SUPPRESSION OF ADJACENT CHANNEL INTERFERENCE BY ADAPTIVE CHANNEL FILTERING IN MOBILE RADIO RECEIVERS

Abstract

A channel filter (200) with a variable pass bandwidth has a first low-pass filter (200.2) on the input side and is driven by a means (30) for controlling the pass bandwidth as a function of the adjacent channel interference. A first input of the control means (30) is connected to the first low-pass filter (200.2), and a second input is connected to a bandpass filter (200.4) which is connected in parallel with the channel filter (200). The signal path which contains the low-pass filter (200.2) contains a means (37) for removing a DC signal component or DC component.

Description

TECHNICAL FIELD

The invention relates to an adaptive channel filter for a receiver unit for a mobile communications system, and to a method for setting a variable pass bandwidth for a channel filter.

BACKGROUND

In many mobile radio systems such as GSM (Global System for Mobile Communications) and its further development EDGE (Enhanced Data Services for GSM Evolution), the overall transmission bandwidth is subdivided into a large number of narrowband subscriber frequency bands (traffic channels). The bandwidth of a subscriber frequency band in GSM and EDGE systems is 200 kHz. FIGS. 1a to 1c show three important interference influences during reception of a narrowband payload signal such as this.

FIGS. 1 a to 1c respectively show the spectral profile of a received signal 1 in the presence of interference 2.1, 2.2 and 2.3. FIG. 1a shows the narrowband received signal 1 in the presence of noise, which represents broadband interference 2.1. FIGS. 1b and 1c show two cases of multiple access interference, of multiple access interference (MAI), specifically cochannel interference (FIG. 1b) and adjacent channel interference (FIG. 1c). While the interference 2.2 in the case of cochannel interference occurs in the same subscriber frequency band as the desired signal 1, and is caused, for example, by an active subscriber in another cell in the network, the interference 2.3 in the case of adjacent channel interference occurs in one of the two adjacent subscriber frequency bands.

The influence of adjacent channel interference is influenced by the channel width of the subscriber frequency bands and the symbol frequency used in the system. In order to achieve a high system subscriber capacity and a high data rate, it is desirable to use narrow channel widths and high symbol frequencies. On the other hand, this results in an increase in the adjacent channel interference which, however, must not exceed a specific limit.

In the case of GSM and EDGE, the symbol frequency is 270.833 kHz and the channel width, as already mentioned, is 200 kHz. This means that the desired signal 1 and the interference 2.3 caused by adjacent channel interference spectrally overlap one another, as is shown in FIG. 1c. It is not possible to completely suppress the adjacent channel interference without constraining the spectrum of the desired signal 1.

In conventional receivers for mobile communications systems, the channel filter which is used to filter out the desired subscriber frequency band has a fixed, predetermined bandwidth. The chosen bandwidth represents a compromise between the mutually contradictory aims of utilization of the subscriber frequency band as well as possible for signal detection and suppression of adjacent channel interference as well as possible. This compromise is necessarily sub-optimal in many receiving situations.

FIG. 1 is based on the assumption that the clock rate of the signal is 40 m×f_T. The symbol frequency is denoted f_T, and is 270.833 kHz for GSM and EDGE. The oversampling factor is denoted m. For a channel filter in baseband, m is typically equal to 2. FIG. 1 shows that, with oversampling using the factor m, optional signal decimation can be carried out in each case between the low-pass filters 200.2 and 200.3 and the selection switch 210. The (optional) decimation is carried out by the decimators 211. The method of operation of each decimator 211 is to pass onto the output only one sample value from a group of m sample values, with the remaining m−1 sample values being rejected. The signal decimation is required only when signal processing is carried out at the symbol clock rate downstream from the adaptive channel filter.

The adaptive channel filter in FIG. 1 has the following disadvantage, however.

Owning to non-linearities in the RF receiver, each signal component (the payload signal or interference signal) at the input of the RF receiver also leads to a corresponding DC component (direct current, DC offset) in the quadrature-demodulated I and Q signals at the output, as is shown in FIG. 2a. In certain receiving situations, the DC offset can also vary within one burst. In this case, interference in the form of a step, a “DC step” is superimposed on the I and Q output signals, as is illustrated by way of example in FIG. 2b. The superimposition of a DC offset or DC step in the I and Q signals corrupts the estimate of the payload signal energy, while the DC interference in the energy estimate for the adjacent channel interference is largely suppressed owing to the bandpass filtering. Residual interference admittedly remains in the transition area of the DC step even after the bandpass filtering, but this lasts for a negligible time in comparison to the burst duration for energy estimation. The corruption of the energy measurement in one of the two paths leads to an increased error rate in the detection of the adjacent channel interference source, and thus to a deterioration in the reception quality.

SUMMARY

In consequence, the object of the present invention is to specify an adaptive channel filter which allows adequately good suppression of adjacent channel interference, while maintaining an adequate bandwidth, even in the presence of a DC signal component or DC component in the quadrature-demodulated I and Q signals in the receiver, and to specify a corresponding method for adaptive channel filtering having the stated characteristics.

This object can be achieved by an adaptive channel filter for a receiver unit for a mobile communications system, comprising a channel filter with a variable pass bandwidth, which comprises a first low-pass filter on the input side, a bandpass filter which is connected in parallel with the channel filter, a means for controlling the pass bandwidth of the channel filter as a function of the adjacent channel interference, having a first input which is connected to an output of the first low-pass filter, and having a second input which is connected to an output of the bandpass filter, and a means for removal of any DC signal component or DC component in the signal path which contains the first low-pass filter.

The means for removal of the DC signal component can be a notch filter. The means for removal of the DC signal component may comprise a DC signal estimator for estimation of the DC signal from an input signal, and may comprise an adder for subtraction of the estimated DC signal component from the input signal. The channel filter can be a digital low-pass filter in the baseband section of the receiver unit. The means for controlling the pass bandwidth also may take account of the noise, in addition to the adjacent channel interference. The channel filter with a variable pass bandwidth may comprise two or more filters which are arranged in series with one another and limit the bandwidth in steps, and may comprise a selection switch, at least some of whose inputs are connected to signal taps between the filters. The means for controlling the pass bandwidth may comprise in each case one energy estimator, which is connected to the two inputs and in each case calculates a variable which is representative of the power supplied to this input, and a comparison means which compares the variables calculated for different inputs with one another. The means for removal of the DC signal component can be arranged downstream of the first low-pass filter.

The object can also be achieved by a method for setting the variable pass bandwidth of a channel filter, having the steps of filtering of an input signal with a bandpass filter for production of a signal which is characteristic of an interference signal in an interference signal path, and having a low-pass filter for production of a signal which is characteristic of a payload signal in a payload signal path and for removal of the DC signal component in the payload signal path, calculating two variables which are characteristic of the signal powers of the two filtered signals, and setting of the pass bandwidth of the channel filter as a function of a comparison of the calculated variables.

The DC signal component in the payload signal path can be removed by means of a notch filter. The DC signal component in the payload signal path can be removed by first of all estimating it on the basis of the input signal and then subtracting it from the input signal.

Controlling the pass bandwidth of the channel filter as a function of the adjacent channel interference, that is to say in general as a function of a variable which is influenced by the strength of the adjacent channel interference, results in an adaptive channel filter using which the desired signal can always be filtered optimally in different receiving and interference situations. The means which is provided for removal of a DC signal component or DC component from the signal component passing through the first low-pass filter also ensures that the DC signal component does not lead to any corruption of the estimation of the payload signal energy.

The means for removal of the DC signal component may, in one embodiment, be formed by a notch filter. This is a special high-pass filter with high attenuation at the frequency zero. In order to achieve optimum detection of the adjacent channel interference source, it is desirable to use a notch filter with as narrow a stop band as possible, in order that the spectrum of the desired signal is filtered out as little as possible.

In another embodiment, the correction for the DC signal component is achieved by subtraction of an estimated DC signal value or DC value from the payload signal. In this case, a DC value is first of all estimated on a burst basis from the input signal. This estimated DC value is then subtracted from the input signal.

The adaptive channel filter according to the invention is preferably a digital low-pass filter in the baseband section of the receiver unit. In this case, the variable pass bandwidth is achieved by the low-pass filter having a variable upper cut-off frequency.

In addition to the (absolutely essential) relationship between the control of the pass bandwidth of the channel filter and the adjacent channel interference, it is also possible to take into account further influencing variables in the control of the pass bandwidth of the channel filter. In this context, one advantageous embodiment variant of the adaptive channel filter is characterized in that the means for controlling the pass bandwidth also takes account of the noise, in particular its strength.

In this case, the means for controlling the pass bandwidth is expediently designed to set the channel filter to a first, narrow pass bandwidth when the adjacent channel interference is high, and to set it to a second pass bandwidth, which is wider than the first pass bandwidth, when the adjacent channel interference and the noise are low, and to set a third pass bandwidth when the adjacent channel interference is low and the noise dominates the adjacent channel interference, which third pass bandwidth is wider than the first but narrower than the second pass bandwidth. The (reasonable) reduction in the pass bandwidth when the noise level is high results in the noise bandwidth of the received signal being reduced, but without causing excessively great signal distortion during the process.

BRIEF DESCRIPTION OF THE DRAWINGS

Exemplary embodiments of an adaptive channel filter will be explained in more detail in the following text with reference to the further drawings, in which:

FIGS. 1 a-1c show signal spectra in the presence of various interference sources, specifically broadband noise, cochannel interference and adjacent channel interference;

FIGS. 2 a, b show quadrature-demodulated I and Q signals with a DC component (a) and with a DC step within a burst (b);

FIG. 3 shows a block diagram of one exemplary embodiment of an adaptive channel filter according to the invention; and

FIG. 4 shows an exemplary embodiment of a DC correction device.

DETAILED DESCRIPTION

In the block diagram shown in FIG. 3 of one exemplary embodiment of an adaptive channel filter according to the invention, the reference symbols of the circuit components which are functionally equivalent and which are the same as those used for the adaptive channel filter shown in FIG. 1 have been retained. Furthermore, a DC correction device 37 has been inserted into the payload signal path upstream of the energy estimator 32, with the object of removing the DC signal component or DC component from this signal component. In the exemplary embodiment, the DC correction device 37 is a part of the control device 30. However, this is not of major technical significance for the present invention. It is equally well possible to provide for the control device 30 to be transferred essentially unchanged from the cited prior art and for the additional DC correction device 37 to be arranged upstream of the control device 30 in the payload signal path. It is likewise theoretically feasible for the DC correction device 37 to be arranged in the payload signal path upstream of the first low-pass filter 200.2, in order to remove the DC component even before the low-pass filtering.

In a first embodiment, the DC correction device 37 may be formed by a notch filter. In this case, this is a special high-pass filter which has high attenuation at the frequency 0. For optimum detection of the adjacent channel interference source, the notch filter preferably has as narrow a stop band as possible in order that the spectrum of the desired signal is filtered out as little as possible. On the other hand, this results in the step-function response having a longer decay time which, in the event of interference resulting from a DC step, once again leads to increased corruption of the energy estimate for the payload signal. An optimum notch filter represents a compromise between these mutually contradictory requirements. A low-order FIR or IIR filter is used for a cost-effective solution that takes account of these requirements.

The following equations indicate the transfer functions of two simple notch filters in the form of first and second order FIR filters.

H _FIR1(z)=1−z⁻¹  (2)

H _FIR2(z)=1−2z⁻1+z⁻²  (3)

As is known, both filters have very high attenuation at the frequency 0, and a step-function response with a very short decay time. Both filters thus provide good DC suppression and DC-step suppression. The broad stop band of the two filters has the disadvantage, however, that a relatively high proportion of the payload spectrum is also filtered out in the process.

An even better compromise can be achieved here by means of recursive filters. Even a first-order IIR filter can be used to produce a notch filter with a very narrow stop band. The following equation describes the transfer function of an IR notch filter such as this.

$\begin{array}{cc}{H}_{\mathrm{IIR}}\left(z\right)=\frac{1-z^{-1}}{1-{\mathrm{az}}^{-1}}& \left(4\right)\end{array}$

The parameter a allows the width of the stop band to be interchanged with the decay time period of the step-function response. When a=0, the IIR filter merges into the FIR filter described by equation 2 above. Simulations have shown that a good compromise is achieved with a=0.5.

In a second embodiment, the DC correction device 37 is implemented by estimation of the DC value of the signal X₂ in the payload signal 2, followed by subtraction of the estimate of the DC value from the signal. This is illustrated in FIG. 4. In consequence, the DC correction device 37 contains a DC estimator 37.1 and an adder 37.2. A DC value is estimated from the input signal X₂ on a burst basis in the DC estimator 37.1, and the estimated DC value is subtracted from the input signal X₂ in the adder 37.2. The simplest method for estimation of the DC value is to form the average value of the input signal over a specific time period M:

$\begin{array}{cc}{X}_{\mathrm{DC}}=\frac{1}{M}\sum _{k=1}^Mx_2\left(k\right)& \left(5\right)\end{array}$

where x₂ is the complex input signal and x_DC is the estimated complex DC value. The DC correction can now be written as follows:

x ₂*(i)=x_s(i)−x_DC i=1, 2, . . . , N  (6)

where N represents the number of data samples per burst.

In contrast to the notch filter in the first embodiment, the second embodiment has the disadvantage that it is more complex, since the DC estimate has the required accuracy only with relatively large M<N, and that residual interference always remains in the event of a DC step, whose extent depends on the magnitude of the DC step.

Claims

1. An adaptive channel filter for a receiver unit for a mobile communications system, comprising a channel filter with a variable pass bandwidth, which comprises a first low-pass filter on the input side,a bandpass filter which is connected in parallel with the channel filter,a means for controlling the pass bandwidth of the channel filter as a function of the adjacent channel interference, having a first input which is connected to an output of the first low-pass filter, andhaving a second input which is connected to an output of the bandpass filter, and a means for removal of a DC signal component or DC component in the signal path which contains the first low-pass filter.

2. The adaptive channel filter as claimed in claim 1, wherein the means for removal of the DC signal component is a notch filter.

3. The adaptive channel filter as claimed in claim 1, wherein the means for removal of the DC signal component comprises a DC signal estimator for estimation of the DC signal from an input signal, and comprises an adder for subtraction of the estimated DC signal component from the input signal.

4. The adaptive channel filter as claimed in claim 1, wherein the channel filter is a digital low-pass filter in the baseband section of the receiver unit.

5. The adaptive channel filter as claimed in claim 1, wherein the means for controlling the pass bandwidth also takes account of the noise, in addition to the adjacent channel interference.

6. The adaptive channel filter as claimed in claim 1, wherein the channel filter with a variable pass bandwidth comprises two or more filters which are arranged in series with one another and limit the bandwidth in steps, and comprises a selection switch, at least some of whose inputs are connected to signal taps between the filters.

7. The adaptive channel filter as claimed in claim 1, wherein the means for controlling the pass bandwidth comprises: in each case one energy estimator, which is connected to the two inputs and in each case calculates a variable which is representative of the power supplied to this input, anda comparison means which compares the variables calculated for different inputs with one another.

8. The adaptive channel filter as claimed in claim 1, wherein the means for removal of the DC signal component is arranged downstream of the first low-pass filter.

9. A method for setting the variable pass bandwidth of a channel filter, having the following steps: filtering of an input signal with a bandpass filter for production of a signal which is characteristic of an interference signal in an interference signal path, and having a low-pass filter for production of a signal which is characteristic of a payload signal in a payload signal path and for removal of the DC signal component in the payload signal path, calculating two variables which are characteristic of the signal powers of the two filtered signals, andsetting of the pass bandwidth of the channel filter as a function of a comparison of the calculated variables.

10. The method as claimed in claim 9, wherein the DC signal component in the payload signal path is removed by means of a notch filter.

11. The method as claimed in claim 9, wherein the DC signal component in the payload signal path is removed by first of all estimating it on the basis of the input signal and then subtracting it from the input signal.

12. An adaptive channel filter for a receiver unit for a mobile communications system, comprising a channel filter with a variable pass bandwidth, which comprises a first low-pass filter on the input side,a bandpass filter which is connected in parallel with the channel filter,a pass bandwidth controller which controls the pass bandwidth as a function of the adjacent channel interference, comprising a first input connected to an output of the first low-pass filter, and a second input connected to an output of the bandpass filter, anda DC component removal unit arranged in the signal path which contains the first low-pass filter.

13. The adaptive channel filter as claimed in claim 12, wherein the DC component removal unit is a notch filter.

14. The adaptive channel filter as claimed in claim 12, wherein the DC component removal unit comprises a DC signal estimator for estimation of the DC signal from an input signal, and comprises an adder for subtraction of the estimated DC signal component from the input signal.

15. The adaptive channel filter as claimed in claim 12, wherein the channel filter is a digital low-pass filter in the baseband section of the receiver unit.

16. The adaptive channel filter as claimed in claim 12, wherein the pass bandwidth controller also takes account of the noise, in addition to the adjacent channel interference.

17. The adaptive channel filter as claimed in claim 12, wherein the channel filter with a variable pass bandwidth comprises two or more filters which are arranged in series with one another and limit the bandwidth in steps, and comprises a selection switch, at least some of whose inputs are connected to signal taps between the filters.

18. The adaptive channel filter as claimed in claim 12, wherein the pass bandwidth controller comprises:in each case one energy estimator, which is connected to the two inputs and in each case calculates a variable which is representative of the power supplied to this input, anda comparison means which compares the variables calculated for different inputs with one another.

19. The adaptive channel filter as claimed in claim 12, wherein the DC component removal unit is arranged downstream of the first low-pass filter.