following that at a fifth stage, an interfer- ence covariance matrix C_m is calculated by subtracting the signal covariance matrix estimate from the sample covariance matrix. Thus, the exact formula for

C_ml is the same as already mentioned in equation (9) :

r> — n

"" ⁿ EJT 'Mi ' (14)

where C_n. is the sample covariance matrix, C_ω is the symbol covariance matrix and E_c/I_or is the set power ratio estimate of the pilot channel.

The interference covariance matrix contains an error and therefore, the accuracy of the interference covariance matrix must be estimated in the sixth stage of this embodiment. An iteration round consist- ing the third (with changed power ratio estimate) , fourth, fifth and sixth stage is repeated as long as the minimum is not found. When the error is substantially accurately minimized, the iteration stops and the current power ratio estimate is the measured power ratio.

In one embodiment of the invention, the accuracy of the interference covariance matrix is estimated by investigating positive definiteness of the matrix. The positive definiteness of the matrix can be investigated by calculating the eigenvalues of the matrix and if they all are greater than zero, the matrix is positive definite.

Another embodiment of investigating the positive definiteness of the matrix is to decompose the matrix according to Cholesky decomposition. If the composition is possible into upper and lower triangular matrices, the matrix is equivalently positive definite. The decomposition can succeed or fail and this knowledge is the relevant information achieved from this stage rather than the decomposition matrices themselves . In one embodiment of the invention, the iteration round includes changing the power ratio estimate before calculating the interference covariance matrix and estimating the accuracy of that matrix. If the interference covariance matrix has been positive definite in the previous iteration round, the power ratio estimate is decreased for the next iteration round. Correspondingly, if the interference covariance matrix has not been positive definite in the previous iteration round, the power ratio estimate is increased for the next iteration round.

In one embodiment of the invention, a step size is defined for the power ratio estimate. The magnitude of the change (decrease or increase) of the power ratio estimate between two iterations is a step size. The iteration ends in this embodiment when two consecutive iterations include both the positive definite interference covariance matrix and the non- positive definite interference covariance matrix. Then the error is substantially accurately minimized. The inventive idea includes also a receiver, a circuit implementation and a terminal device implementation performing the claimed method steps. The calculation can also be implemented with software.

The basic advantage of the invention is the more accurate measurement result for the pilot channel power ratio. The method provides better variance and less biased estimates of the power ratio than the reference method disclosed in prior art. An another advantage with the present invention is a more accurate soft handover (SHO) procedure, which improves the performance of the system.

Another advantage with the present invention is that the method works with a receiver that has single receive antenna or with a receiver that supports antenna diversity.

Another advantage of the invention is that it is not limited to the 3rd generation mobile terminals only. The method can be used in any system that utilizes code division techniques and has a pilot channel .

BRIEF DESCRIPTION OF THE DRAWINGS

The accompanying drawings, which are included to provide a further understanding of the invention and constitute a part of this specification, illustrate the method and apparatus of one embodiment of the invention and together with the description help to explain the principles of the invention. In the drawings:

Figure 1 illustrates one embodiment according to the invention as a flow diagram showing the estimation procedure of the power ratio of a pilot channel power to a total transmit power in a receiver, Figure 2 discloses one embodiment of the invention by showing the iteration round of Figure 1 in more detail and including eigenvalue calculation,

Figure 3 discloses one embodiment of the invention by showing the iteration round of Figure 1 in more detail and including Cholesky decomposition,

Figures 4a and 4b represent the power ratio measured according to prior art and according to the present invention in one implemented system while the radio channel is the AWGN; with low SNR and high SNR, respectively,

Figures 4c and 4d represent the power ratio measured according to prior art and according to the present invention in the implemented system while the radio channel is a 2-tap fading channel; with low SNR and high SNR, respectively,

Figure 5 discloses three implementation possibilities for using the present invention with dif- ferent kinds of applications,

Figure 6 discloses a simple structural view of a terminal , and

Figure 7 discloses a simple example of the UMTS radio link components and the radio channel ob~ jects affecting to link quality.

DETAILED DESCRIPTION OF THE INVENTION

The present invention describes an efficient way of measuring the ratio of the pilot channel power to a total transmit power of the transmitter in a receiver. The measured power ratio can be used for example in IRC receiver or in the soft handover procedure in WCDMA system. The receiver is preferably implemented in the mobile terminal (concerning downlink) but the receiver can also be the base station (concerning uplink) . Later only WCDMA is referred in the examples but the invention is not limited for WCDMA system only. The present invention can be used for example in IS-95 or CDMA2000 systems. The problem of measuring the mentioned power ratio is closely related to the channel estimation because the channel estimates can be achieved from the pilot symbols, as discussed earlier in the background of the invention. As explained earlier in the text, the channel estimates in (W)CDMA can be expressed by:

(15)

where h is a vector containing coefficients of the propagation channel, n is a vector containing interference and noise and E_c /I_or is the power ratio of the pilot channel and the total transmit power of the base station.

In the equation (15) there is only one known quantity, h . On the contrary, two quantities of (15), E_c/lor and h, are unknown. The determination procedure of the E_c/l_Or is the core of the present invention. With the channel estimates it is possible to calculate a symbol covariance matrix of the detected pilot symbols as in the equation (4) :

C_hh=M* (16)

Similarly compared to the equation (16) , the received wideband signal samples in the chip level can

Λ be processed to a sample covariance matrix C_n.. It can be done as disclosed in the background of the inven- tion, equation (6) .

The problems of the pilot power determination are due to the fact that the power level of the pilot channel depends on the configuration of the used network and the operator can decide what the pilot chan- nel power is. The total power per chip of the base station, I_or, varies according to the number of users and their services. In other words, J_or is a characteristic of a cell load. Even though it can be assumed that the power of the pilot channel is kept constant, ■Bc/Ior varies over time due to variations on the cell load.

As described earlier in the text (equations 8 and 9), the actual desired signal, which was transmitted, creates error in the interference covariance matrix. The principal issue in the present invention is that the correct power ratio estimate of E_c/I_or minimizes the error the desired signal covariance matrix produces in the interference covariance matrix. Therefore, according to the invention, the narrowband interference covariance matrix is estimated by using the received wideband sample covariance matrix and covariance matrix of pilot symbols (or, alternatively, the channel estimates) , which is scaled by the E_c/I_or according to equation 9. The core of the invention is that once the interference covariance matrix has been produced with some power ratio estimate (E_c/I_or) , the accuracy of the interference covariance matrix will be checked. According to the invention, the accuracy will be checked by investigating positive definiteness of the matrix. When the power ratio is correctly estimated, the error is at its minimum (i.e. the desired signal is subtracted correctly from the sample covari- ance matrix) , and the interference covariance matrix is positive definite, yet close to be non-positive definite .

A preferred embodiment of the method according to the invention has the following sequence of steps. This embodiment is illustrated as a flow chart in Figure 1. At first wideband signal samples are received in a receiver. This chip level sampled signal is a basis for calculating a sample covariance matrix

C_n. 10. The knowledge of the allocated delays in the receiver is included in the sample covariance matrix C_n.. A second step is to calculate a symbol covariance matrix C_hh 11 from the pilot symbols (or alternatively, the channel estimates) achieved by the re- ceiver.

A third step is to set an initial estimate for the power ratio {E_c/I_or) 12, which is to be determined in this invention. Naturally it is preferable if the initial power ratio estimate is within the range where it practically would be in any real network configuration.

As a start of an iteration round, a fourth step 13 is performed where a signal covariance matrix is estimated by using the previously achieved data. The signal covariance matrix is achieved by dividing the symbol covariance matrix by the set power ratio estimate .

The following step is to calculate the inter- ference covariance matrix C_uu as expressed in equa- tions (9) and (14) . Thus, the signal covariance matrix estimate is subtracted from the sample covariance matrix to produce the interference covariance matrix.

As mentioned earlier, the interference covariance matrix contains an error if the power ratio is set incorrectly and when the power ratio is correctly set, the error would be minimized. Therefore in the present invention the accuracy of the interference matrix must be estimated 15. This can be done by calculating the error itself or with some other method or algorithm. The proposed inventive method uses positive definiteness of the interference covariance matrix to check when the error is minimized. This is explained in more detail with Figures 2 and 3. When the accuracy is estimated, it must be checked whether the substantially accurate interference covariance matrix has been found 16. If the error is not yet minimized, the method changes the power ra- tio estimate 17 and jumps back to signal covariance matrix calculation step 13. The magnitude of change in the power ratio 17 can be based on the error found out in step 15. Another possibility is to change the power ratio 17 with a given step size. The iteration is performed until the substantially error-free interference covariance matrix has been found. At that stage, the power ratio used with the C_uu with error minimum, is determined to be the measured power ratio of pilot channel power to a total transmit power 18.

A more detailed description of the accuracy estimating step 15 is presented in the following. Two embodiments of the present invention are shown and references are made to Figures 2 and 3. Figure 2 presents finding the error minimum by calculating eigenvalues. The following is based on the matrix theory whereby a positive definite matrix has equivalently only positive eigenvalues. The interference covariance matrix is estimated by performing steps 10-14 of Figure 1. In Figure 2 only the last two steps 13, 14 are shown, which are the signal covariance matrix estimation and the interference covariance matrix estimation. The steps 20-23 of Figure 2 are functioning equivalently with steps 15-17 of Figure 1. The idea of the iteration round 13, 14, 20-23 is to find a positive definite interference covariance matrix, which is very close to be non-positive definite. This condition is achieved when at least one eigenvalue of the matrix is very close to zero. Thus, the eigenvalues of the interference covariance matrix are calculated in step 20. In the next step 21 the calculated eigenvalues are examined. If every eigenvalue is clearly positive, the used power ratio estimate is too big. Therefore the power ratio estimate must be decreased 22 for the next iteration round. Respectively, if at least one eigenvalue is clearly negative, the used power ratio estimate is too small. In this case, the power ratio estimate must be increased 23 for the next iteration round. The iteration continues by calculating new signal and interference covariance matrix estimates 13, 14 on the new power ratio estimate and calculating the eigenvalues 20 again.

The iteration goes on until at least one ei- genvalue is very close to zero. Thus, a limit can be set for the absolute value of the eigenvalues. This limit absolute value can be chosen according to the accuracy requirements of the power ratio estimation.

When the eigenvalue condition succeeds, in other words, the absolute value of at least one eigenvalue is below the limit, the iteration ends and the used power ratio is chosen to be the measured power ratio of the pilot power and the total transmit power 18.

One numerical example of Figure 2 is pre- sented in the following. The power ratio is set for instance to 0,08 in step 12. The matrices are calculated with the help of received signal samples and the first power ratio estimate. The resulting matrix C_uu is finally achieved in step 14 as discussed earlier. The eigenvalues of C_uu are λi, ... , λ_L, where L is the number of allocated delays. Assume the eigenvalues, which are sorted in increasing order, to be λ =

[-0,5; 1; 1,5; ... ; 10]. Then we assume the limit to be 0,01 in this example. Because none of the eigenvalues' absolute value is smaller than 0,01, and one eigenvalue is negative, must the power estimate be increased, by choosing the step 23 of Figure 2. The new power ratio can be chosen to be 0,10. The signal and interference covariance matrices are estimated again 13, 14 and the eigenvalues of the new interference covariance matrix is calculated 20. The resulting eigenvalues are in this example λ = [0,008; 0,5; 1; 1,5; ... ; 5] . The following decision is that at least one eigenvalue is close to zero because the smallest absolute value 0,008 is smaller than the limit 0,01. The iteration stops and the power ratio is achieved with the result E₀/I_or ~ 0,10. Figure 3 presents finding the error minimum by performing the Cholesky decomposition. The Cholesky decomposition is known from prior art and disclosed in detail in the background of the invention in context with equations (10) , (11) and (12) . The Cholesky de- composition is an equivalent procedure to the eigenvalue calculation in the perspective of this invention. The interference covariance matrix is estimated here as well by performing steps 10-14 of Figure 1. In Figure 3 only the last two steps 13, 14 are shown which are the signal covariance matrix estimation and the interference covariance matrix estimation. The steps 30-33 of Figure 3 are functioning equivalently with steps 20-23 of Figure 2. Also here the idea of the iteration round 13, 14, 30-33 is to find a posi- tive definite interference covariance matrix, which is close to be non-positive definite. This condition is achieved when the Cholesky just succeeds but with minor modifications to the elements of the matrix, it would fail. The condition is also achieved when the Cholesky just fails but with minor modifications to the elements of the matrix, it would succeed. Thus, the Cholesky decomposition on the interference covari- ance matrix is tried 30. Next, the succession of the Cholesky decomposition is examined in step 31. This succession data is the relevant part of examination and the decomposition matrices themselves are not relevant in this embodiment. If the decomposition succeeds clearly, the used power ratio estimate is too big. Therefore it must be decreased 32 for the next iteration round. Respectively, if the decomposition fails, the used power ratio estimate is too small. In this case, the power ratio estimate must be increased 33 for the next iteration round. The iteration contin- ues by calculating new signal and interference covari- ance matrix estimates 13, 14 on the new power ratio and trying the Cholesky decomposition 30 again.

The iteration goes on until the previously described target result has been reached, in other words, the decomposition still succeeds but is just about to fail. The accuracy requirements of the power ratio estimation define, how close should the theoretical limit between succession and failure be from the actual decomposition result. When the decomposi- tion succeeds but is just about to fail, the iteration ends and the used power ratio is chosen to be the measured power ratio of the pilot power and the total transmit power 18. The success of the decomposition means that the interference covariance matrix is posi- tive definite and the failing decomposition indicates the matrix to be non-positive definite. Thus, in the end of the iteration, the interference covariance matrix is positive definite but very close to be non- positive definite. The basis for the procedure in step 31 is that the Cholesky decomposition always succeeds and the interference covariance matrix is clearly positive definite if the estimate of B_c/I_or is high enough. Re- spectively, the decomposition process always fails and the interference covariance matrix is clearly non- positive definite if the estimate of E_c/I_or is too low. Within the limit of a successful and non-successful Cholesky decompositions, the power ratio result 18 gives a good estimate of the actual power ratio value.

In one embodiment of the invention, the method is used in a CDMA or WCDIVLA. communication system. The technical specifications for the third generation mobile system stand as the main information source for CDMA and WCDMA technology.

In one embodiment of the invention, at least one of the functional means configured to implement the method steps, is implemented in at least one of a programmable device, dedicated hardware, programmable logic and any other processing device. In addition, it is possible to use combination of the aforementioned devices. An Application Specific Integrated Circuit (ASIC) and Digital Signal Processing (DSP) unit are examples of these. The estimation calculation frequency can be chosen relatively freely. The changes in the base station load affect the calculation frequency.

Measurement results concerning the present invention and the reference method from prior art in one implemented system are presented in Figures 4a-4d.

The. estimated pilot channel power ratio value measured in an AWGN channel and low SNR is presented in Figure 4a with lots of separate measurements shown in the same figure. The same measurement with 12 dB higher SNR is presented in Figure 4b. Diversity antenna reception is used in the system, which in this case means two receiving antenna branches. The calculated pilot channel power ratio according to prior art is presented by histograms 40, 42. The present invention gives the results for histograms 41, 43. The x-axis represents the pilot channel power ratio (which is actually set to 0,1 in this system) and the y-axis shows the probability distribution of the resulting power ratio values. The present invention has better (smaller) variance and standard deviation than the reference method according to prior art. Furthermore, the present invention gives less biased values (closer to 0,1) than the reference method. When the SNR in- creases, both methods approach the actual pilot channel power ratio value (= 0,1) .

The distribution of the estimates with a mul- tipath propagation conditions is presented in Figures 4c and 4d. The prior art method is shown as histograms 44, 46 and the present invention is shown as histograms 45, 47. In Figure 4c the SNR is 12 dB lower than in Figure 4d. As shown, the present invention works clearly better in a multipath environment compared to the reference method. Figure 5 shows a few embodiments where the present invention can be used. All the features of the invention are included in this example in calculation means 50 which estimate the ratio of a pilot channel power and a total transmit power as described earlier in detail. The receiving means for receiving wideband signal samples can be implemented in various ways. Such applications are the rake receiver 51, the IRC receiver 53 and the soft handover procedure 55 (SHO) of a receiver 54, for example. Each receiver implemen- tation is a part of the mobile terminal (not included in the figure) , which communicates with the base station over the radio channel by using the antennas 52. The receiver implementations are explained in the background of the invention in more detail. The calculation means 50 can be a programmable device, dedicated hardware, programmable logic and any other processing device, or a combination of the previous means, as mentioned earlier. Figure 6 presents one embodiment of a terminal according to the invention in a simple form. The terminal 60 includes a circuit 61, which performs the functionalities of the invention. The terminal 60 has two antenna branches 62 in this example for making the diversity reception possible.

Figure 7 shows the radio channel 70 between the mobile terminal 71 and the base station 72. The base station 72 is shown as Node B as part of the UMTS network. The antenna 73 of the Node B is in the other end of the radio link and the user's mobile terminal 71 is in another end. The radio channel 70 includes different kinds of scatterers . These scatterers can be fixed physical barriers such as walls of buildings 75 or smaller particles such as cars 74. It is obvious to a person skilled in the art that with the advancement of technology, the basic idea of the invention may be implemented in various ways. The invention and its embodiments are thus not limited to the examples described above, instead they may vary within the scope of the claims.

Claims

1. A method for measuring a ratio of a pilot channel power to a total transmit power of a transmitter, in a receiver, comprising the steps of: receiving wideband signal samples in the receiver; calculating a sample covariance matrix from the received wideband signal samples; detecting pilot symbols from the received wideband signal samples; calculating a symbol covariance matrix of the detected pilot symbols; setting an estimate of the power ratio; producing an estimate of signal covariance matrix by dividing the symbol covariance matrix by the power ratio estimate; estimating an interference covariance matrix, which contains an error, by subtracting the estimate of signal covariance matrix from the sample covariance matrix; estimating accuracy of the interference covariance matrix; repeating the setting step by changing the power ratio estimate, the signal covariance matrix producing step, the interference covariance matrix estimating step and the accuracy estimating step until the error is substantially accurately minimized; and deciding the power ratio estimate used with the error minimum estimate to be the measured power ratio.

2. The method according to claim 1, further comprising the step of: estimating accuracy of the interference covariance matrix by investigating positive definiteness of the interference covariance matrix.

3. The method according to claim 2, further comprising the step of : investigating the positive definiteness of the interference covariance matrix by checking if all eigen- values of the interference covariance matrix are greater than zero.

4. The method according to claim 2 , further comprising the step of: investigating the positive definiteness of the in- terference covariance matrix by checking if the interference covariance matrix can be decomposed into upper and lower triangular matrices according to the Chole- sky decomposition.

5. The method according to claim 2 , further comprising the step of: performing the setting step, comprising the steps of: decreasing the estimate of the power ratio when the interference covariance matrix is positive defi- nite; and increasing the estimate of the power ratio when the interference covariance matrix is not positive definite .

6. The method according to claim 5, further comprising the steps of: defining a step size for the power ratio estimate; changing the power ratio estimate by the defined step size; and performing the deciding step when two consecutive interference covariance matrix estimates include both the positive definite interference covariance matrix and the non-positive definite interference covariance matrix, thus minimizing the error substantially accurately.

7. The method according to claim 1, wherein the pilot symbols are common pilot symbols .

8. The method according to claim 1, wherein the pilot symbols are dedicated pilot symbols.

9. The method according to claim 1, wherein the pilot symbols are channel estimates.

10. A receiver for measuring a ratio of a pilot channel power to a total transmit power of a transmitter, in a communication system, comprising: receiving means configured to receive the wideband signal samples; and calculating means configured to: calculate a sample covariance matrix from the received wideband signal samples; detect pilot symbols from the received wideband signal samples; calculate a symbol covariance matrix of the detected pilot symbols; set an estimate of the power ratio; produce an estimate of signal covariance matrix by dividing the symbol covariance matrix by the power ratio estimate; estimate an interference covariance matrix, which contains an error, by subtracting the estimate of signal covariance matrix from the sample covariance matrix; estimate accuracy of the interference covariance matrix; repeat the setting step by changing the power ratio estimate, the signal covariance matrix producing step, the interference covariance matrix estimating step and the accuracy estimating step until the error is substantially accurately minimized; and decide the power ratio estimate used with the error minimum estimate to be the measured power ratio.

11. The receiver according to claim 10, further comprising: the calculating means configured to estimate accuracy of the interference covariance matrix by investigating positive definiteness of the interference covariance matrix.

12. The receiver according to claim 11, fur- ther comprising: the calculating means configured to investigate the positive definiteness of the interference covariance matrix by checking if all eigenvalues of the interference covariance matrix are greater than zero.

13. The receiver according to claim 11, further comprising: the calculating means configured to investigate the positive definiteness of the interference covariance matrix by checking if the interference covariance matrix can be decomposed into upper and lower triangular matrices according to the Cholesky decomposition.

14. The receiver according to claim 11, further comprising the calculating means configured to: perform the setting step, wherein the calculating means is configured to: decrease the estimate of the power ratio when the interference covariance matrix is positive definite; and increase the estimate of the power ratio when the interference covariance matrix is not positive definite.

15. The receiver according to claim 14, further comprising: the calculating means configured to define a step size for the power ratio estimate; the calculating means configured to change the power ratio estimate by the defined step size; and the calculating means configured to perform the deciding step when two consecutive interference co- variance matrix estimates include both the positive definite interference covariance matrix and the non- positive definite interference covariance matrix, thus minimizing the error substantially accurately.

16. An electric circuit for measuring a ratio of a pilot channel power to a total transmit power of a transmitter, in a receiver, configured to: receive wideband signal samples in the receiver; calculate a sample covariance matrix from the received wideband signal samples; detect pilot symbols from the received wideband signal samples; calculate a symbol covariance matrix of the de- tected pilot symbols; set an estimate of the power ratio; produce an estimate of signal covariance matrix by dividing the symbol covariance matrix by the power ratio estimate; estimate an interference covariance matrix, which contains an error, by subtracting the estimate of signal covariance matrix from the sample covariance matrix; estimate accuracy of the interference covariance matrix; repeat the setting step by changing the power ratio estimate, the signal covariance matrix producing step, the interference covariance matrix estimating step and the accuracy estimating step until the error is substantially accurately minimized; and decide the power ratio estimate used with the error minimum estimate to be the measured power ratio.

17. The electric circuit according to claim

16, further configured to: estimate accuracy of the interference covariance matrix by investigating positive definiteness of the interference covariance matrix.

18. The electric circuit according to claim

17, further configured to: investigate the positive definiteness of the interference covariance matrix by checking if all eigenvalues of the interference covariance matrix are greater than zero.

19. The electric circuit according to claim 17, further configured to: investigate the positive definiteness of the interference covariance matrix by checking if the inter- ference covariance matrix can be decomposed into upper and lower triangular matrices according to the Chole- sky decomposition.

20. The electric circuit according to claim 17, further configured to: perform the setting step, wherein the electric circuit is configured to: decrease the estimate of the power ratio when the interference covariance matrix is positive definite; and increase the estimate of the power ratio when the interference covariance matrix is not positive definite .

21. The electric circuit according to claim 20, further configured to: define a step size for the power ratio estimate; change the power ratio estimate by the defined step size; and perform the deciding step when two consecutive interference covariance matrix estimates include both the positive definite interference covariance matrix and the non-positive definite interference covariance matrix, thus minimizing the error substantially accurately.

22. A terminal device for measuring a ratio of a pilot channel power to a total transmit power of a transmitter, in a receiver, comprising: an electric circuit, comprising: receiving means for receiving wideband signal samples ,- calculating means for calculating a sample covariance matrix from the received wideband signal sam- pies; calculating means for detecting pilot symbols from the received wideband signal samples,- calculating means for calculating a symbol covariance matrix of the detected pilot symbols; calculating means for setting an estimate of the power ratio; calculating means for producing an estimate of signal covariance matrix by dividing the symbol co- variance matrix by the power ratio estimate; calculating means for estimating an interference covariance matrix, which contains an error, by subtracting the estimate of signal covariance matrix from the sample covariance matrix,- calculating means for estimating accuracy of the interference covariance matrix; calculating means for repeating the setting step by changing the power ratio estimate, the signal co- variance matrix producing step, the interference covariance matrix estimating step and the accuracy estimating step until the error is substantially accurately minimized; and calculating means for deciding the power ratio estimate used with the error minimum estimate to be the measured power ratio.

23. The terminal device according to claim

22, further comprising the electric circuit comprising the calculating means for: estimating accuracy of the interference covariance matrix by investigating positive definiteness of the interference covariance matrix.

24. The terminal device according to claim

23, further comprising the electric circuit comprising the calculating means for: investigating the positive definiteness of the interference covariance matrix by checking if all eigenvalues of the interference covariance matrix are greater than zero.

25. The terminal device according to claim

23, further comprising the electric circuit comprising the calculating means for: investigating the positive definiteness of the interference covariance matrix by checking if the inter- ference covariance matrix can be decomposed into upper and lower triangular matrices according to the Chole- sky decomposition.

26. The terminal device according to claim 23, further comprising the electric circuit comprising the calculating means for: performing the setting step, wherein: decreasing the estimate of the power ratio when the interference covariance matrix is positive definite; and increasing the estimate of the power ratio when the interference covariance matrix is not positive definite.

27. The terminal device according to claim 26, further comprising the electric circuit comprising the calculating means for: defining a step size for the power ratio estimate; changing the power ratio estimate by the defined step size; and performing the deciding step when two consecutive interference covariance matrix estimates include both the positive definite interference covariance matrix and the non-positive definite interference covariance matrix, thus minimizing the error substantially accurately.