Login| Sign Up| Help| Contact|

Patent Searching and Data


Title:
METHOD FOR CONVERTING AND METHOD FOR PROCESSING ELECTRONIC MULTIPLEX SIGNALS, AND APPARATUS FOR PROCESSING SUCH SIGNALS
Document Type and Number:
WIPO Patent Application WO/1990/005417
Kind Code:
A1
Abstract:
Method for converting electronic signals from frequency-division multiplex (FDM format) into time-division multiplex (TDM format) while retaining substantially all signal information (amplitude and phase information) of a band-limited signal. The method comprises multiplication and convolution, e.g. filtering with frequency sweep signals. According to the invention there is first performed a multiplication (2) and then a convolution (3) with substantially ideal frequency sweep signals. A corresponding method for converting from TDM format to FDM format is also described. Moreover there is described a method and apparatus for processing electronic signals in FDM format, comprising conversion according to methods mentioned above and intermediate processing.

Inventors:
Ronnekleiv, Arne
Application Number:
PCT/NO1989/000115
Publication Date:
May 17, 1990
Filing Date:
November 06, 1989
Export Citation:
Click for automatic bibliography generation   Help
Assignee:
Frobe Radio, A/s Ronnekleiv Arne
International Classes:
H04J4/00; H04L27/26; (IPC1-7): H04J4/00
Foreign References:
US4737952A1988-04-12
US4578790A1986-03-25
EP0214649A21987-03-18
US4796253A1989-01-03
US4813035A1989-03-14
EP0280161A21988-08-31
GB2030822A1980-04-10
DE2811576C21983-12-01
GB2095515A1982-09-29
Download PDF:
Claims:
C l a i m s
1. Method for converting electronic signals from frequencydivision multiplex (FDM format) into time division multiplex (TDM format) while retaining substantially all signal information (amplitude and phase information) of a bandlimited signal, comprising multiplication and convolution (filtering) with frequency sweep signals, characterized in that a multiplication is first performed and then a convolution with substantially ideal frequency sweep signals.
2. Method for converting electronic signals from time division multiplex (TDM format) into frequencydivision multiplex (FDM format) while retaining substantially all signal information (amplitude and phase information) of a bandlimited signal, comprising multiplication and convolution (filtering) with frequency sweep signals, characterized in that a convolution is first performed and then a multiplication with substantially ideal frequency sweep signals.
3. Method according to claim 1 or 2, characterized in that the filter providing for said convolution is defined by an impulse response equal to hQ(t)* exp[j(ω0tμt2/2)J where h~(t) is the baseband impulse response of the desired filter function and where ω and μ are constants, whereby μ can be positive or negative, but not zero.
4. Method for processing electronic signals in FDM format (frequencydivision multiplex), comprising conversion into TDM format (timedivision multiplex) and conversion back to FDM format, whereby said conversion comprises multiplication and convolution with frequency sweep signals, characterized in that a first multiplication is performed and then a first convolution with substantially ideal frequency sweep signals, that the processing concerned is carried out, and that the processed signals are subjected to a second convolution and then a second multiplication.
5. Method according to claim 4, characterized in that said processing comprises removal of at least some TDM samples so as to eliminate corresponding frequency components in the signals.
6. Method according to claim 4, characterized in that said processing comprises multiplication of at least some TDM samples by a constant or variable factor'^so as to modify, possibly modulate the signals in amplitude and/or phase.
7. Method according to claims 4, 5 or β, characterized in that said processing comprises time displacement of the TDM samples within the TDM frame, so as to move corresponding signal components in frequency in relation to the remaining signal components.
8. Apparatus for processing electronic signals in FDM format (frequencydivision multiplex), comprising conversion into TDM format (timedivision) multiplex and conversion back to FDM format, whereby said conversion comprises multiplication and convolution with frequency sweep signals, characterized in that in front of a processor unit for said processing there are provided a corresponding number of units for performing said first multiplication and said first convolution each on their separate signals, and that behind the processor unit there are provided a corresponding number of units for performing said second convolution and said second multiplication each on their separate signals, these signals being derived from one or more of said units in front of the processor unit.
9. Apparatus according to claim 8, characterized in that each of said units in front of the processor units is permanently associated with one or more of said units behind the processor unit.
10. Apparatus according to claim 8, characterized in that there is provided a control unit for controlling the coordination between said units in front of the processor unit and said units behind the processor unit.
Description:
METHOD FOR CONVERTING AND METHOD FOR PROCESSING ELECTRONIC MULTIPLEX SIGNALS, AND APPARATUS FOR PROCESSING SUCH SIGNALS.

This invention relates to a method for converting and a method for processing electronic multiplex signals and an apparatus for processing such signals. Thus, at one hand the invention relates to filtering of electronic signals combined with conversion from frequency-division multiplexing (FDM) of the signals to time-division multiplexing (TDM) of the signals, whereby many filter functions must be performed in parallel in order to separate the signals from each other. On the other hand the invention is concerned with filtering of signals combined with conversion from time-division multiplexing of the signals into frequency-division multiplexing of the signals, whereby many filter functions must be performed in parallel in order to avoid frequency overlap of the signals in an undesired manner. In this connection the invention can be considered to relate to a combined electronic filter bank.

The invention also comprises a combination of conversion from FDM to TDM and subsequent conversion from TDM back to FDM, both taking place in one or more units having intermediate analogue and/or digital signal processing. This makes possible bandpass filtering of the signals with great flexibility in the choice of centre frequency ' and bandwidth for implementing filter functions, and with a low processing requirement, power consumption and space requirement compared to other, known solutions based on present-day technology.

Conversion between FDM and TDM formats is today known by filtering and sampling respectively modulation of sample sequences, for each individual signal, and combination of sample sequences or signals to TDM or FDM formats. This results in large circuit arrangements with a high volume and power consumption. Moreover, so-called transmultiplexer

techniques are known [1, 2] (See list of references at the end of this description) . These are to a large extent based on digital processing for carrying out the filter functions and require a relatively high processing capacity.

Further, techniques are known being based on the analogue-chirp-Z Fourier transform [3, 4], in which a Fourier transform of an analogue signal sequence is performed either by multiplication (M) , convolution (C) and finally multiplication (M) with so-called chirp signals (FM sweep signals), i.e. an MCM transform; or also convolution, multiplication and convolution with such signals, i.e. a CMC transform.

From the patent literature reference can be made to European patent publication No. 214,649 which relates to conversion from FDM to TDM by complete Fourier transform, and therefore is comprised by the field of known techniques as mentioned above. More of interest is US patent 4,578,790 which describes a system having in principle a structure somewhat related to that which the present invention is directed to. The US patent, however, is specialized in the direction of single sideband multiplexing and purely digital filtering processes. It utilizes the digital technique in order to obtain simplifications and does not at all discuss analogue processing in order to perform the individual functions in the transform.

The invention may be regarded to be most closely related to the above mentioned CMC transform, but is distinguished therefrom in that for the FDM-to-TDM case the first convolution (filtering) is deleted while at the same time the response of the filter employed for the last convolution is somewhat modified, but without substantially changing the complexity of the filter. In the TDM-to-FDM case the last filter is omitted and the first filter is modified.

Closer statements of the invention as well as the novel and specific features thereof are found in the claims, of which claim 1 and 2 is related to each one of the two cases just mentioned.

The change in the frequency response of the corresponding chirp filters from the MCM circuits to the circuits proposed here, in both cases can be described as follows, presuming that a filtering of the signals is desired which corresponds to an impuls response h Q (t) in the baseband of the filter. For the CMC transforms the filter on that side where the time samples are outputted and fed in respectively, then have a frequency response given by ω - ω (ω-ω Λ ) 2 ώ h Q ( ) exp [3 2 μ ] '

in which the exponential function represents an "ideal" chirp in the frequency domain. In the time domain an

"ideal" chirp is a frequency sweep signal having a frequency varying liniarly with time and having a constant amplitude, exp[j (ω t-μtV2) 3 • Hereω Q and μ are constants; μ may be positive or negative, but not zero, and determines the

2 frequency change per time unit (rad/s ) for the ideal chirp.

According to a particular feature of the invention the ideal filter is here defined by having an impulse response equal

2 to h«(t)* exp[j(ω Q t-μt /2)]. In systems with few signals it will be particularly useful to employ a filter response as stated here. The periodicity of the chirp signals gives the sampling rate for the individual signal.

The advantage obtained with the invention in relation to the circuits based on the Fourier transform is that it is sufficient with two chirp operations, i.e. a convolution (C) and a multiplication (M) compared to three operations, i.e. CMC or MCM with a full Fourier transform. This means that the invention results in a reduced volume, complexity and power consumption, especially in narrowband systems in which the chirp filters will be physically large. Note that for the MCM transform the chirp filter has an impulse response length closely corresponding to the sum of the response lengths of the two filters in the CMC structure for the same capacity.

The invention together with additional particular

features and advantages shall be explained more closely in the following description with reference to the drawings, in which:

Fig. 1 schematically and in the principle shows a first example of a circuit for conversion from FDM to TDM formats.

Fig. 1A-D show successive signal forms during conversion in the circuit of Fig. 1.

Fig. 2 schematically and in the principle shows another example of a circuit for signal conversion, namely from TDM to FDM format.

Fig. 2A-D show successive signal forms during conversion in the circuit of Fig. 2.

Fig. 3 shows in the principle an apparatus for processing electronic signals by conversion as in Fig. 1 and in Fig. 2.

Example 1: FDM-to-TDM conversion

Fig. 1 schematically shows a circuit for FDM-to-TDM conversion. The input signal first passes through a prefiltering filter ("cover-filter") 1, which removes noise and signals outside the band to be processed. It is then multiplied with one or mostly several parallel running chirp signals in one or more multipliers 2 before it is filtered in a chirp filter 3. The chirp rate of the chirp signals shall have an absolute value equal to the chirp rate of the ideal chirp signal which modulates the impulse response of the chirp filter, for an ideal operation. The sign of the chirp rates of the chirp and the filter as well as signal frequences and the selection of sideband after the multiplication, are so chosen that an input pure tone is outputted as a band-limited pulse.

Fig. 1A-1D show the signals in the circuit of Fig. 1 in

time-frequency diagrams. The input FDM signal is denoted 4. The signal ocmponents in time and frequency which influence the output TDM signal within a time slot, i.e. the processing window, are indicated (for a given choice of chirp rate and frequences). The chirp signals are shown at 5 and at 6 the signal after multiplication and an assumed selection of one sideband (the selection takes place in the chirp filter). The output TDM signal is shown at 7, i.e. a pulse for each input signal. It is presumed here that both in the chirp signals and in the impulse response of the chirp filter B the frequency decreases with time, that the signal has a higher centre frequency than the chirp signals and that the lower sideband is selected after multiplication. This means that the pulses corresponding to the signals having the highest frequency, are first outputted. Other combinations of chirp rate and centre frequences can give the same result - or also the highest frequency outputted last.

A circuit for TDM-to-FDM conversion is schematically shown in Fig. 2, comprising a chirp filter 11, a multiplicator 12 and a bandpass filter 13.

In the principle the TDM signal can be as shown at 7 in Fig. ID. After passage through the chirp filter 11, which is considered to have an increasing time delay with decreasing frequency, the signal will appear as shown at 14 in Fig. 2A. The signal is then multiplied by a series of periodic chirp signals having an increasing frequency versus time and the same chirp rate as the filter (absolute value), see Fig. 2B at 15. At 16 (Fig. 2C) the signal is shown after multiplication of the signal in one TDM window with the set of chirp signals and selection of the desired sideband (assumed). Here it is presumed that the chirp signals have a higher centre frequency than the signals 14, and that the lower sideband is chosen after multiplication. The same signal is shown at 17 after filtering in the filter 13 in Fig. 2.

It is presumed here that the filter function implemented by the CM circuit is adapted to the sampling

rates of the TDM signal so that repeated sidebands in the (re)constructed FDM signal are sufficiently supressed in accordance with design requirements and known theory. It is also presumed that the samples of the TDM signal are phase shifted so that they will have the centre frequency at which they shall be generated. This is trivial if the frequency is equal to an integer x 1/T in which is the time separation between the samples for each signal.

Based on conversion as in Examples 1 and 2 discussed above, there is now described a combination for signal processing as illustrated in the principle by Fig. 3.

Units which can perform FDM-to-TDM and TDM-to-FDM conversion can be combined via an intermediate processor unit which can be an analogue and/or a digital unit, and which performs such operations dependent upon the desired total processing, see Fig. 3. Units 21A, 21B... in Fig. 3 are FDM-to-TDM units having the same sampling rate. 22 denotes said processor unit and 23A, 23B... are TDM-to-FDM units adapted to the sampling rate for the units 21A, 2IB.-, and so forth.

If the FDM signals are individually completely characterized in accordance with known sampling theory, each by its series of samples in the TDM signal, with one sample per signal per TDM time slot, the processing and control of the signal operation of the processing unit 22 in Fig. can take place as follows:

If necessary desired TDM sequences are filtered further so that possible undesired signal components belonging to neighbouring FDM signals are removed. This will require intermediate storing of a number of TDM frames depending on the length of the digital filter.

TDM samples which have been processed can then be routed from a random unit 21A - 2IB and so forth to a random unit 23A - 23B, and be put into a random unoccupied time slot in the TDM signal here, by using known switching techniques. In this way a given arriving FDM signal at a random input unit 21A, 21B... can be reconstructed at a random frequency in a random input unit. The system

performs a selection of FDM signals, switching and frequency translation of the signals with a random complicated pattern given by the processor unit 22. The switching can take place in a control unit which is a part of the processor unit.

With the MC and CM circuits shown in Fig. 1 and 2, the spectre of each individual signal will be inverted from the input to the output. This can be avoided by complex conjugation of the FDM samples in the processor units 22 (or by modifying the FDM-to-TDM circuits 21A and so forth or the TDM-to-FDM circuit 23A and so forth) .

If a single TDM signal does not characterize an individual FDM signal (because it has too large bandwidth or wrong frequency) continued filtering, sampling and switching and reconstruction of signals within a flexible chosen band is possible. However, the processing must be somewhat modified in order to obtain a correct phase relationship at the output between signals from adjacent TDM frames. A possible modification is to let the processor unit 22 in addition to possible filtering and switching of the signals, perform complex conjugation thereof before reconstruction. This may be verified by known theory. Modifications of the FDM-to-TDM circuits 21A and so forth or the TDM-to-FDM circuits 23A and so forth offer further possibilities.

Selection (filtering-out) of a band from an input unit 21A and so forth, routing to a chosen output unit 23A and so forth and possible frequency translation of the band now takes place by having a corresponding group or TDM samples processed and conveyed in parallel to the desired output unit while maintaining their order, and possibly being complex conjugated and reconstructed therein. If any frequency translation in the unit is restricted to steps of an integer x 1/ this is sufficient. If smaller steps are desired, the samples must be displaced in frequency as by usual frequency translation before reconstruction included complex conjugation. Filtering in the processor unit 22 may now be necessary in order to obtain a sufficiently flat frequency response within the bands transferred, and it can

be utilized to sharpen band edges with insignificant extra processing (since only signals at the band edges must be processed) . This results in a significant increase in the intermediate storing of data ( TDM samples for all desired signals must be buffered in order to compensate for the time delay in the filters, presuming filtering with a linear phase) . Both aspects can be analyzed by means of known theory.

By having the TDM samples corresponding to certain output frequency bands multiplied by a variable m the processor unit, information can be modulated into signals in these frequency bands. The information may be made to be different in the various frequency bands by letting the variable multiplier be different with respect to the corresponding TDM samples. The frequency selectivity by the modulation will correspond to what is obtained by filtering in such a configuration. The modulation must be slow so that it will result in a small degree of band broadening of the signals, in order to avoid substantial removal or changing thereof by the filtering in the TDM-to-FDM conversion, and further to " avoid errors arising from TDM samples in the same frame representing FDM signals having different time delays. The latter can De avoided by delaying the modulating signal correspondingly for the individual TDM samples in the same frame. Both effects can be analyzed by means of known theory.

In addition to other processing as mentioned the TDM sample from various FDM-to-TDM units can be multiplied with different constants in the processor units and the results be added togetner, and then this can be used as an input signal for a TDM-to-FDM unit. Thereby a weighed f itereα sum of several arriving signals can be formed. This corresponds to frequency selective beam forming in an array antenna during reception.

Frequency selective beam forming in an array antenna during transmission can be obtained in an analogue manner by multiplying samples from a FDM-to-TDM unit with different constants in a processor unit, in addition to other

necessary processing as mentioned above, and using each sample as an input signal to a separate TDM-to-FDM unit together with other samples for other frequency ranges or added to samples for other beam direction at the same frequencies.

REFERENCES

1) H. Scheuerman, H. Gόckler: "A comprehensive survey of digital transmultiplexing methods"

Proc. IEEE, Vol 69, pp 1419-1450, Nov. 1981.

2) T.G. Marshall Jr., C . Puckette: "Guest Editors' Prologue, Special Issue on Transmultiplexers",

IEEE trans. Communications, Vol COM-30, No. 7, Part I, pp 1457-1459, July 1982.

3) M.A. Jack, E.G.S. Paige: "Fourier Transform Processors based on Surface Acoustic Wave Filters",

Wave Electronics _3, 1978 pp 229-247.

4) R.C. Williamson, V.S. Dolat, R.R. Rhodes and

D.M. Boroson: "A satellite-borne SAW chirp-transform system for uplink demodulation of FSK communication signals"

1979 Ultrasonics Symposium Proceedings, pp 741-747.