CA1336445C - Fsk discriminator - Google Patents
Fsk discriminatorInfo
- Publication number
- CA1336445C CA1336445C CA000575252A CA575252A CA1336445C CA 1336445 C CA1336445 C CA 1336445C CA 000575252 A CA000575252 A CA 000575252A CA 575252 A CA575252 A CA 575252A CA 1336445 C CA1336445 C CA 1336445C
- Authority
- CA
- Canada
- Prior art keywords
- discriminator
- pair
- filters
- signal
- filter
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L27/00—Modulated-carrier systems
- H04L27/10—Frequency-modulated carrier systems, i.e. using frequency-shift keying
- H04L27/14—Demodulator circuits; Receiver circuits
- H04L27/144—Demodulator circuits; Receiver circuits with demodulation using spectral properties of the received signal, e.g. by using frequency selective- or frequency sensitive elements
- H04L27/152—Demodulator circuits; Receiver circuits with demodulation using spectral properties of the received signal, e.g. by using frequency selective- or frequency sensitive elements using controlled oscillators, e.g. PLL arrangements
- H04L27/1525—Demodulator circuits; Receiver circuits with demodulation using spectral properties of the received signal, e.g. by using frequency selective- or frequency sensitive elements using controlled oscillators, e.g. PLL arrangements using quadrature demodulation
Abstract
The outputs of two filter channels for accepting two keying frequencies are compared to produce the output.
Each channel uses finite impulse response transversal filters; preferably with an impulse response of duration comparable with the symbol period of the incoming signal.
Each channel has a pair of filters for accepting phase quadrature components (or, alternatively, quadrature synchronous demodulators followed by filters): the filter coefficients differ from sinusoidal by (or, in the alternative case, are formed by) a modification function which is the same for both filters of the pair.
Each channel uses finite impulse response transversal filters; preferably with an impulse response of duration comparable with the symbol period of the incoming signal.
Each channel has a pair of filters for accepting phase quadrature components (or, alternatively, quadrature synchronous demodulators followed by filters): the filter coefficients differ from sinusoidal by (or, in the alternative case, are formed by) a modification function which is the same for both filters of the pair.
Description
l336~5 FSK DISCRIMINATOR
The present invention relates to discriminators S for frequency shifted keyed signals.
Reference is made to the accompanying drawings, in which:
Figure la is a block diagram of discriminator, Figure lb is a block diagram of a transversal filter, Figure 2 is a block diagram of one form of FSK
discriminator according to the invention, Figure 3 illustrates a typical idealised input waveform, Figure 4 illustrates a typical eye diagram, Figure 5 is a block diagram of a second form of FSK discriminator according to the invention, and appears out of consecutive order on the same page as Figure 6, Figures 5a, 5b and 5c illustrate the curves resulting from the functions shown in Figure 5, and Figure 6 is a block diagram of a filter section which may be used in the discriminator of Figure 2.
In 2-frequency fsk modulation, an on-off modulating signal is used to switch between a higher (fH) and a lower (fL) frequency, usually in a continuous phase manner. The general structure of a discriminator (Figure lA) is well known and comprises bandpass filters 1, 2 tuned to the two frequencies fH, fL respectively. The modulus of the filter outputs is taken by rectifiers or (as shown) squaring circuits 3, 4, whose outputs are then subtracted 5 to provide the demodulated output which may be further filtered 6 to reduce high frequency ripple.
The filtered signal is then converted to a square wave by a thresholding device 7.
It has been proposed (see U.S. Patent No.
4,437,068) to employ finite impulse response transversal S filters, as illustrated in Figure lB. That the input signal is in the form of signal samples is illustrated by sampler 10.
In the general case, because of the phase-sensitive nature of the FIR filter, quadrature pairs with vector summation of the outputs are used. Thus in Figure lB the higher frequency is handled by filters 12, 13 having tap coefficients which are sine and cosine functions of the tap number. The filter outputs are squared (14,15) and summed (16).
Filters, squarers and an adder (17-21) are provided in like manner for the low frequency channel, the coefficients again being sine and cosine function of tap number but with a period corresponding to the lower frequency.
As the filter outputs have already been squared, the outputs are supplied directly to the subtractor 5.
According to one aspect of the present invention there is provided a discriminator for frequency shift keyed signals comprising first filter means for accepting a first keying frequency, second filter means for accepting a second keying frequency, and a subtractor for generating the difference between the outputs of the filter means, each filter means comprising a pair of finite impulse response filters for accepting signal components which are mutually in phase quadrature and means for forming the sum of an even function of the output of each of the two filters of the pair, ..
~ 3 ~ 1336445 characterized in that each filter response is defined by a set of coefficients and the filters of the first pair have coefficients.
aHk = rHk Sin(2~fH k/fS + ~H) bHk = rHk cos(2~fH k/fs + ~H) and those of the second set aLk = KrLk Sin(2~fL k/fS + ~L) bLk = KrLk cS(2~fL k/fs + ~L) where k is the coefficient number, fH and fL are substantially equal to the keying frequencies, 0H and PL
are constants, K is a normalizing factor and rHk, rLk are sets of values, for the respective pairs, representing a modification function.
In another aspect, the invention provides a discriminator for frequency shift keyed signals comprising first filter means for accepting a first keying frequency, second filter means for accepting a second keying frequency, and a subtractor for generating the difference between the outputs of the filter means, characterized in that each filter means comprises: a pair of multipliers and means for supplying to the multipliers of the pair reference signals, mutually in phase quadrature, at the keying frequency which that filter means is to accept, each multiplier being followed by a finite impulse response filter, the two filters having the same response, representing a modification function, and means for forming the sum of an even function the output of each of the two filters of the pair.
other optical features of the invention are defined in the sub-claims appended.
- 3a -The discriminator shown in Figure 2 is basically similar to that depicted in Figure lB; similar items carry the same reference numerals. The preferred implementation is digital and therefore an analogue to digital converter ll is also included, although filters employing charge coupled devices (ccd) which use sampled analogue voltage (or charge) values could also be used.
To assist further description, the format of a 2fsk signal is shown in Figure 3. Successive bits of the modulating signal are represented in successive symbol periods, of duration T (i.e. a baud rate of l/T) by the higher or lower frequency FH or fL according to whether the bit is '0' or 'l' (or vice versa). In general, the frequencies used need not be harmonically related to the baud rate and hence the signal phase at the intersymbol transitions will vary, as can be seen in the figure.
Usually, the phase is continuous across transitions (i.e. no amplitude jump occurs). One symbol period may contain less th~n one, one, or ~ore t~an one, c~mpletle cycle of the keyed frequency.
Clearly, in order to discern the fre~uency wlthin a ~ymbol period, the sampling rate used must proYide sever~l sa~ples of ~he signal during that period. The samp~lng r~te may, but need not, ~e a mult1ple of the baud rate;
- likew~se its phase need not be related to the phase of the inter-symbol tr~nsit~ons (the latter usu~lly being unkncwn a~ the di~crlminator input). As an example for a VZ3 o (~CITT standard) forward channel at 1200 baud with ~requencies fH = 2100 ~æ and fL - 1300 Hz a ~600 Hz sampling ~requency may ~e used (ie 8 samples per sy~bol period~.
One objective of the design of the di~criminator shown in ~igure 2 is to reduce distortion 1n the demodulated output. There are ~ num~er of ways of expre~sing the dis~or~i~n. Where, in thi~ specification~ reference i~
mad~ to a distortion criterion~ any desired measure of distortion m~y be employed. Quoted distortion figures, however, are for isochronous distortion - viz ~e peak-to-peak temporal scatter ~T of ~he zero-cros~ings of the dis~riminator output, expressed as a percen~ge of the symbol period T - this is illustr~ted in the eye diagram of figure 4.
2S In order to identify the frequency tran~ition during for ~xample the time interval tl to t3 ~ar~ed in figure 3, it is necessary ~o observe the behaviour of the input signal during this inter~al. Although it is not intended that the invention be constr~ed with reference to any particular theory, it is believed that the use of analogue lu~ped filters, or their diyital equivalent, IIR
(infinite impulse response) filters, whose output depends on t~e history of the input for a t~me which is theoretically infinlte (though limited in pra~tice by - ~ - 133644S
noi~e o~ ~run~tlon e~ror), contribute~ to di~or~ion.
The ~is~ri~inator of figure ~ therefore employs flnite i~pulse re~ponse filter~ with a re6~ri~ted length (ie the ti~ differ~nce be~ween ~he earliest and latest input 6a~ples which ~ontribute to one s~mp~e output fro~ the fil~er ~ . Prefe~a~ly the length i~ le~ than ~wice the symbvl perio~, ~nd more prefer~bly ls app~oximately equal to t~le ~ymbol period for e~mple, ~or the V23 figures quote~ above, an 8-tap filter ~ight be used.
The prin~i~al difference bet~een the discrim~nator of f igure ~ and that o~ figure lB i~ th~t ~he fil~er coefficien~ deviate ~rum ~he sine/coslne form~ mentioned in the illtrodllction by ~ modiftcation functlon R. Thu~
in figure 2 the higher ~equenc~ is handled by fi~ters 12, 13, havin~ kap coe~ficients.
a~k - rHk sin(2~fH k/f~ ~ ~H) ....~l) bHX ~ rH~ co5(2~fH klf6 ~ ~H) ....(2) where k i~ the tap number. ~ i.~ an arbitra~y phase shift whlch can be ~ero but ~an be chosen ~o make the ~rgument 4f the sine and cosine f~nction~ zero ~ the mid-point ~f the filter (for an e~en number of taps, this refer~ to a no~ional half-tap v~lue), so th~t, at least in the ~bsence ~ ~he ~odific~tion function, the coefficients are symmetri~al (co~) or antisymm~trical (s~ne) ~ie ~cr an 8~tap filter ~l = +a8, a~ =~a7 etc). If the modii~tion is also symmetrical, this can reduce the nu~her o~ ~ultipliers required~ ~s befvre, the f~lter c~utputs are squ~red (l~, ~5) and ~ummed ~16).
~ er~, squa~rs ~nd ~n adder (17-~) are provided in like ~anner or the low ~requency ~hannel, coefficients aL~, bLK ~ein~ given by the a~ove exp~essi~n, fL, ~L ~eillg sub~ uted for ~H~ ~H - viz ~ ,k = KrLk Sin(~fL k~f5 ~ ~L) ~ (3) bL~ rL,k GO5~2~L ~/fS ~ IJ) 13364~
where K is a scaling factor (which can be unity) such that the output amplitude from the adder 16 for a signal input at fH is the same as that from the s adder 21 for a signal input at fL.
Thus each filter coefficient is modified by a factor rHk (for the higher frequency filters) or rLk (for the lower). It is noted that the object of using sine and cosine filters whose outputs are squared and summed is to ensure a constant output at the adder output in response to a continuous input at the relevant frequency and hence the same function RH = [rHk] (or RL ~ [rLk]) is used for both sine and cosine filters in order to preserve this. The functions RH, RL used for the upper and lower frequency filters may be the same, but this is not essential.
As will be seen from the examples given below, by selection of a suitable modification function, a significant improvement in the discriminator distortion can be obtained. The reason for this is not known.
When a zero-crossing occurs in the discriminator output, each filter contains portions of input signal at both frequencies fH,fL. The modification function concentrates most of the importance (or tap weight) into the central region of each FIR filter. The concentration ensures that an input frequency transition is tracked quite quickly by the filter system. With no modification function, the transition output is liable to dither about the threshold, giving rise to considerable distortion. An analytical determination of the optimum form of the function has not been found.
Some functions which have been found useful in practice are illustrated in Figures 5a to 5c (for an 8-tap filter) where the functions are plotted against tap 3s ~ 7 ~ 1 33 64 ~ 5 number k for clarity, these are shown as curves: in fact, of course, rk exists only for integer values of k. Mathematically expressed:
rk = hp - (k-4.5)2 (Fig. 5a) ...(5) rk = P + sin ~k/9 (Fig. 5b) ...(6) rk = ht ~ ¦k~4 5~ (Fig. Sc) --(7) It is thought desirable that the functions be substantially symmetrical about the filter mid-point, which is intuitively consistent with the fact that they have to discern transition from fH to fL and from fL to fH; however, the optimisation methods discussed do produce a degree of asymmetry. It is noted that the functions of Figures 5a to 5c show a monotonic increase from k=0 to the filter mid-point; this also appears to be a desirable feature, but a quite different shape is found to be preferable where certain types of receive filter (i.e. a filter at the discriminator input) are employed.
The output at this point is still in the form of digitally coded samples representing a waveform of the type shown in Figure 4. The zero-crossing can be identified by identifying a sign-change in the samples;
however this fixes the zero-crossing only to an accuracy of + half the sampling period (+6.75% in this case) which for many applications is inadequate. A second possibility would be to use a digital to analogue converter, with a filter to remove components at fs followed by a thresholding circuit as in the case of Figure 1. However, the preferred option is to employ an interpolator 22 which interpolates (e.g. linearly) between the sample values on either side of the zero crossing, to fix the zero-crossing more precisely.
An example of a filter, which results obtained 5 by a computer program will now be described.
Example 1 S The discriminator had the form of Figure 2, with the omission of the filter 6, for 1200 baud with fH = 2100 Hz, fL = 1300 Hz, fs = 9600 Hz. No receive filter was used. 8-tap filters were employed, as defined by equations (5) to (8), with K = 1.
lo Modification functions according to Figure 5 were used (the same for all four filters).
Discriminator 1: rk = 1 Discriminator 2: rk = 17.2 - (k-4.5)2 Discriminator 3: rk = ~0-9 + sin ~k/9 Discriminator 4: rk = 4.1 - k-4.5 The interpolator 22 carried out a linear interpolation between the samples flanking the zero-crossing, using floating point numbers with 9 decimal digit accuracy.
Results were as follows, using an input signal modulated with a 63 bit pseudo-random sequence.
Discriminator 1: Distortion = 16.394%
Discriminator 2: Distortion = 1.168%
Discriminator 3: Distortion = 0.961%
Discriminator 4: Distortion = 0.390%
*****
As mentioned above, an optimisation procedure may be applied to the functions. Suppose a function as defined by equation 7 is to be used as the starting point. This function has only one degree of freedom (ht) and ht can be varied on a trial and error basis to minimise distortion. A simple optimisation method for the coefficients rk can then be iteratively adjusted, as follows:
`- ` 1336~45 ~i) adiust ~in separate te~ts) ea~h rk by a small amount in both direction~
(eg +10/o)~
(ii) if any of those ~lxteen tests (for an 8-tap filter) re~ult~ in improvement, retaln that one change which gives the biggest improvement as the new ~a~ue for the relevant term.
(i~i) repeat (i) and (ii) until no ~urther improve~ent is obta~ned.
~ f de6ired ~he proce~s can be ~epeated for a smaller increment/decrement.
This proce~s will ~e illustrated by a further example.
ExamPle II
The d~scriminator had the same form a~ ln example I, except that an elllptic p~e-filter was used, defined as fOllows:
The ilter had four sections in cascade. One section ~s shown in fig~re 6, and has an input 4~ for the input signal or the output of the preceding section. A~
adder 41 has an output which feeds t~o delay stages 42,43 whose outputs are mult~plied in multipliers 4~, 45 by coefficients B~,B2. ~he inputs ~o the adder 4~ are the input 40, and the outpu~s of the two ~ultipliers. The section output (to the next ~ection or to the discriminato~) is formed ~y an adder 4~ whose three lnputs are the output of the adde~ 41 and the outputs of two multipliers 47,48 in which the outputs of the delay sta~es 42,43 are mult~plied ~y coeffici~nts Al,A2.
3~ ~he val~es of the coefficients (where A2 (3) indicates coefficient A2 for the third section, etc) we~e 133644~
Al(l)- .528102: A2(1)~1: Bl(l)- ~0.24~46: 8~ eO.464~1 A1~2)= -1. 74852: A~(2)~1: Bl(2)= -1.14611s 82(2)~0.59096 Al(3)= 1.423245: ~2(3)=1: ~1(3)= 0.10045: B2(~ .B4652 A1~4)= ~1. 92~2: A2(4)-1: Bl(4)= -1.55~17: 82(4)-0.90748 With rk defined by equation 11, the optlmum value of ht was found to be 4.8 ¦ie r~ = 4.8 - ¦k-4.5¦ ) wi~h a distortion ~f 6.90l0 . These r~ were then used as st~rtin~ points in the optimisation proce~
as descrtbed above (~ 10/O changes only), whlch o gave an improvement to 3.774/o distortion. The corresponding v~lue~ for r~ were:
k 1 2 3 4 5 6 7 rk 1~0059 2.6225 3.3 ~.3 4.3 3.3 2.3 1.0~59 Tests were also conducted w~th a post-detect f~lter 6, ~5 which was a fourth order FIR filte~ ~or rejecting tones at 2600 Hz and 4200 ~z (twice ~he signalling ~requencies).
The fllter ~ran~er function was ~z2 + 0.2~1~5z + 1) (z2 + 1.8477~z + 1) Optimisation with the filter in place ga~e a fina~
~o - distortion of 3.6~/o.
* ~ ~
Where a discriminator is ~o be used ~or a lower as well a~ a high baud rate, wlth lower keyin~ frequencies (~g for a backward signalllng channel as in Y23) lt is convenient to employ the ~ame sampling rate. However, assumin~ the sa~e filter ~rder, th~s can result in an undesirable short filter relative to the keying frequency and therefore a modified version of the discriminator s includes means for sub-sampling the signal, so that the effective filter length is increased.
Figure 5 shows an alternative embodiment of the invention. Those components which are the same as those in Figure 2 are given the same reference numerals.
The difference between Figures 2 and 5 is that instead of employing a pair of FIR filters (12,13) which both provides the necessary phase quadrature frequency discrimination by multiplying the signal samples by coefficients which are sine or cosine functions of filter tap position and impose the desired modification function by modifying those coefficients, the first of these requirements is met by a pair of multipliers 30,31 in which the input signal is multiplied by respective phase quadrature reference signals at frequency fH from suitable source (not shown), viz.
sin (2~fHt+~H) ...8 COS ( 27~fHt+~3H) where ~H is an arbitrary constant.
The second requirement is met by FIR filters, 32,33 which follow the multipliers 30,31 and differ from the filters 12,13 of Figure 2 only in that their coefficients now are determined only by the modificiation function -i.e.
aHk=bHkrHk ...10 It will be apparent of course that if the amplitudes of the multiplying signals were different, this could be corrected by varying the ratio of aHk to bHk (aHk/bHk being the same for all k) or vice versa, provided that the overall gain in the two paths is the same.
13364~
The filter outputs are squared and summed, as before. Although the filter outputs themselves are not S the same as in Figure 2, it can be shown that the output of the adder 16 is the same - i.e. the two arrangements are mathematically identical (although in practice in a digital implementation there may be small differences due to the use of finite precision arithmetic).
The lower channel is modified in similar manner, with mulipliers 34,35 for multiplying by sin (2~fLt+~L) and cos (2~fLt+~L) and filters 36,37 with coefficients aLK=bLK=KrLk.
Although the invention has been described in terms of a 2fsk discriminator, it may of course be applied to 3fsk (or higher order) systems by inclusion of an appropriate number of filter arrangements are suitable "majority decision" circuitry.
~,, ~
The present invention relates to discriminators S for frequency shifted keyed signals.
Reference is made to the accompanying drawings, in which:
Figure la is a block diagram of discriminator, Figure lb is a block diagram of a transversal filter, Figure 2 is a block diagram of one form of FSK
discriminator according to the invention, Figure 3 illustrates a typical idealised input waveform, Figure 4 illustrates a typical eye diagram, Figure 5 is a block diagram of a second form of FSK discriminator according to the invention, and appears out of consecutive order on the same page as Figure 6, Figures 5a, 5b and 5c illustrate the curves resulting from the functions shown in Figure 5, and Figure 6 is a block diagram of a filter section which may be used in the discriminator of Figure 2.
In 2-frequency fsk modulation, an on-off modulating signal is used to switch between a higher (fH) and a lower (fL) frequency, usually in a continuous phase manner. The general structure of a discriminator (Figure lA) is well known and comprises bandpass filters 1, 2 tuned to the two frequencies fH, fL respectively. The modulus of the filter outputs is taken by rectifiers or (as shown) squaring circuits 3, 4, whose outputs are then subtracted 5 to provide the demodulated output which may be further filtered 6 to reduce high frequency ripple.
The filtered signal is then converted to a square wave by a thresholding device 7.
It has been proposed (see U.S. Patent No.
4,437,068) to employ finite impulse response transversal S filters, as illustrated in Figure lB. That the input signal is in the form of signal samples is illustrated by sampler 10.
In the general case, because of the phase-sensitive nature of the FIR filter, quadrature pairs with vector summation of the outputs are used. Thus in Figure lB the higher frequency is handled by filters 12, 13 having tap coefficients which are sine and cosine functions of the tap number. The filter outputs are squared (14,15) and summed (16).
Filters, squarers and an adder (17-21) are provided in like manner for the low frequency channel, the coefficients again being sine and cosine function of tap number but with a period corresponding to the lower frequency.
As the filter outputs have already been squared, the outputs are supplied directly to the subtractor 5.
According to one aspect of the present invention there is provided a discriminator for frequency shift keyed signals comprising first filter means for accepting a first keying frequency, second filter means for accepting a second keying frequency, and a subtractor for generating the difference between the outputs of the filter means, each filter means comprising a pair of finite impulse response filters for accepting signal components which are mutually in phase quadrature and means for forming the sum of an even function of the output of each of the two filters of the pair, ..
~ 3 ~ 1336445 characterized in that each filter response is defined by a set of coefficients and the filters of the first pair have coefficients.
aHk = rHk Sin(2~fH k/fS + ~H) bHk = rHk cos(2~fH k/fs + ~H) and those of the second set aLk = KrLk Sin(2~fL k/fS + ~L) bLk = KrLk cS(2~fL k/fs + ~L) where k is the coefficient number, fH and fL are substantially equal to the keying frequencies, 0H and PL
are constants, K is a normalizing factor and rHk, rLk are sets of values, for the respective pairs, representing a modification function.
In another aspect, the invention provides a discriminator for frequency shift keyed signals comprising first filter means for accepting a first keying frequency, second filter means for accepting a second keying frequency, and a subtractor for generating the difference between the outputs of the filter means, characterized in that each filter means comprises: a pair of multipliers and means for supplying to the multipliers of the pair reference signals, mutually in phase quadrature, at the keying frequency which that filter means is to accept, each multiplier being followed by a finite impulse response filter, the two filters having the same response, representing a modification function, and means for forming the sum of an even function the output of each of the two filters of the pair.
other optical features of the invention are defined in the sub-claims appended.
- 3a -The discriminator shown in Figure 2 is basically similar to that depicted in Figure lB; similar items carry the same reference numerals. The preferred implementation is digital and therefore an analogue to digital converter ll is also included, although filters employing charge coupled devices (ccd) which use sampled analogue voltage (or charge) values could also be used.
To assist further description, the format of a 2fsk signal is shown in Figure 3. Successive bits of the modulating signal are represented in successive symbol periods, of duration T (i.e. a baud rate of l/T) by the higher or lower frequency FH or fL according to whether the bit is '0' or 'l' (or vice versa). In general, the frequencies used need not be harmonically related to the baud rate and hence the signal phase at the intersymbol transitions will vary, as can be seen in the figure.
Usually, the phase is continuous across transitions (i.e. no amplitude jump occurs). One symbol period may contain less th~n one, one, or ~ore t~an one, c~mpletle cycle of the keyed frequency.
Clearly, in order to discern the fre~uency wlthin a ~ymbol period, the sampling rate used must proYide sever~l sa~ples of ~he signal during that period. The samp~lng r~te may, but need not, ~e a mult1ple of the baud rate;
- likew~se its phase need not be related to the phase of the inter-symbol tr~nsit~ons (the latter usu~lly being unkncwn a~ the di~crlminator input). As an example for a VZ3 o (~CITT standard) forward channel at 1200 baud with ~requencies fH = 2100 ~æ and fL - 1300 Hz a ~600 Hz sampling ~requency may ~e used (ie 8 samples per sy~bol period~.
One objective of the design of the di~criminator shown in ~igure 2 is to reduce distortion 1n the demodulated output. There are ~ num~er of ways of expre~sing the dis~or~i~n. Where, in thi~ specification~ reference i~
mad~ to a distortion criterion~ any desired measure of distortion m~y be employed. Quoted distortion figures, however, are for isochronous distortion - viz ~e peak-to-peak temporal scatter ~T of ~he zero-cros~ings of the dis~riminator output, expressed as a percen~ge of the symbol period T - this is illustr~ted in the eye diagram of figure 4.
2S In order to identify the frequency tran~ition during for ~xample the time interval tl to t3 ~ar~ed in figure 3, it is necessary ~o observe the behaviour of the input signal during this inter~al. Although it is not intended that the invention be constr~ed with reference to any particular theory, it is believed that the use of analogue lu~ped filters, or their diyital equivalent, IIR
(infinite impulse response) filters, whose output depends on t~e history of the input for a t~me which is theoretically infinlte (though limited in pra~tice by - ~ - 133644S
noi~e o~ ~run~tlon e~ror), contribute~ to di~or~ion.
The ~is~ri~inator of figure ~ therefore employs flnite i~pulse re~ponse filter~ with a re6~ri~ted length (ie the ti~ differ~nce be~ween ~he earliest and latest input 6a~ples which ~ontribute to one s~mp~e output fro~ the fil~er ~ . Prefe~a~ly the length i~ le~ than ~wice the symbvl perio~, ~nd more prefer~bly ls app~oximately equal to t~le ~ymbol period for e~mple, ~or the V23 figures quote~ above, an 8-tap filter ~ight be used.
The prin~i~al difference bet~een the discrim~nator of f igure ~ and that o~ figure lB i~ th~t ~he fil~er coefficien~ deviate ~rum ~he sine/coslne form~ mentioned in the illtrodllction by ~ modiftcation functlon R. Thu~
in figure 2 the higher ~equenc~ is handled by fi~ters 12, 13, havin~ kap coe~ficients.
a~k - rHk sin(2~fH k/f~ ~ ~H) ....~l) bHX ~ rH~ co5(2~fH klf6 ~ ~H) ....(2) where k i~ the tap number. ~ i.~ an arbitra~y phase shift whlch can be ~ero but ~an be chosen ~o make the ~rgument 4f the sine and cosine f~nction~ zero ~ the mid-point ~f the filter (for an e~en number of taps, this refer~ to a no~ional half-tap v~lue), so th~t, at least in the ~bsence ~ ~he ~odific~tion function, the coefficients are symmetri~al (co~) or antisymm~trical (s~ne) ~ie ~cr an 8~tap filter ~l = +a8, a~ =~a7 etc). If the modii~tion is also symmetrical, this can reduce the nu~her o~ ~ultipliers required~ ~s befvre, the f~lter c~utputs are squ~red (l~, ~5) and ~ummed ~16).
~ er~, squa~rs ~nd ~n adder (17-~) are provided in like ~anner or the low ~requency ~hannel, coefficients aL~, bLK ~ein~ given by the a~ove exp~essi~n, fL, ~L ~eillg sub~ uted for ~H~ ~H - viz ~ ,k = KrLk Sin(~fL k~f5 ~ ~L) ~ (3) bL~ rL,k GO5~2~L ~/fS ~ IJ) 13364~
where K is a scaling factor (which can be unity) such that the output amplitude from the adder 16 for a signal input at fH is the same as that from the s adder 21 for a signal input at fL.
Thus each filter coefficient is modified by a factor rHk (for the higher frequency filters) or rLk (for the lower). It is noted that the object of using sine and cosine filters whose outputs are squared and summed is to ensure a constant output at the adder output in response to a continuous input at the relevant frequency and hence the same function RH = [rHk] (or RL ~ [rLk]) is used for both sine and cosine filters in order to preserve this. The functions RH, RL used for the upper and lower frequency filters may be the same, but this is not essential.
As will be seen from the examples given below, by selection of a suitable modification function, a significant improvement in the discriminator distortion can be obtained. The reason for this is not known.
When a zero-crossing occurs in the discriminator output, each filter contains portions of input signal at both frequencies fH,fL. The modification function concentrates most of the importance (or tap weight) into the central region of each FIR filter. The concentration ensures that an input frequency transition is tracked quite quickly by the filter system. With no modification function, the transition output is liable to dither about the threshold, giving rise to considerable distortion. An analytical determination of the optimum form of the function has not been found.
Some functions which have been found useful in practice are illustrated in Figures 5a to 5c (for an 8-tap filter) where the functions are plotted against tap 3s ~ 7 ~ 1 33 64 ~ 5 number k for clarity, these are shown as curves: in fact, of course, rk exists only for integer values of k. Mathematically expressed:
rk = hp - (k-4.5)2 (Fig. 5a) ...(5) rk = P + sin ~k/9 (Fig. 5b) ...(6) rk = ht ~ ¦k~4 5~ (Fig. Sc) --(7) It is thought desirable that the functions be substantially symmetrical about the filter mid-point, which is intuitively consistent with the fact that they have to discern transition from fH to fL and from fL to fH; however, the optimisation methods discussed do produce a degree of asymmetry. It is noted that the functions of Figures 5a to 5c show a monotonic increase from k=0 to the filter mid-point; this also appears to be a desirable feature, but a quite different shape is found to be preferable where certain types of receive filter (i.e. a filter at the discriminator input) are employed.
The output at this point is still in the form of digitally coded samples representing a waveform of the type shown in Figure 4. The zero-crossing can be identified by identifying a sign-change in the samples;
however this fixes the zero-crossing only to an accuracy of + half the sampling period (+6.75% in this case) which for many applications is inadequate. A second possibility would be to use a digital to analogue converter, with a filter to remove components at fs followed by a thresholding circuit as in the case of Figure 1. However, the preferred option is to employ an interpolator 22 which interpolates (e.g. linearly) between the sample values on either side of the zero crossing, to fix the zero-crossing more precisely.
An example of a filter, which results obtained 5 by a computer program will now be described.
Example 1 S The discriminator had the form of Figure 2, with the omission of the filter 6, for 1200 baud with fH = 2100 Hz, fL = 1300 Hz, fs = 9600 Hz. No receive filter was used. 8-tap filters were employed, as defined by equations (5) to (8), with K = 1.
lo Modification functions according to Figure 5 were used (the same for all four filters).
Discriminator 1: rk = 1 Discriminator 2: rk = 17.2 - (k-4.5)2 Discriminator 3: rk = ~0-9 + sin ~k/9 Discriminator 4: rk = 4.1 - k-4.5 The interpolator 22 carried out a linear interpolation between the samples flanking the zero-crossing, using floating point numbers with 9 decimal digit accuracy.
Results were as follows, using an input signal modulated with a 63 bit pseudo-random sequence.
Discriminator 1: Distortion = 16.394%
Discriminator 2: Distortion = 1.168%
Discriminator 3: Distortion = 0.961%
Discriminator 4: Distortion = 0.390%
*****
As mentioned above, an optimisation procedure may be applied to the functions. Suppose a function as defined by equation 7 is to be used as the starting point. This function has only one degree of freedom (ht) and ht can be varied on a trial and error basis to minimise distortion. A simple optimisation method for the coefficients rk can then be iteratively adjusted, as follows:
`- ` 1336~45 ~i) adiust ~in separate te~ts) ea~h rk by a small amount in both direction~
(eg +10/o)~
(ii) if any of those ~lxteen tests (for an 8-tap filter) re~ult~ in improvement, retaln that one change which gives the biggest improvement as the new ~a~ue for the relevant term.
(i~i) repeat (i) and (ii) until no ~urther improve~ent is obta~ned.
~ f de6ired ~he proce~s can be ~epeated for a smaller increment/decrement.
This proce~s will ~e illustrated by a further example.
ExamPle II
The d~scriminator had the same form a~ ln example I, except that an elllptic p~e-filter was used, defined as fOllows:
The ilter had four sections in cascade. One section ~s shown in fig~re 6, and has an input 4~ for the input signal or the output of the preceding section. A~
adder 41 has an output which feeds t~o delay stages 42,43 whose outputs are mult~plied in multipliers 4~, 45 by coefficients B~,B2. ~he inputs ~o the adder 4~ are the input 40, and the outpu~s of the two ~ultipliers. The section output (to the next ~ection or to the discriminato~) is formed ~y an adder 4~ whose three lnputs are the output of the adde~ 41 and the outputs of two multipliers 47,48 in which the outputs of the delay sta~es 42,43 are mult~plied ~y coeffici~nts Al,A2.
3~ ~he val~es of the coefficients (where A2 (3) indicates coefficient A2 for the third section, etc) we~e 133644~
Al(l)- .528102: A2(1)~1: Bl(l)- ~0.24~46: 8~ eO.464~1 A1~2)= -1. 74852: A~(2)~1: Bl(2)= -1.14611s 82(2)~0.59096 Al(3)= 1.423245: ~2(3)=1: ~1(3)= 0.10045: B2(~ .B4652 A1~4)= ~1. 92~2: A2(4)-1: Bl(4)= -1.55~17: 82(4)-0.90748 With rk defined by equation 11, the optlmum value of ht was found to be 4.8 ¦ie r~ = 4.8 - ¦k-4.5¦ ) wi~h a distortion ~f 6.90l0 . These r~ were then used as st~rtin~ points in the optimisation proce~
as descrtbed above (~ 10/O changes only), whlch o gave an improvement to 3.774/o distortion. The corresponding v~lue~ for r~ were:
k 1 2 3 4 5 6 7 rk 1~0059 2.6225 3.3 ~.3 4.3 3.3 2.3 1.0~59 Tests were also conducted w~th a post-detect f~lter 6, ~5 which was a fourth order FIR filte~ ~or rejecting tones at 2600 Hz and 4200 ~z (twice ~he signalling ~requencies).
The fllter ~ran~er function was ~z2 + 0.2~1~5z + 1) (z2 + 1.8477~z + 1) Optimisation with the filter in place ga~e a fina~
~o - distortion of 3.6~/o.
* ~ ~
Where a discriminator is ~o be used ~or a lower as well a~ a high baud rate, wlth lower keyin~ frequencies (~g for a backward signalllng channel as in Y23) lt is convenient to employ the ~ame sampling rate. However, assumin~ the sa~e filter ~rder, th~s can result in an undesirable short filter relative to the keying frequency and therefore a modified version of the discriminator s includes means for sub-sampling the signal, so that the effective filter length is increased.
Figure 5 shows an alternative embodiment of the invention. Those components which are the same as those in Figure 2 are given the same reference numerals.
The difference between Figures 2 and 5 is that instead of employing a pair of FIR filters (12,13) which both provides the necessary phase quadrature frequency discrimination by multiplying the signal samples by coefficients which are sine or cosine functions of filter tap position and impose the desired modification function by modifying those coefficients, the first of these requirements is met by a pair of multipliers 30,31 in which the input signal is multiplied by respective phase quadrature reference signals at frequency fH from suitable source (not shown), viz.
sin (2~fHt+~H) ...8 COS ( 27~fHt+~3H) where ~H is an arbitrary constant.
The second requirement is met by FIR filters, 32,33 which follow the multipliers 30,31 and differ from the filters 12,13 of Figure 2 only in that their coefficients now are determined only by the modificiation function -i.e.
aHk=bHkrHk ...10 It will be apparent of course that if the amplitudes of the multiplying signals were different, this could be corrected by varying the ratio of aHk to bHk (aHk/bHk being the same for all k) or vice versa, provided that the overall gain in the two paths is the same.
13364~
The filter outputs are squared and summed, as before. Although the filter outputs themselves are not S the same as in Figure 2, it can be shown that the output of the adder 16 is the same - i.e. the two arrangements are mathematically identical (although in practice in a digital implementation there may be small differences due to the use of finite precision arithmetic).
The lower channel is modified in similar manner, with mulipliers 34,35 for multiplying by sin (2~fLt+~L) and cos (2~fLt+~L) and filters 36,37 with coefficients aLK=bLK=KrLk.
Although the invention has been described in terms of a 2fsk discriminator, it may of course be applied to 3fsk (or higher order) systems by inclusion of an appropriate number of filter arrangements are suitable "majority decision" circuitry.
~,, ~
Claims (33)
1. A discriminator for frequency shift keyed signals comprising first filter means for accepting a first keying frequency, second filter means for accepting a second keying frequency, and a subtractor for generating the difference between the outputs of the filter means, each filter means comprising a pair of finite impulse response filters for accepting signal components which are mutually in phase quadrature and means for forming the sum of an even function of the output of each of the two filters of the pair, characterized in that each filter response is defined by a set of coefficients and the filters of the first pair have coefficients aHk = rHk sin(2.pi.fH k/fs + .PHI.H) bHk = rHk cos(2.pi.fH k/fs + .PHI.H) and those of the second pair aLk = KrLk sin(2.pi.fL k/fs + .PHI.L) bLk = KrLk cos(2.pi.fL k/fs + .PHI.L) where k is the cofficient number, fH and fL are substantially equal to the keying frequencies, .PHI.H and .PHI.L are constants, fs is the sampling frequency at which the first and second keying frequencies are sampled, K is a normalizing factor and rHk, rLk are sets of values, for the respective pairs, representing a modification function.
2. A discriminator according to claim 1 characterized in that the first set of values rHk for the first pair is the same as that rLk for the second pair.
3. A discriminator according to claim 1, in which the constants .PHI.H, .PHI.L are such that the arguments of the sine and cosine functions are zero at the mid-point of the filter.
4. A discriminator according to claim 2, in which the constants .PHI.H, .PHI.L are such that the arguments of the sine and cosine functions are zero at the mid-point of the filter.
5. A discriminator for frequency shift keyed signals comprising first filter means for accepting a first keying frequency, second filter means for accepting a second keying frequency, and a subtractor for generating the difference between the outputs of the filter means, characterized in that each filter means comprises: a pair of multipliers and means for supplying to the multipliers of the pair reference signals, mutually in phase quadrature, at the keying frequency which that filter means is to accept, each multiplier being followed by a finite impulse response filter, the two filters having the same response, representing a modification function, and means for forming the sum of an even function of the output of each of the two filters of the pair.
6. A discriminator according to claim 1, 2, 3, 4 or 5 characterized in that the or each modification function is substantially symmetrical about the mid-point of the relevant filter.
7. A discriminator according to any one of claims 1, 2, 3, 4 or 5 characterized in that the or each modification function has been derived by an iterative adjustment procedure such as to reduce distortion in the discriminator output.
8. A discriminator according to any one of claims 1, 2, 3, 4 or 5 in which the even function is the square.
9. A discriminator according to any one of claims 1, 2, 3, 4 or 5 characterized in that the filters have an impulse response of duration less than twice the symbol period of the signal to be demodulated.
10. A discriminator according to any one of claims 1, 2, 3, 4 or 5 characterized in that the filters have an impulse response of duration substantially equal to the symbol period of the signal to be demodulated.
11. A discriminator according to any one of claims 1, 2, 3, 4 or 5 including an interpolator for interpolating between successive samples output from the subtractor which are of different sign, to determine a zero-crossing instant.
12. A discriminator according to any one of claims 1, 2, 3, 4 or 5 characterized in that it is operable at a first baud rate employing a first pair of keying frequencies and at a second baud rate lower than the first employing a second pair of keying frequencies lower than the first pair and includes means operable when the discriminator operates at the second baud rate to subsample the signal to be demodulated before it is supplied to the first and second filter means.
13. A discriminator according to any one of claims 1, 2, 3, 4 or 5 characterized in that the or each modification function is substantially symmetrical about mid-point of the relevant filter, and in that the or each modification function has been derived by an iterative adjustment procedure such as to reduce distortion in the discriminator output.
14. A discriminator according to claims 1, 2, 3, 4 or 5 characterized in that the or each modification function is substantially symmetrical about the mid-point of the relevant filter, in which the even function is the square.
15. A discriminator according to any one of claims 1, 2, 3, 4 or 5 characterized in that the or each modification function has been derived by an iterative adjustment procedure such as to reduce distortion in the discriminator output, in which the even function is the square.
16. A discriminator according to claims 1, 2, 3, 4 or 5 characterized in that the or each modification function is substantially symmetrical about the mid-point of the relevant filter, and in that the filters have an impulse response of duration less than twice the symbol period of the signal to be demodulated.
17. A discriminator according to any one of claims 1, 2, 3, 4 or 5 characterized in that the or each modification function has been derived by an iterative adjustment procedure such as to reduce distortion in the discriminator output, and in that the filters have an impulse response of duration less than twice the symbol period of the signal to be demodulated.
18. A discriminator according to any one of claims 1, 2, 3, 4 or 5 in which the even function is the square, and in which the filters have an impulse response of duration less than twice the symbol period of the signal to be demodulated.
19. A discriminator according to claims 1, 2, 3, 4 or 5 characterized in that the or each modification function is substantially symmetrical about the mid-point of the relevant filter, and in which the filters have an impulse response of duration substantially equal to the symbol period of the signal to be demodulated.
20. A discriminator according to any one of claims 1, 2, 3, 4 or 5 characterized in that the or each modification function has been derived by an iterative adjustment procedure such as to reduce distortion in the discriminator output, and in which the filters have an impulse response of duration substantially equal to the symbol period of the signal to be demodulated.
21. A discriminator according to any one of claims 1, 2, 3, 4 or 5 in which the even function is the square, and in which the filters have an impulse response of duration substantially equal to the symbol period of the signal to be demodulated.
22. A discriminator according to any one of claims 1, 2, 3, 4 or 5 characterized in that the filters have an impulse response of duration less than twice the symbol period of the signal to be demodulated, and in which the filters have an impulse response of duration substantially equal to the symbol period of the signal to be demodulated.
23. A discriminator according to claims 1, 2, 3, 4 or 5 characterized in that the or each modification function is substantially symmetrical about the mid-point of the relevant filter, and including an interpolator for interpolating between successive samples output from the subtractor which are of different sign to determine a zero-crossing instant.
24. A discriminator according to any one of claims 1, 2, 3, 4 or 5 characterized in that the or each modification function has been derived by an iterative adjustment procedure such as to reduce distortion in the discriminator output, and including an interpolator for interpolating between successive samples output from the subtractor which are of different sign to determine a zero-crossing instant.
25. A discriminator according to any one of claims 1, 2, 3, 4 or 5 in which the even function is the square, and including an interpolator for interpolating between successive samples output from the subtractor which are of different sign to determine a zero-crossing instant.
26. A discriminator according to any one of claims 1, 2, 3, 4 or 5 characterized in that the filters have an impulse response of duration less than twice the symbol period of the signal to be demodulated, and including an interpolator for interpolating between successive samples output from the subtractor which are of different sign to determine a zero-crossing instant.
27. A discriminator according to any one of claims 1, 2, 3, 4 or 5 characterized in that the filters have an impulse response of duration substantially equal to the symbol period of the signal to be demodulated, and including an interpolator for the subtractor which are of different sign to determine a zero-crossing instant.
28. A discriminator according to claims 1, 2, 3, 4 or 5 characterized in that the or each modification function is substantially symmetrical about the mid-point of the relevant filter, and in that it is operable at a first baud rate employing a first pair of keying frequencies and at a second baud rate lower than the first employing a second pair of keying frequencies lower than the first pair and includes means operable when the discriminator operates at the second baud rate to subsample the signal to be demodulated before it is supplied to the first and second filter means.
29. A discriminator according to any one of claims 1, 2, 3, 4 or 5 characterized in that the or each modification function has been derived by an iterative adjustment procedure such as to reduce distortion in the discriminator output and in that it is operable at a first baud rate employing a first pair of keying frequencies and at a second baud rate lower than the first employing a second pair of keying frequencies lower than the first pair and includes means operable when the discriminator operates at the second baud rate to subsample the signal to be demodulated before it is supplied to the first and second filter means.
30. A discriminator according to any one of claims 1, 2, 3, 4 or 5 in which the even function is the square and in that it is operable at a first baud rate employing a first pair of keying frequencies and at a second baud rate lower than the first employing a second pair of keying frequencies lower than the first pair and includes means operable when the discriminator operates at the second baud rate to subsample the signal to be demodulated before it is supplied to the first and second filter means.
31. A discriminator according to any one of claims 1, 2, 3, 4 or 5 characterized in that the filters have an impulse response of duration less than twice the symbol period of the signal to be demodulated, and in that it is operable at a first baud rate employing a first pair of keying frequencies and at a second baud rate lower than the first employing a second pair of keying frequencies lower than the first pair and includes means operable when the discriminator operates at the second baud rate to subsample the signal to be demodulated before it is supplied to the first and second filter means.
32. A discriminator according to any one of claims 1, 2, 3, 4 or 5 characterized in that the filters have an impulse response of duration substantially equal to the symbol period of the signal to be demodulated, and in that it is operable at a first baud rate employing a first pair of keying frequencies and at a second baud rate lower than the first employing a second pair of keying frequencies lower than the first pair and includes means operable when the discriminator operates at the second baud rate to subsample the signal to be demodulated before it is supplied to the first and second filter means.
33. A discriminator according to any one of claims 1, 2, 3, 4 or 5 including an interpolator for interpolating between successive samples output from the subtractor which are of different sign, to determine a zero-crossing instant, and in that it is operable at a first baud rate employing a first pair of keying frequencies and at a second baud rate lower than the first employing a second pair of keying frequencies lower than the first pair and includes means operable when the discriminator operates at the second baud rate to subsample the signal to be demodulated before it is supplied to the first and second filter means.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| GB8719849 | 1987-08-21 | ||
| GB878719849A GB8719849D0 (en) | 1987-08-21 | 1987-08-21 | Fsk discriminator |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CA1336445C true CA1336445C (en) | 1995-07-25 |
Family
ID=10622647
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CA000575252A Expired - Fee Related CA1336445C (en) | 1987-08-21 | 1988-08-19 | Fsk discriminator |
Country Status (14)
| Country | Link |
|---|---|
| US (1) | US5065409A (en) |
| EP (1) | EP0305125B1 (en) |
| JP (1) | JP2910855B2 (en) |
| AT (1) | ATE77188T1 (en) |
| AU (1) | AU615475B2 (en) |
| CA (1) | CA1336445C (en) |
| DE (1) | DE3871880T2 (en) |
| DK (1) | DK194489A (en) |
| ES (1) | ES2033438T3 (en) |
| GB (1) | GB8719849D0 (en) |
| GR (1) | GR3005339T3 (en) |
| HK (1) | HK129096A (en) |
| IE (1) | IE61886B1 (en) |
| WO (1) | WO1989001726A1 (en) |
Families Citing this family (35)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| DE4234337C1 (en) * | 1992-10-12 | 1993-12-23 | Ant Nachrichtentech | Demodulator for FSK signals - uses digital comb filter for dividing input signal into partial signals of differing frequency |
| DE4234336C1 (en) * | 1992-10-12 | 1993-10-21 | Ant Nachrichtentech | Demodulator for FSK signals - uses multiphase filter with separate Fourier transformation network for each frequency shift frequency |
| DE19614979C2 (en) | 1995-04-20 | 2001-05-17 | Fujitsu Ltd | Radio frequency transceiver for data communication |
| US5748036A (en) * | 1996-07-30 | 1998-05-05 | United Microelectronics Corporation | Non-coherent digital FSK demodulator |
| US7515896B1 (en) | 1998-10-21 | 2009-04-07 | Parkervision, Inc. | Method and system for down-converting an electromagnetic signal, and transforms for same, and aperture relationships |
| US6694128B1 (en) | 1998-08-18 | 2004-02-17 | Parkervision, Inc. | Frequency synthesizer using universal frequency translation technology |
| US6091940A (en) | 1998-10-21 | 2000-07-18 | Parkervision, Inc. | Method and system for frequency up-conversion |
| US6061551A (en) | 1998-10-21 | 2000-05-09 | Parkervision, Inc. | Method and system for down-converting electromagnetic signals |
| US7236754B2 (en) | 1999-08-23 | 2007-06-26 | Parkervision, Inc. | Method and system for frequency up-conversion |
| US6813485B2 (en) | 1998-10-21 | 2004-11-02 | Parkervision, Inc. | Method and system for down-converting and up-converting an electromagnetic signal, and transforms for same |
| US6542722B1 (en) | 1998-10-21 | 2003-04-01 | Parkervision, Inc. | Method and system for frequency up-conversion with variety of transmitter configurations |
| US6370371B1 (en) | 1998-10-21 | 2002-04-09 | Parkervision, Inc. | Applications of universal frequency translation |
| US6560301B1 (en) | 1998-10-21 | 2003-05-06 | Parkervision, Inc. | Integrated frequency translation and selectivity with a variety of filter embodiments |
| US6061555A (en) | 1998-10-21 | 2000-05-09 | Parkervision, Inc. | Method and system for ensuring reception of a communications signal |
| US6049706A (en) | 1998-10-21 | 2000-04-11 | Parkervision, Inc. | Integrated frequency translation and selectivity |
| US7039372B1 (en) | 1998-10-21 | 2006-05-02 | Parkervision, Inc. | Method and system for frequency up-conversion with modulation embodiments |
| EP1014637B1 (en) | 1998-12-23 | 2006-05-03 | Asulab S.A. | Direct conversion receiver for FSK modulated signals |
| US6704558B1 (en) | 1999-01-22 | 2004-03-09 | Parkervision, Inc. | Image-reject down-converter and embodiments thereof, such as the family radio service |
| US6704549B1 (en) | 1999-03-03 | 2004-03-09 | Parkvision, Inc. | Multi-mode, multi-band communication system |
| US6853690B1 (en) | 1999-04-16 | 2005-02-08 | Parkervision, Inc. | Method, system and apparatus for balanced frequency up-conversion of a baseband signal and 4-phase receiver and transceiver embodiments |
| US6879817B1 (en) | 1999-04-16 | 2005-04-12 | Parkervision, Inc. | DC offset, re-radiation, and I/Q solutions using universal frequency translation technology |
| US7110444B1 (en) | 1999-08-04 | 2006-09-19 | Parkervision, Inc. | Wireless local area network (WLAN) using universal frequency translation technology including multi-phase embodiments and circuit implementations |
| US7693230B2 (en) | 1999-04-16 | 2010-04-06 | Parkervision, Inc. | Apparatus and method of differential IQ frequency up-conversion |
| US7065162B1 (en) | 1999-04-16 | 2006-06-20 | Parkervision, Inc. | Method and system for down-converting an electromagnetic signal, and transforms for same |
| US6785347B1 (en) * | 1999-04-22 | 2004-08-31 | 3Com Corporation | Nonlinear-least squares based method and apparatus for FSK signal decoding |
| US8295406B1 (en) | 1999-08-04 | 2012-10-23 | Parkervision, Inc. | Universal platform module for a plurality of communication protocols |
| US7010286B2 (en) | 2000-04-14 | 2006-03-07 | Parkervision, Inc. | Apparatus, system, and method for down-converting and up-converting electromagnetic signals |
| FR2815211A1 (en) * | 2000-10-11 | 2002-04-12 | Satellite Terminal Access | Digital numerical word filtering mechanism having transfer function with P coefficients cq with P whole number having period T such that cq equals cq plus T when q is smaller than P minus T and larger than one. |
| US7454453B2 (en) | 2000-11-14 | 2008-11-18 | Parkervision, Inc. | Methods, systems, and computer program products for parallel correlation and applications thereof |
| US7072427B2 (en) | 2001-11-09 | 2006-07-04 | Parkervision, Inc. | Method and apparatus for reducing DC offsets in a communication system |
| US7460584B2 (en) | 2002-07-18 | 2008-12-02 | Parkervision, Inc. | Networking methods and systems |
| US7379883B2 (en) | 2002-07-18 | 2008-05-27 | Parkervision, Inc. | Networking methods and systems |
| US7672356B2 (en) | 2005-05-19 | 2010-03-02 | Itt Manufacturing Enterprises, Inc. | Method and apparatus for detection of a frequency coded sequence in the presence of sinusoidal interference |
| EP2469783B1 (en) * | 2010-12-23 | 2017-12-13 | The Swatch Group Research and Development Ltd. | FSK radio signal receiver with high-sensitivity demodulator and method for operating said receiver |
| US11515863B2 (en) * | 2021-03-08 | 2022-11-29 | Rohde & Schwarz Gmbh & Co. Kg | Comb signal generator and method of providing a phase and amplitude reference |
Family Cites Families (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US3803501A (en) * | 1972-11-17 | 1974-04-09 | Ibm | Frequency discriminator using digital non-recursive filters |
| US4437068A (en) * | 1981-02-19 | 1984-03-13 | Thomson-Csf | FSK Demodulator for frequency-modulation modem |
| US4479092A (en) * | 1981-11-20 | 1984-10-23 | At&T Bell Laboratories | Digital frequency-shift keyed demodulator |
| US4475219A (en) * | 1982-08-18 | 1984-10-02 | General Electric Company | Demodulator for frequency-shift-keyed binary data signals |
| US4568882A (en) * | 1984-10-01 | 1986-02-04 | National Semiconductor Corporation | Digital FSK demodulator circuit |
| ATE56573T1 (en) * | 1985-07-03 | 1990-09-15 | Siemens Ag | DIGITAL FILTERS, ESPECIALLY FOR A DATA RECEIVER. |
| US4755761A (en) * | 1986-06-26 | 1988-07-05 | General Electric Company | Zero intermediate-frequency demodulator |
-
1987
- 1987-08-21 GB GB878719849A patent/GB8719849D0/en active Pending
-
1988
- 1988-08-19 AU AU22569/88A patent/AU615475B2/en not_active Ceased
- 1988-08-19 JP JP63506697A patent/JP2910855B2/en not_active Expired - Lifetime
- 1988-08-19 WO PCT/GB1988/000687 patent/WO1989001726A1/en unknown
- 1988-08-19 DE DE8888307727T patent/DE3871880T2/en not_active Expired - Fee Related
- 1988-08-19 AT AT88307727T patent/ATE77188T1/en not_active IP Right Cessation
- 1988-08-19 CA CA000575252A patent/CA1336445C/en not_active Expired - Fee Related
- 1988-08-19 ES ES198888307727T patent/ES2033438T3/en not_active Expired - Lifetime
- 1988-08-19 US US07/354,435 patent/US5065409A/en not_active Expired - Lifetime
- 1988-08-19 EP EP88307727A patent/EP0305125B1/en not_active Expired - Lifetime
- 1988-08-22 IE IE255888A patent/IE61886B1/en not_active IP Right Cessation
-
1989
- 1989-04-20 DK DK194489A patent/DK194489A/en not_active Application Discontinuation
-
1992
- 1992-08-04 GR GR920401677T patent/GR3005339T3/el unknown
-
1996
- 1996-07-18 HK HK129096A patent/HK129096A/en not_active IP Right Cessation
Also Published As
| Publication number | Publication date |
|---|---|
| IE882558L (en) | 1989-02-21 |
| EP0305125A1 (en) | 1989-03-01 |
| EP0305125B1 (en) | 1992-06-10 |
| WO1989001726A1 (en) | 1989-02-23 |
| JPH02501181A (en) | 1990-04-19 |
| DE3871880T2 (en) | 1993-01-14 |
| AU615475B2 (en) | 1991-10-03 |
| IE61886B1 (en) | 1994-11-30 |
| GB8719849D0 (en) | 1987-09-30 |
| DE3871880D1 (en) | 1992-07-16 |
| ATE77188T1 (en) | 1992-06-15 |
| JP2910855B2 (en) | 1999-06-23 |
| HK129096A (en) | 1996-07-26 |
| DK194489D0 (en) | 1989-04-20 |
| DK194489A (en) | 1989-04-20 |
| AU2256988A (en) | 1989-03-09 |
| ES2033438T3 (en) | 1993-03-16 |
| US5065409A (en) | 1991-11-12 |
| GR3005339T3 (en) | 1993-05-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CA1336445C (en) | Fsk discriminator | |
| US4726041A (en) | Digital filter switch for data receiver | |
| CA2035785C (en) | Method and apparatus for optimally autocorrelating an fsk signal | |
| US4887280A (en) | System for detecting the presence of a signal of a particular data rate | |
| RU2099892C1 (en) | Method and device for relative phase modulated signal demodulation | |
| US4608540A (en) | Phase-shift keying demodulator | |
| US4794556A (en) | Method and apparatus for sampling in-phase and quadrature components | |
| JPS62115951A (en) | Demodulator | |
| Santhanam et al. | Energy demodulation of two-component AM-FM signal mixtures | |
| US4755961A (en) | Digital tank circuit | |
| US4439737A (en) | Phase locked loop, as for MPSK signal detector | |
| EP0643511B1 (en) | Synchronization circuit for subcarrier signal | |
| US4535295A (en) | Method and device for controlling the phase of timing signal | |
| DE3015217C2 (en) | Transmission method and corresponding transmitter and receiver for the transmission of two-valued data symbols | |
| US5554955A (en) | Method and apparatus for removing the effects of co-channel interference from the message on a dominant frequency modulated carrier and for recovering the message from each of two co-channel carriers | |
| US7096244B2 (en) | Digital filter and reference signal cancelling device and method using the same | |
| EP0294738A2 (en) | A method, a transmitter and a receiver for the transmission of digital auxiliary information | |
| RU2039416C1 (en) | Digital adaptive receiver of digital signals | |
| US5373246A (en) | Digital FFSK demodulator | |
| SU1046906A1 (en) | Phase demodulator | |
| Kouvaras et al. | A technique for a substantial reduction of the quantization noise in the direct processing of delta-modulated signals | |
| Santhanam et al. | Energy demodulation of two-component AM-FM signals with application to speaker separation | |
| KR890001387B1 (en) | Phuse shift keying detector circuit | |
| KR940004196B1 (en) | Gmsk digital modulate apparatus and method | |
| Plotkin et al. | On estimation of noise variance in the presence of a narrow-band signal |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| MKLA | Lapsed |