HK1079006A - Method and apparatus for determining signal-to-interference ratio with reduced bias effect - Google Patents

Description

Method and apparatus for determining sir with reduced bias effect

Technical Field

The present invention relates to a method and apparatus for finding a signal-to-interference ratio (SIR) in a digital communication system. And more particularly to SIR estimation with reduced bias contribution.

Background

SIR measurement is an important metric for the performance of digital communication systems. For wireless communication systems, such as third generation (3G) wireless systems, SIR measurements are used in many link adaptation Techniques (LinkAdaptation Techniques) such as transmit power control and adaptive modulation and coding. Typically, SIR measurements at the receiving device are more meaningful than at the transmitting device because SIR measurements at the receiving device directly reflect the quality of the communication link signal, especially in the presence of multiple access interference or multipath fading channels.

By definition, a received signal contains a desired signal and interference. The interference may include other signals and thermal noise (thermal noise) at the receiving end. However, the receiving device generally has no information about the signal power or the interference power, so the receiving device needs to perform signal and interference power estimation based on the received signal using a blind method (blind method). In SIR measurement, the blind approach for a given received signal refers to the signal power and interference power (finally SIR) obtained only from the observed samples of the received signal without any training sequence or any prior information of the desired signal and interference in the received signal.

There are many ways to perform the SIR measurement of the reception. In the prior art, the received signal is averaged over time to estimate the signal power of a given signal, and the total power of the received signal is measured and then subtracted from the estimated signal power of total power to estimate the interference power. The SIR is then determined as the ratio between the estimated signal power and the interference power.

The SIR estimation for a given received signal may be performed at different points of view of the receiver structure, such as at the receiver antenna, at the input of the data demodulator, or at the output of the data demodulator. However, SIR estimation measurements at different locations typically have different levels of accuracy because the signal gain or the amount of interference at one measurement location may be different from readings at other locations.

The main problem with measuring the SIR of the data signal is that the SIR measurement may deviate from the corresponding true SIR value. Such errors in SIR measurements are due to the following two main reasons. First, a signal and its interference cannot be completely separated. Second, the desired signal is data-modulated, so the SIR measurement is done using a blind method, i.e., without previous data for the data signal. This increases the uncertainty in the measured signal power.

In many prior art systems, SIR measurements rely primarily on an averaging filter to calculate signal and noise power, which results in an unexpectedly large bias result. Generally, when the SIR value is small, the SIR measurement becomes more overestimated, mainly due to a larger bias result.

Typically, the kth demodulation codeElement, y_kWhen an input is to a demodulator based on the SIR estimator, it can be expressed as:

equation one

Wherein s is_k ^dRepresents the kth demodulated QPSK signal, and n_k ^eRepresenting the total effective interference (including residual intra-cell interference, inter-cell interference and background noise contributions). Then the average signal power P_SAnd effective interference power P_IFor example, the SIR is measured as:

equation two

By comparing equation two with the SIR definition (i.e., RSCP) used in 3GPP^*SF/interference), the measured RSCP and ISCP are not unambiguously determined. In other words, equation two expresses SIR measurement for a DPCH more clearly than the definition of 3 GPP. In addition, because the SIR measurements are performed in the data portion of the received signal, a blind (blind) estimation is required due to the unknown transmitted signal at the receiving device.

The SIR measurement in equation two is performed on the demodulator output when the SIR definition used in 3GPP implies that it is independent of the data demodulation type used in the receiving device. Thus, the SIR given by equation two may be different for different demodulator formats. For example, for a given received signal that is primarily distorted by interference, the SIR measured in a conventional matched filter receiver may be less than that in an advanced receiver, such as a Canceller (canceler), due to reduced interference effects. Note that the SIR at the demodulated output is the primary determinant of communication link performance. However, SIR measurements on the data portion of the received signal must deal with unknown transmitted data.

FIG. 1 depicts a typical constellation of transmitting QPSK signals, where E_sRepresenting the transmitted QPSK symbol energy. For wireless systems, such as 3GPP systems, after spreading the QPSK signal, the resulting spread signal is achieved over a wireless channel at the receiver. The received signal is then processed by the demodulator, which provides demodulated symbols y_kFrom k 1, 2_burstIn which N is_burstIs the number of symbols in a data burst (burst) of the received signal.

Considering the impulse of fading channel (fading channel) and the demodulator gain, in the absence of significant interference, the typical constellation of soft-valued demodulated symbols can be observed on average as shown in fig. 2, where S is_mRepresenting the mth demodulated signal symbol.

A typical demodulated output symbol may be as shown in fig. 3 when interference is present. For a given transmission symbol, S_kWhose output symbols may fall at any point in the QPSK constellation centered on the associated average demodulated symbol, E { S }_k ^d}. In this case, a blind based average power estimation on the demodulator output will be performed. When making each sent demodulated symbol decision, some erroneous decisions may occur due to the effective interference and fading channel concerns. For example, as shown in FIG. 3, even though S is actually S₂ ^dIs sent to the k-th symbol, which interference may result in the demodulated output symbol, denoted y_kBecomes more than actually transmitted symbolS₂ ^dCloser to S in the first quadrant₁ ^d. As a result, one may make a decision at y_kThe determination of the error (i.e., a determination error). This decision error is a major source of error in the average signal power estimation and thus causes the SIR estimate to be overestimated. At lower SIR ranges (high raw BER ranges), the average signal power (or SIR) estimate may be more overestimated.

It is therefore desirable to provide a method of performing SIR estimation that obviates the disadvantages of the prior art methods.

Disclosure of Invention

The present application provides a method and apparatus for estimating SIR that is more accurate than the estimates available in the prior art. In an exemplary case, the present application implements a method that preferably uses the demodulator output for SIR estimation, noting that the primary determinant of communication link performance appears to be the SIR at the data demodulator output. The method reduces the bias in the SIR estimate resulting in the SIR estimate being as close as possible to the true SIR.

The method of the present invention preferably uses a median filter and an average filter and combines the outputs from the median and average in the SIR estimation. It is also advantageous that a correction term as a function of the mean and median number is introduced to further mitigate the effects of the bias.

In connection with the preferred embodiment of the present application, a new SIR estimator instrument based on the use of data symbols is provided. An output from a data demodulator, such as a Rake output or a multi-user detection (MUD) output, is input to the SIR estimator. As indicated above, the advantage of using a data demodulator output as the output is that the demodulator output directly and efficiently reflects the quality of the received signal. In particular, when SIR measurements are used in connection control techniques, such as power control, such a data demodulator based on SIR estimation is highly desirable. In addition, the proposed SIR estimator device is able to reduce the effects of bias on SIR estimation, resulting in a more reliable and accurate SIR estimate than conventional techniques.

Drawings

The present application will become more fully understood from the detailed description and the accompanying drawings, wherein:

FIG. 1 is a representative signal set for transmitting QPSK symbols for one of the present applications;

fig. 2 is a typical signal set for average demodulation of QPSK symbols for the present application;

FIG. 3 is a representative spatial representation of demodulated symbols in the presence of interference;

FIG. 4 is a functional block diagram of a SIR estimator operating in accordance with a preferred embodiment of the present application;

FIG. 5 is a block diagram of a system including the SIR estimator of FIG. 4; and

fig. 6 is a flow chart of method steps performed for a SIR estimator including the SIR estimator of fig. 4.

And (5) contracting the words:

3 GPP: third generation partnership project (third generation partnership project)

BER: block error Rate (block error rate)

BPSK: binary phase shift key (binary phase shift keying)

DPCH: dedicated physical channel (dedicated physical channel)

ISCP: interference signal code power (interference signal code power)

MUD: multi-user detection (Multi-user detection)

PSK: phase shift key (phase shift keying)

QAM: quadrature amplitude modulation (quadrature amplitude modulation)

QPSK: quadrature phase shift key (quadrature phase shift keying)

RSCP: received signal code power (received signal code power)

SF: spread spectrum factor (spreading factor)

SIR: signal-to-interference ratio (signal-to-interference ratio)

UE: user equipment (user equipment)

Detailed Description

The preferred embodiment of the present application is described below, providing a novel SIR estimation process based on a data demodulator output. The present application also provides a SIR estimation apparatus. By definition, the data demodulator output term means the output provided at the last stage of the data demodulator under consideration. The data demodulator processes the received baseband signal and provides soft-valued (soft-valued) estimates of the transmitted symbols. The estimated symbols are further processed by other receiver functions, such as a channel decoder, in order to resolve the receiving device that transmitted the data information.

In the context of 3GPP systems, the demodulator may be configured as a multi-user detection (MUD) receiver or a single-user detection (SUD) receiver, such as a matched filter, Rake receiver and equalizer (equalizer). Even though Binary Phase Shift Key (BPSK) and Quadrature Phase Shift Key (QPSK) demodulation methods are presented in the preferred embodiments of the present application, the present application is applicable to higher order demodulation methods such as 8-PSK and 16-QAM.

The present application estimates the average signal power such that the effects of skew are reduced. When transmitting QPSK, the average signal power y output by the demodulator_kCan be estimated as follows:

equation three

Wherein Q_iRepresents the ith quadrant in the QPSK signal set; n is a radical of_QiRepresents the number of demodulator outputs belonging to the ith quadrant after individual blind symbol-based decisions are made; and y is_k(Q_i) Is the kth output symbol in the ith quadrant.

Equation three is used to determine the average number of demodulator output symbols in each quadrant of the QPSK set. Second, equation three determines the average signal power based on the amount of average signal points in the individual quadrants. This two-step averaging technique can provide a good estimate over a fairly high SIR range (equivalently low symbol error rate). However, as mentioned earlier, when the actual SIR is low, the average SIR value is biased (overestimated) due to more symbol decision errors, which also results in overestimated SIR values (see equation two). To reduce the effect of bias in the signal power estimation, another statistical parameter, called the "median" (center of a distribution), is used as detailed below.

The mean and median are symmetrically distributed. Thus, due to the high SIR values, the average and median number of demodulator outputs in each quadrant is exactly the same, since the interference experienced in each quadrant can approximate a normal distribution over a range of SIRs.

The sensitivity of the median to the maximum sample value is less than the mean. The characteristic of the median may be such that the median based on average power is closer to the true average power than the mean based on average power, particularly at high skew distributions (skewed distributions) such as Log normal distribution (Log normal distribution), or when the SIR is low and the distribution of demodulator output samples in each quadrant is close to the skew distribution.

The standard deviation of the median of the large samples with normal distribution is larger than the mean. The median is more susceptible to sampling variations (fluctuations). Therefore, when the number of random number samples having a normal distribution is large, the standard deviation of the median is generally larger than that of the average number.

Taking into account the statistical nature of the above average and median, the present application determines the average signal power estimate as the smallest function value between the average and median, as follows:

equation four

Wherein mean (y)_k(Q_i) And mean (y)_k(Q_i) Individually represent in the ith quadrant Q_iBecause of y, the median and mean of the output symbols of the demodulator_k(Q_i) Is a complex symbol and mean (y)_k(Q_i))＝median(real(y_k(Q_i)))+j.median(imag(y_k(Q_i) ()) and the same mean (y)_k(Q_i))＝mean(real(y_k(Q_i)))+j.mean(imag(y_k(Q_i))). This represents that the average signal power is equal to the square of the minimum absolute value of the median and mean averaged over all quadrants. In equation four, the main reason for finding the minimum value between the median and the average value of the demodulated symbol is as follows. The main reason for using the minimum between the median and mean is to reduce the effect of bias. Note that the estimated SIR value of the present application cannot be greater than the mean estimated SIR alone, since the minimum between the median and mean is the medianThe smallest of the number and average.

Selecting the median as the minimum reduces the bias effect in estimating the average signal power, particularly in the low SIR range. On the other hand, selecting the average as the minimum compensates for the disadvantages in median calculations, such as being subject to sample variation. Thus, by combining the median and average of the output symbols of the demodulator, which is effective for each quadrant of the QPSK set, the estimation performance of the averaged signal is substantially improved. Even though the preferred embodiment of the present application employs the minimum between the median and the mean, other combined values from the median and the mean may be used to determine the average signal power estimate. For example, one weighting (combining) method is as follows:

equation five

Wherein 0 < α < 1.

Then, the average effective interference power based on equation four is estimated. From equation one and equation four, the average effective interference power can be expressed as:

equation six

WhereinIs regarded as the amplitude of the average signal, andrepresenting four QPSK symbols with unit energy.

Now SIR estimation has to be performed. From equations four and six, the SIR estimate based demodulator can be expressed as:

equation seven

This SIR estimate has been confirmed by link-level simulations, which indicate that the SIR estimation performance based on equation seven is acceptable within a reasonably operating SIR range. But in the low SIR range (e.g., from 5dB to 0dB or less), bias effects due to symbol errors still occur in the SIR estimate, causing the estimated SIR to deviate from the true SIR. The minimum value cannot fully estimate the bias effect especially in the low SIR range because some symbol decision errors may be present in blindly deciding the relevant basic symbol decisions. In this case, some of the modifications in equation seven are needed to comply with the current requirements standard of 3GPP working group four (WG 4). The application (including the correction terms) is beyond this requirement.

By means of a heuristic method by Monte-Carlo simulation, a correction term is introduced into the numerator of the above equation (single power term) as a function of the deviation between the median and mean of the calculation, as follows:

equation eight

Wherein

The basis for using such a correction term is in a high SIR range, where the median and mean values are as close to each other as possible. The correction term can therefore be ignored insofar as it is not needed (because without it the estimated SIR is already within the accuracy requirements). However, when the true SIR is low, the skew distribution due to the symbol error effect of the demodulator output samples may cause the correction term to increase, since the difference between the corresponding median and mean values may gradually increase. Also, the correction term may help to reduce the bias effect on the SIR estimate when the estimated signal power (and ultimately SIR) is overestimated (biased).

Although the SIR measurement method described above is derived on the assumption that the demodulator output is a complex-valued QPSK symbol string, for an actual MUD implementation, the MUD provides a real data bit string with each pair of two consecutive data bits, which can be mapped to a complex symbol, such as QPSK modulation in a transmitter.

Fig. 4 shows a block diagram of an SIR estimator 400 that uses data bits as input instead of QPSK symbols in accordance with a preferred embodiment of the present application. The SIR estimator includes an input port 405 for inputting the input bit string, a hard limiter 410, a multiplier 415, a median filter 420, a first average filter 425, a minimization process block 430, a signal power process block 435, a correction term process block 440, a second average filter 445, an SIR calculation process block 450, a summer and comparator 455, 460, and a process block 465.

The input port 405 receives a bit string of soft values. The SIR estimator 400 processes the absolute value of the symbol input to the SIR estimator in the form of a bit string which is received via the input port 405. The bit string is arranged to the hard slicer 410 and the multiplier 415. The hard slicer 410 provides a +1 bit to the multiplier 415 if a soft-valued bit is greater than or equal to zero. In addition, the hard limiter 410 provides a-1 bit to the multiplier 415. The multiplier 415 then inputs each soft-valued input bit r_iMultiplying by the corresponding hard-limited bit to obtain the absolute value | r of the individual input bit_i|。

The absolute value | r_iI confirm that has been made at each input bit r_iAnd if the resulting bit decision (resetting bit decision) becomes-1, then the input bit is phase shifted by 180 degrees. Furthermore, the input bit remains unchanged. Therefore, the calculation of the average signal power and the interference power is based on a blind basis decision. The multiplier 415 outputs the absolute value | r_iTo the median filter 420, the first average filter 425, and the sum/comparator 460. Based on a number of consecutive samples, the median filter 420 and the first mean filter 425 individually determine the median and mean of the string of absolute values. From the median filter 420 (m)_d) And the first average number filter 425 (m)_e) Is compared in the minimization process block 430 to determine the mean power value m based on the median_dAnd average power value m based on the average_eThe minimum value m in between. The correction term processing block 440 also receives the correction term from the median filter 420(m_d) And the first average number filter 425 (m)_e) And determining a correction term C, wherein:

C＝|m_d-m_e|²equation nine

The output m of the minimization process block 430 is arranged to the signal power process block 435 and the sum/comparator 460. The sum/comparator 455 compares the output Ps of the signal power processing block 435 with the correction term C, where:

P_s＝(m)²equation ten

To determine the average interference power, the processing block 465 first receives the sum/comparator 460 output and performs (| r)_i|-m)²Wherein the function solves for interfering components from the output bit string. The second averaging filter 445 receives the output of the processing block 465 and combines P_NOutput to the SIR calculation processing block 450. The SIR is calculated by the SIR calculation processing block 450 based on the sum/comparator 455 and the second average filter 445, where:

equation eleven

Fig. 5 shows a system 500 comprising a demodulator 505 and a known soft symbol to soft bit mapper (mapper)510, wherein the mapper 510 inputs a string of soft-valued bits to the SIR estimator 400. The estimator 400 can also be used for higher order demodulation such as 8-PSK, 16-QAM, and 64-QAM, provided that the complex-valued demodulated symbols are converted to soft-valued bits by the soft-symbol-to-soft-bit correspondences 510.

Fig. 6 is a flow chart of a method 600 performed by SIR estimator 400. The SIR estimator blockDetermining SIR of a symbol/bit by receiving the symbol/bit (step 605), estimating the average signal power of a symbol/bit as the median-based average power value m of the symbol/bit_dAnd an average power value m based on the average value_eThe average effective interference power of the symbols/bits is estimated (step 615) and the SIR is calculated by dividing the estimated average signal power of the symbols/bits by the estimated average effective interference power of the symbols/bits (step 620), the function of the median based average power value and the mean based average power value being a minimum function value for determining a minimum value m between the median based average power value and the mean based average power value. The average signal power is equal to the square of the minimum absolute value of the median and mean averaged over all quadrants.

The new SIR estimator described previously is preferably built on the data symbols. The output of the data demodulator, such as the Rake output or MUD output, is provided to the SIR estimator. As indicated above, one advantage of using a data demodulator output as the output is that it optimally and directly reflects the quality of the received data signal. Especially when SIR measurements are used in connection control techniques such as power control, a basic SIR estimate for a data demodulator as described above is highly desirable. In addition, the proposed SIR estimator can reduce the effects of bias on SIR estimation, resulting in a more reliable and accurate SIR estimate than conventional techniques. All such modifications and variations are within the scope of the invention as determined by the appended claims.

While the invention of this application has been described in terms of the preferred embodiment, other variations and applications of the invention as described in the claims below will be apparent to those skilled in the art.

Claims (15)

1. A method of determining a signal-to-interference ratio (SIR) of a data signal based on a data demodulator output, comprising:

(a) receiving demodulated symbols output from the data demodulator;

(b) for each quadrant of a Quadrature Phase Shift Keying (QPSK) set, the average signal power of the demodulated symbol is estimated as a median-based average power value m for the symbol_dAnd an average power value m based on the average value_eA function of (a);

(c) estimating an average effective interference power for the symbol; and

(d) the SIR is calculated by dividing the estimated average signal power of the symbol by the estimated average effective interference power of the symbol.

2. The method of claim 1, wherein the function of the median based average power value and the mean based average power value is a minimum function value used to determine a minimum value m between the median based average power value and the mean based average power value.

3. The method of claim 2, wherein the average signal power is equal to the square of the minimum of the absolute value of the median and the absolute value of the mean averaged over all quadrants.

4. The method of claim 1, wherein the data signal comprises demodulated data symbols.

5. The method of claim 4 wherein the data symbols are contained in a transmit data group of a Dedicated Physical Channel (DPCH).

6. The method of claim 4, wherein the data symbols are Quadrature Phase Shift Keying (QPSK) data symbols.

7. The method of claim 4, wherein the data symbols are Binary Phase Shift Key (BPSK) data symbols.

8. The method of claim 1, wherein the step (b) further comprises performing the following calculations:

whereinRepresents the kth demodulated QPSK signal, and i represents the quadrant of the QPSK set, and mean (y)_k(Q_i) And mean (y)_k(Q_i) Respectively represent in the ith quadrant Q_iThe median and mean of the symbol in (1).

The method of claim 8, wherein the step (c) further comprises performing the following calculations:

whereinIs regarded as the average signal amplitude, and

10. the method of claim 9, wherein the step (d) further comprises performing the following calculations:

wherein

11. A method according to claim 1 further comprising calculating a correction term C, wherein C ═ m_d-m_e|²。

12. A method for determining a signal-to-interference ratio (SIR) of a bit of data, comprising:

(a) receiving the data bits from the memory;

(b) estimating the average signal power of the bit as a median-based average power value m for the bit_dAnd an average power value m based on the average value_eA function of (a);

(c) estimating the average effective interference power of the bit; and

(d) the SIR is calculated by dividing the estimated average signal power of the bit by the estimated average effective interference power of the bit.

13. The method of claim 12, wherein the function of the median based average power value and the mean based average power value is a minimum function value used to determine a minimum value m between the median based average power value and the mean based average power value.

14. The method of claim 13, wherein the average signal power is equal to the square of the minimum of the absolute value of the median and the absolute value of the mean averaged over all quadrants.

15. The method of claim 12 further comprising calculating a correction term C, wherein C ═ m_d-m_e|²。