A METHOD AND APPARATUS FOR TUNING A COMBINER FILTER
TECHNICAL FIELD
The present invention relates to radio transmission, and in particular, to a method and an apparatus for minimising modulation degradation and signal distortion when tuning a combiner filter in a transmitting radio station.
BACKGROUND OF THE INVENTION AND PRIOR ART In a transmitting radio station, such as a base station of a mobile communication network, it is typically required that signals from plural transmitters are transmitted simultaneously to receiving radio stations, using a single antenna shared by the transmitters. This is achieved by using a combiner for combining the signals from the transmitters into a common signal, which is fed to a transmitting antenna. For example, a base station may be in radio connection with plural mobile terminals, located within its radio coverage area, over plural channel frequencies transmitted simultaneously over a single base station transmit antenna.
The combiner may be a cavity channel filter combiner comprising a number of cavity filters which are tuneable by adjusting the cavity resonance. Each active transmitter in the radio station is connected to a tuneable filter in the combiner for filtering the signal coming from the transmitter before being combined with other signals for transmission over the antenna. This signal filtering is necessary in order to prevent that radio energy enters other transmitters in the radio station instead of being radiated from the antenna. If each transmit frequency can be finely adjusted, it is possible to reduce the carrier spacing between transmitters in the radio station, which will increase the number of possible carrier frequencies within a given bandwidth. Thereby, network capacity is enhanced since it is possible to
choose from a relatively large number of frequencies when allocating channel frequencies to radio connections in individual radio stations, e.g., base stations of a cellular mobile communication network. Further, a relatively short reuse distance may be employed when performing cell planning in cellular networks, which will further increase the network capacity.
However, when using conventional tuneable filters in the combiner, it is a problem that the modulation accuracy is degraded as signals pass through the filter. Further, the modulation accuracy degradation becomes more prominent if a relatively close channel spacing is used in the combiner. A close channel spacing requires that the combiner filters are designed with a narrow passband filter shape, typically resulting in signal distortion, such as a greater group delay variation and amplitude response variation of the transmitted signal over the modulation bandwidth. This signal distortion in turn degrades the modulation accuracy, which makes it more difficult for the receiving radio station to demodulate and detect the received signal properly.
In order to minimise the signal degradation imposed by the tuneable filter combiner, it is important to tune the combiner filters with high accuracy. When tuning a combiner filter, a conventional method is to maximise the signal output and input power ratio, which somewhat reduces the amplitude distortion. Various solutions for maximising the output to input power ratio are described in the international patent applications WO 9416495, WO 9416497 and in the US patent US 5,473,292. However, maximising the output to input power ratio is often not sufficient for combating the modulation accuracy degradation. In particular, when using high modulation forms which are relatively sensitive to signal distortion, such as 8-PSK (Phase Shift Keying) modulation, and when a narrow
channel spacing is used, the tuning accuracy is more crucial. A solution is therefore desirable for tuning combiner filters with high accuracy and for reducing signal distortion, in order to enable the use of high modulation forms and/or a narrow carrier spacing in radio stations when transmitting plural signals over a shared antenna.
SUMMARY
It is an object of the present invention to overcome the problems as outlined above, by minimising the degradation of modulation accuracy and also reducing signal distortion when plural signals are transmitted over a single antenna, wherein combiner filters can be tuned with high accuracy.
This object and others are obtained by a method and an apparatus, wherein the modulation accuracy degradation is measured and used for tuning filters in the combiner.
An Error Vector Magnitude (EVM) may preferably be used as a measure for the modulation accuracy degradation. However, other parameters may also be used, such as phase error, amplitude error, amplitude variation, or any combination thereof.
Thus, the modulation accuracy degradation, in particular the EVM, of a transmitted signal is measured by comparing an input signal, entering the tuneable filter, with the output signal leaving the filter. The combiner filter is then tuned based on the measured modulation accuracy degradation, and preferably corresponding to a minimum of the modulation accuracy degradation.
In a preferred embodiment, the signal to be filtered and transmitted is pre-filtered in order to compensate for any signal distortion arising in the transmission chain of the signal, thereby further reducing the modulation accuracy degradation in the combiner. The signal to be filtered in the combiner is thus first pre-filtered by determining the
transmitter chain characteristic, and in a filter corresponding to an inverse filter of the transmission chain.
BRIEF DESCRIPTION OF THE DRAWINGS The present invention will now be described in more detail by way of non-limiting examples and with reference to the accompanying drawings, in which:
Fig. 1 is a general block diagram of a filter combiner serving a plurality of transmitters. - Fig. 2 is a basic block diagram of an apparatus for tuning a combiner filter.
Fig. 3 is a block diagram of an exemplary tuneable combiner filter using a homodyne receiver for determining the modulation accuracy degradation. - Fig. 4 is a block diagram of an exemplary apparatus for pre- filtering a transmission signal before entering a tuneable combiner filter, based on the transmitter chain characteristics .
Fig. 5 is a diagram illustrating a measured EVM as a function of a signal frequency offset error in a combiner filter with and without using pre-filtering, respectively, and
Fig. 6 is an exemplary flow chart illustrating steps performed in a method for pre-filtering a signal and tuning a combiner filter.
DESCRIPTION OF PREFERRED EMBODIMENTS
Fig. 1 illustrates a general view of the transmitting part 100 of a radio station, such as a base station of a cellular mobile communication network, comprising a filter combiner 102 serving a plurality of transmitters 104. Each transmitter 104 generates a signal comprising a baseband information signal being modulated onto a carrier wave of a specific frequency. There are a multitude of well-known
techniques for generating such signals, which are of no particular concern as such to the present invention, and will hence not be described further.
Each transmitter 104 is connected to a tuneable bandpass filter 106 in the combiner 102, such that signals coming from the transmitters 104 are filtered by the filters 106 before being combined into a common signal which is finally transmitted over a single shared antenna 108. Each transmitter 104 may in turn serve a limited number of simultaneously ongoing connections by using a suitable multiplexing technique, such as TDMA (Time Division Multiplex Access) .
The present invention is concerned with minimising the degradation of modulation accuracy, and also reducing signal distortion. According to one aspect of the invention, this is accomplished by enhancing the tuning accuracy of the tuneable combiner filters 106, whereby demodulation and detection of the transmitted signals is facilitated at receiving radio stations, such as a mobile terminals. In this way, the use of high modulation forms and/or a narrow channel spacing is enabled, as described above.
Fig. 2 illustrates an inventive apparatus, in a basic and schematic form, for tuning a combiner filter in a radio station. An input signal 200 coming from a transmitter 104 enters a tuneable bandpass filter 106 which is arranged to filter the input signal 200 into a predetermined bandwidth around a carrier frequency centre, for reasons described above. A filtered output signal 202 leaving the filter 106 can then be combined with other signals for transmission over a shared single antenna 108.
The tuneable filter 106 may be a cavity channel filter. Tuning the filter involves adjusting the bandpass width as well as the centre frequency. Various methods for tuning a bandpass filter are well-known in the art, and will
therefore not be described further. The present invention is not limited to any particular filter type or tuning method.
The input and output signals 200, 202 are extracted and demodulated, which is schematically illustrated in this figure by demodulation units 204 and 206, respectively. In order to control the filter tuning process, a filter controlling unit 208 is connected to the filter 106. The filter controlling unit 208 receives the demodulated input and output signals from the demodulators 204, 206, and is adapted to compare the demodulated signals for determining the modulation accuracy degradation of the signal imposed by the filter 106. The determined degradation is then used for adjusting the tuning of the filter 106 such that the modulation accuracy degradation is minimised. Fig. 3 illustrates in more detail an exemplary apparatus for tuning a combiner filter. An input signal 200 enters a tuneable bandpass filter 106 producing a filtered output signal 202, similar to Fig. 2. The input and output signals 200, 202 are extracted by interception devices 300 and 302, respectively, which sample the input and output signals 200, 202 on either side of the filter 106. By way of example, the interception devices 300, 302 may be using directional couplers, resistor elements, capacitor elements or any other technique for sampling the signals . In this example, the extracted input and output signals 200, 202 are fed to a homodyne receiver 304 for demodulating the signals before being compared. By using a homodyne receiver, simple lowpass filters can be used for the extracted signals, imposing a minimum of distortion on the signals to be compared. The homodyne receiver 304 comprises a local oscillator 306 working at the carrier frequency of the transmitted signal, phase shifters 308 and mixers 310. . The extracted signals 200, 202 are each being split into two branches which are mixed in the mixers 310 with signals at
carrier frequency from the local oscillator 306. The signals 300, 302 are thus each divided into an in-phase (I) signal and a quadrature phase (Q) signal, the Q signals being phase- shifted -90 degrees by means of the phase shifters 308. Hence, the signals 300, 302 have been divided into quadrature components, I and Q.
After some suitable filtering and analogue-to-digital conversion of the mixed signals, not shown, demodulated I and Q signals in digital form are created from each of the input and output signals 200, 202. The homodyne receiver 304 thus outputs the demodulated input I-signal Iin and Q-signal Qin as well as the demodulated output I-signal Iout and Q-signal Qout, which are all fed into a Digital Signal Processor (DSP) 312. The signals are then compared in the DSP 312 for determining the modulation accuracy degradation of the signal imposed by the filter 106.
In a preferred embodiment, the modulation accuracy degradation is determined by using a parameter known as Error Vector Magnitude (EVM) . The EVM may be used for any modulation form, such as 8-PSK modulation. The EVM is defined in, e.g., the ETSI standard for GSM, see document 3GPP TS 05.05 V8.10.0 (2001-06) , Annex G.
The modulation degradation of the output signal 202 compared to the input signal 200 may be determined in the DSP 312 by using the change of EVM imposed by the filter 106, δEVM(t), at a certain point in time t, according to the formula :
δEVM(t) = |k*Sout(t-τ)-Sin(t) I / |Sin(t) I (1)
where
Sout(t) is the output complex baseband signal from the tuneable filter,
Sin(t) is the input complex baseband signal to the tuneable filter, δEVM(t) is the change of Error Vector Magnitude resulting from filter characteristics of the tuneable cavity filter, - k is a normalisation factor, and τ is a combiner filter delay.
The DSP 312 is adapted to adjust the tuning of the tuneable cavity filter 106 such that δEVM(t) is minimised, k and τ are characteristics of the filter 106 that should be selected for minimising δEVM(t) . The tuning accuracy can also be improved by increasing the sampling rate in A/D converters used in the homodyne receiver 304.
In another preferred embodiment, the modulation accuracy can be further improved by pre-filtering the filter input signal, in order to compensate for signal distortion in a transmitter chain of the signal before being transmitted over the antenna. The transmitter chain mainly includes the combiner filter, but typically also includes various further components, such as cables, amplifiers and other filters, not shown.
With reference to Fig. 4, a true signal 400, S_true, is pre-filtered by a complex filter 402 having filter parameters mainly corresponding to an inverse filter of the transmitter chain characteristics . The true signal is the signal that would be transmitted if the input signal is not distorted and is optimally filtered by the tuneable filter 106. The true signal 400 can be extracted from the corresponding transmitter, not shown, as an unmodulated baseband signal in digital form, which is thus fed into the complex filter 402. The complex filter 402 may be implemented in a DSP or the like.
The output signal 404 from the tuneable combiner filter 106 is extracted and demodulated in a demodulation unit 406, which may comprise a homodyne receiver as in the example of Fig. 3. The resulting demodulated baseband signal S__out is also fed in digital form into the complex filter 402, which then can calculate the transmitter chain characteristics based on the true signal 400 and the output signal 404 for determining its filter parameters. Furthermore, the unmodulated digital true and output signals S_true, S_out entering the complex filter 402 may be divided into I and Q components, not shown. The transmitter chain characteristics can thus be continuously calculated from the output signal 404 and the true signal 400, for determining the parameters of the complex filter 402 in an adaptive manner. Alternatively, the characteristics of the complex filter 402 may be set by predetermined filter parameters.
In the example of Fig. 4, the inverse filter or the complex filter 402 is applied to the incoming true signal 400 for producing a pre-filtered signal S_in, which is modulated onto a carrier frequency in a modulation unit 408 to become an input signal 410 to the tuneable combiner filter 106. In this way, a pre-filtering is imposed to the transmitter signal by way of the inverse pre-filter or complex filter 402, in order to compensate for signal distortion in the transmitter chain, in particular in the combiner filter 106. Simultaneously with this process, the combiner filter 106 is tuned according to the examples of Figs 2 or 3 , as described above .
It will now be described in more detail how the transmitter chain characteristics can be calculated for determining the filter parameters of the complex filter 402. The following formulas may be used:
S_out = convolute (S_in, f) (2)
S in = convolute (S_true, p) (3)
where
S_in is the signal fed to the transmitter chain, S_out is the signal output from the transmitter chain, S_true is the ideal desired signal, p is the pre-filter characteristics, and f is the transmitter chain characteristics .
In order to make
S_out = S_true (4)
p is selected as the inverse of f .
In one embodiment, the combiner filter is tuned for minimising the modulation accuracy degradation, by comparing the input and output signals. In another embodiment, the signal is also pre-filtered before entering the combiner filter for compensating for signal distortion in the transmitter chain, thereby making S_out and S__true as equal as possible. If the characteristics of the transmitter chain are stable enough, e.g., with respect to filter tuning, pre-set or calibrated pre-filter parameters may be used.
In Fig. 5, the EVM is shown as a function of an offset from the carrier frequency centre. The upper continuous curve is the EVM when no pre-filter is used, and the lower dashed curve is the EVM when using an ideal pre-filter, as described above. As can be seen in Fig. 5, both curves display a minimum EVM at the frequency centre, i.e. when the frequency offset is zero. However, the EVM minimum is deeper and more pronounced when an ideal pre-filter is used. Thus, an optimised tuning of the filter is easier to obtain when using a pre-filter. The EVM can be used for estimating the frequency offset in order to determine the pre-filter characteristics. By determining the frequency offset and re-calculating the pre-filter
characteristics, mechanical impairments in the tuning parts can for example be compensated for. Other measures indicative of distortion may also be used, such as group delay variation and amplitude response variation. The EVM can thus be measured for determining the frequency offset in the combiner filter. However, in conventional filter combiners, such as tuneable channel cavity filter combiners, it has been difficult to determine whether the frequency offset is positive or negative since the EVM is substantially symmetric around the correct carrier frequency. This can be determined by detecting a filter response of the channel cavity filter combiner in a known synchronisation sequence, such as a training sequence, which is normally included in traffic bursts. The synchronisation sequence is then filtered by the combiner filter and a phase shift can be detected from consecutive bursts, which is different depending on whether the bandpass centre of the filter is above or below the centre frequency. Hence, the phase shift is not a symmetric function around the centre frequency, being negative below and positive above the centre frequency. The phase shift information may thus be used to adjust the pre-filter characteristics based on the frequency offset for improving the modulation accuracy degradation.
Finally, Fig. 6 illustrates steps in accordance with an exemplary procedure for tuning a combiner filter when using the frequency offset as a distortion measure, as described above. First in a step 600, the transmitter chain characteristics are calculated for a signal filtered by the combiner filter. Next in a step 602, parameters for a pre- filter are determined, corresponding to the inverse filter of the calculated transmitter chain, for pre-filtering the signal .
The EVM of the transmitted signal, as filtered by the pre-filter and the combiner filter, is measured in a next step
604. In a further step 606, the frequency offset of the combiner filter is determined, preferably based on the EVM measured in step 604. It is then checked in a step 608 whether the determined frequency offset exceeds a pre- set threshold value. If the pre- set threshold value is exceeded, the procedure continues by a step 610 where the combiner filter tuning is adjusted. From the step 610, the procedure returns to the step 604 where the EVM is measured again and the frequency offset is determined in step 606, after the tuning adjustment of step 610.
On the other hand, if the frequency offset calculated in step 606 does not exceed the pre-set threshold value, the procedure returns to step 600 where the process is repeated and the transmitter chain characteristics are re-calculated. Thus, Fig. 6 illustrates an ongoing procedure for tuning the combiner filter and for calculating the pre-filter parameters in an adaptive manner. If the frequency offset exceeds the pre-set threshold value, it is possible to correct the frequency offset by tuning the filter in step 610. However, if the frequency offset is smaller, it can only be corrected by means of the pre-filter parameters according to steps 600 and 602. As mentioned above, it is possible to use other distortion measures than the frequency offset, such as group delay variation and amplitude response variation. By using the method and apparatus as described herein for tuning a tuneable combiner filter and for pre- filtering a transmit signal, by minimising the degradation of modulation accuracy, and also reducing signal distortion, the filter tuning can be improved, which will allow higher modulation forms and/or a closer channel spacing for the transmitters .
While the invention has been described with reference to specific exemplary embodiments, the description is only intended to illustrate the inventive concept and should not be taken as limiting the scope of the invention. Various
alternatives, modifications and equivalents may be used without departing from the spirit of the invention, which is defined by the appended claims .