WO1993002506A1 - Programmable noise bandwidth reduction by means of digital averaging - Google Patents
Programmable noise bandwidth reduction by means of digital averaging Download PDFInfo
- Publication number
- WO1993002506A1 WO1993002506A1 PCT/US1992/005849 US9205849W WO9302506A1 WO 1993002506 A1 WO1993002506 A1 WO 1993002506A1 US 9205849 W US9205849 W US 9205849W WO 9302506 A1 WO9302506 A1 WO 9302506A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- samples
- signal
- sample
- digital signal
- weighted
- Prior art date
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B1/00—Details of transmission systems, not covered by a single one of groups H04B3/00 - H04B13/00; Details of transmission systems not characterised by the medium used for transmission
- H04B1/06—Receivers
- H04B1/10—Means associated with receiver for limiting or suppressing noise or interference
- H04B1/1027—Means associated with receiver for limiting or suppressing noise or interference assessing signal quality or detecting noise/interference for the received signal
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B1/00—Details of transmission systems, not covered by a single one of groups H04B3/00 - H04B13/00; Details of transmission systems not characterised by the medium used for transmission
- H04B1/06—Receivers
- H04B1/10—Means associated with receiver for limiting or suppressing noise or interference
- H04B1/1027—Means associated with receiver for limiting or suppressing noise or interference assessing signal quality or detecting noise/interference for the received signal
- H04B2001/1054—Means associated with receiver for limiting or suppressing noise or interference assessing signal quality or detecting noise/interference for the received signal by changing bandwidth
Definitions
- the invention is in the field of signal demodula ⁇ tion and has application in information transmission systems where predemodulation, that is, predetection, noise bandwidth reduction is advantageous to increase the signal-to-noise (S/N) ratio prior to signal detection. More specifically, the present invention is directed towards a variable-rate data filter for digital data transmission having optimum adjacent channel rejection characteristics.
- Modulated signals carrying information such as video, data, music and speech are generally contami ⁇ nated by noise. Efficient demodulation requires distinguishing the information from the noise.
- the demodulation process includes several steps.
- the receiver may receive, at its antenna, an informa- tion signal modulated on a radio frequency (RF) carrier.
- the signal may then undergo frequency conversion to the intermediate frequency (IF) band.
- the information signal, at baseband is recovered from the IF signal by a suitable detector.
- the IF signal produced from a received RF signal by subjecting the RF signal to a mixing or filtering process, is subse- quently applied to a data detector for recovering, at baseband, the information content of the input signal.
- the system must be responsive to a variable rate signal, thus the IF bandwidth must be broad enough to process the highest expected data rate, although at any point in time the receiver may be detecting a lower rate and thus narrower band signal.
- the noise bandwidth is not limited to the frequency spectrum, that is, bandwidth, of the received signals
- the bandwidth of the receiver's front end that is, prior to detection, must be scaled with the received signalling rate to prevent noise overload, signal suppression, and distortion in subsequent digital processing stages.
- filter switching mechanism limiting the IF bandwidth based on. the receive-signalling rate.
- a conventional filter switching arrangement for limiting the noise bandwidth at a receiver front end is illustrated in Fig. 1.
- the Fig. l arrangement may be used in a receiver of a digital data transmission system to select a bandwidth at IF sufficient to pass data signals transmitted at a selected one of several data rates, while suppressing noise outside that bandwidth.
- the Fig. l arrangement includes an input terminal 6 receiving the incoming modulated signal and noise at IF.
- the input terminal 6 is connected to a commutator 4 of a rotary switch 2.
- the switch 2 has a number of fixed contacts 8 1 - 8 n each selectively connected to the commutator 4 through rotation of the commutator.
- Each fixed contact 8, - 8 n is electrically connected to a respective IF filter 10 1 -10 n .
- the center frequen ⁇ cies F T - F n and bandwidths BW, - BW n of the IF filters 10, - 10 n are selected on the basis of the data rates the receiver is designed to accept.
- the outputs from the IF filters are input to a power combiner 12.
- the output from the power combiner is an IF signal whose bandwidth is scaled to the signalling rate of the received signal, that is, somewhat greater than, but proportional to, the bandwidth of the received data or symbol rate, thereby reducing the noise bandwidth prior to data detection in a detector 14.
- the reduced noise bandwidth prevents noise overload, signal suppression and distortion in the latter processing stages of the detector 14.
- an RF signal modulated by a data signal at the selected symbol rate, is converted to IF by conventional mixing or filtering and then applied to input terminal 6.
- One of the parallel sets of filter paths is selected by rotating commutator 4 based on the symbol rate of the data signal modulating the IF signal.
- the selected one of the IF filters 10, - 10 n limits the bandwidth of the IF signal prior to detection, thereby reducing the noise bandwidth which initially extends over the entire IF spectrum.
- This conventional arrangement suffers from several disadvantages. For example, it is expensive and cumbersome to implement, and it produces gain and phase variations from one path to another as well as from one unit to another.
- Boxcar filtering is another technique which may be used to reduce predetection noise.
- Boxcar filter ⁇ ing involves averaging the incoming signal, with noise reduction the expected result since noise is theoreti ⁇ cally random. Over time many random signals have substantially equal positive and negative components, and thus averaging will tend to reduce the noise component of such a signal toward zero.
- boxcar filtering is not applicable to digital data demodulation since with the boxcar technique averaging must be done over many symbols and the exact period of the signal to be averaged must be known.
- the present invention is directed to a system including a technique and implementing apparatus which do not experience the aforementioned disadvantages of either the conventional bandwidth switching technique or boxcar filtering technique.
- An object of the invention is to reduce the pre ⁇ detection noise bandwidth of a modulated communication signal. Another object of the invention is to reduce predetection noise bandwidth without expensive and cumbersome equipment.
- Another object of the invention is to reduce the rate at which samples need be handled in subsequent processing.
- a still further object of the invention is to reduce predetection noise bandwidth using averaging over a single data symbol and without prior determina ⁇ tion of the exact signal period. It is also an object of the invention to utilize the predetection averaging of the invention to form a data filter thus combining the function of noise bandwidth reduction and data shaping in a single unit.
- Another object of the present invention is to provide a pre-averager circuit, particularly suited for digital signal processing (DSP) techniques, for programmably reducing the incoming noise bandwidth of a receiver when it is operated over a broad range of potential signal bandwidths.
- DSP digital signal processing
- Still another object of the present invention is to provide a pre-averager circuit having improved operational response in a tightly packed adjacent channel environment.
- a pre-averager is positioned in a receiver front end for processing the input signal. Assuming the information signal to be digital data, averaging of the samples will be over a single symbol. While the invention is not limited to digital data transmission systems, for convenience it will be described hereinafter in connection with such a system since the pre-averager according to the invention may be configured as a data filter for data detection. However, the pre-averager of the invention may also be used in receivers for other types of information signals such as video, speech and music.
- the pre-averager of the invention includes a digital averaging module which samples an incoming signal, converted to baseband, at least at twice the noise bandwidth determined by the single, input IF filter. The samples are averaged over a defined averaging interval, ordinarily set as a function of the input data rate, and are then clocked out of the averager at a lower output data rate.
- An output sampling rate of two samples per symbol has been found acceptable in the preferred embodiment hereinafter described for noise bandwidth reduction, although other output rates may also be acceptable.
- the defined averaging interval is the input sampling rate divided by the output sampling rate.
- the invention appears to conflict with well known sampling theory concepts in that the output sample rate of the pre-averager is typically lower than twice the input noise bandwidth, although it is never less than twice the input signal band ⁇ width. Further, groups of incoming samples may be replaced by a single sample which is representative of their average value. While the output sample will be slightly contaminated with aperture distortion, this may be easily compensated for by transmit side equal ⁇ ization, for example.
- the pre-averager is configured to form a data filter.
- An optimum data filter in digital data transmission must have two fundamental attributes. One, its frequency response should be matched to the transmitted signal spectrum. Two, its combined transmission and reception impulse response should exhibit equally spaced zero crossings so that inter- ference does not occur in the detection of adjacent symbols.
- a data filter approaching the optimum conditions is realized by the pre-averager of the invention when the output sampling rate is reduced to one sample per symbol in the detection path.
- the pre-averager implemented data filter for the detection of asynchronous data may be constructed as two parallel paths per I or Q data channel, each containing a pre-averager.
- asynchronous it is meant digital data for which the exact clock frequency and phase are unknown and must be recovered.
- the first of the two pre-averagers supplies an even sample used for data detection, carrier phase recovery and automatic gain control (AGC) estimation.
- AGC automatic gain control
- the second of the two pre-averagers provides an odd sample used for clock recovery. Since averaging according to the teachings of the invention produces only one output sample per symbol in each of two paths, a feature of the invention is the reuse of input samples to provide the necessary samples for data detection and clock recovery.
- the pre-averager implemented data filter for the detection of asynchronous data may be constructed as two parallel paths, each containing a pair of pre-averagers.
- Each of the pre-averagers in a corresponding path accumulates and performs a weighted average of digital samples over a period equal to twice the symbol interval.
- the pair of pre- averagers is coupled to a multiplexer circuit, which in response to clock signals based on the sampling rate, interlaces the average values produced in each pre-averager to produce an output signal at a rate of one sample per symbol interval.
- accumulating samples over two symbol intervals improves the data filter performance when the desires signal is located in a tightly packed adjacent channel environment.
- the pre-averager according to this latter embodi ⁇ ment may include circuitry for weighting the samples based on finite impulse response (FIR) coefficients stored, for example, in random access memory coupled to the weighting circuitry.
- FIR finite impulse response
- the response of the pre- averager may be optimized by changing the filter coefficients stored in memory.
- the pre-averager performs the function of rate reduction, that is, it reduces the rate at which subsequent circuits must operate in processing the data to a new, lower, fixed, rate. This reduces the complexity and expense of those circuits.
- Fig. 1 illustrates a conventional bandwidth switching arrangement for reducing predetection noise bandwidth in a variable rate data receiver.
- Fig. 2 illustrates a predetection noise bandwidth reducing pre-averager according to a first embodiment of the invention.
- Fig. 3 is a timing diagram for illustrating the operation of the Fig. 2 circuit.
- Fig. 4 is a general representation of the sin x/x versus x relationship for an input information signal and the overall signal bandwidth including noise at several sampling rates.
- Fig. 5 illustrates a pre-averager data filter according to a second embodiment of the invention.
- Fig. 6 is a representation of the sampling point offset and sample reuse features incorporated in the pre-averager data filter of Fig. 5.
- Fig. 7 is a timing diagram for illustrating the operation of the Fig. 5 circuitry.
- Fig. 8 illustrates a pre-averager data filter according to a another embodiment of the invention.
- Fig. 9 is a representation of the time-staggered weighting factors incorporated in the pre-averager data filter of Fig. 8.
- Fig. 10 is a timing diagram for illustrating the operation of the Fig. 8 circuitry.
- an IF filter and a predetection digital averaging module effect noise bandwidth and rate reduction.
- Noise bandwidth reduction according to this invention could take place at baseband, carrier, or IF frequen- cies. However, it is convenient to initiate the noise bandwidth reduction at baseband and therefore, a preferred embodiment of the invention has the pre- averaging according to the invention initiated there. Also, as previously stated herein, noise bandwidth reduction may be accomplished for video, speech, music or other signals besides data signals. Because the present invention is especially useful in data trans ⁇ mission systems, the preferred embodiment is disclosed in relation to a data transmission system. In systems other than data transmission systems, variable rate reduction might be used to narrow the signal bandwidth in response to varying noise conditions.
- a preferred embodiment of the pre-averager apparatus for noise bandwidth reduction includes an IF filter 20 passing the carrier modulated information signals and noise.
- the output from the IF filter is connected to a first input of a mixer 22.
- a second input to the mixer 22 is connected to local oscillator •24 for converting the IF signal to base ⁇ band.
- the output from mixer 22 is an analog baseband signal i(t) which includes noise whose frequency spectrum is limited by the bandwidth of IF filter 20.
- the signal i(t) is input to analog-digital (A/D) converter 26 which converts signal i(t) including its noise components to digital form.
- the input sampling rate Rs AMP is relatively high to prevent aliasing of noise. Typically, the input sampling rate would be at or greater than twice the noise bandwidth of the IF filter 20.
- the sampling rate R ⁇ p for the A/D conver ⁇ sion is set by a clock signal generated by a sample clock (not shown) input at terminal 39.
- the output I k from the A/D converter 26 is applied to an accumulator 28, where it is added to the value, ⁇ k , output from the accumulator in response to the preceding sample I k nourish.,. This accumulated value is fed back from the accumulator output to a second input thereto, through a one sample delay 30.
- the one sample delay 30 may be a latch circuit. It is to be understood that the output from A/D converter 26 is a parallel arrangement of M bits.
- the delay 30 is reset by a reset pulse from a timing generator 38. The reset pulse determines the averaging interval as it sets to zero the second input to the accumulator 26 at the end of the averaging interval.
- the averaging interval is conveniently set to a power of two (2 ) .
- the averaging interval information is input to the binary point shifter 32 at a second input thereto connected to line 33.
- a latching circuit 34 receives the L bits from the binary point shifter.
- the latching circuit functions as a buffer to assure the presence of all L bits representing the average ⁇ I k/H > for further processing.
- the latching circuit which may be comprised of L parallel flip flops, is clocked at the lower output sampling rate of R ⁇ p ⁇ ".
- further processing of the information signal and particularly detection thereof by a suitable detector is at the lower SA ⁇ /2 output sampling rate.
- Fig. 2 includes a digital lowpass filter 36 with sine * compensation. This conventional device is optional and used when further shaping of an averaged data signal is desirable.
- the waveform i(t) represents an input signal at the mixer 22 output terminal. This signal is sampled at the rate Rs AHP and converted to a digital signal in A/D converter 26.
- the output of the A/D converter I k is a stream of M bit-wide samples, designated in Fig. 3 by sample numbers 0, 1, 2, 3 .... That is, the first sample is designated, 0, the second sample, 1, and so forth.
- the first sample, 0, passes through the accumulator, it is delayed by one sampling period, T s , in the delay 30 as illustrated in the ⁇ k ... timing diagram where Ts ⁇ l/R fj ⁇ p .
- the delayed first sample coincides in time with the next sample, 1, as can be readily seen from the I k and ⁇ , diagrams.
- the first sample, 0, is added to the second sample 1 in accumulator 28 as illustrat- ed in timing diagram, ⁇ k , of Fig. 3. Since - Rs AHp /2 and therefore, a reset pulse is applied to the delay 30 after the second sample as shown in the Reset diagram of Fig. 3.
- the output from the delay 30 is thus zero when the third sample, 2, is input to the accumulator.
- the process continues as illustrated in Fig. 3, with every two samples being added and the delay 30 output being reset to zero after the sum is generated.
- the sum of the first and second samples passes through the binary point shifter, clocked at Rs AHP /2 in this example, to produce a signal representing the average value of the sum, ⁇ I k/N >.
- averaging takes place within a single symbol, thus eliminating the prior art requirement for averaging over several symbols and the need to know the exact symbol period to accomplish the averaging process.
- the output bandwidth is narrowed to reduce the noise bandwidth prior to detection.
- Fig. 4 which represents the sin x/x versus x plot for various sampling rates, R SAHP /2 M .
- the cross-hatched portion represents the information signal bandwidth.
- B H is the noise equivalent bandwidth of the IF filter.
- R jy ⁇ p in accordance with conventional sampling theory, is selected to be more than twice B N .
- ⁇ I /n a number of output samples (0+1, 2+3, etc.), equal to one half the number of input samples of the input waveform, each output sample being an average of two input samples, are latched and then may be applied to digital lowpass filter 36 to produce a properly shaped digital baseband output, reduced in noise by the above described averaging process. Note that this filter will operate at a lower sample rate due to the preceding pre-averager and the rate reducing aspect of the invention.
- the noise reduced digital baseband output is applied to a suitable detector (not shown) for demodulation.
- the digital pre-averager described hereinabove is configured as a receive data filter, eliminating the need for a separate data filter following the pre-averager.
- a data filter in a digital data transmission system should have two fundamental attributes. One, its frequency response should be matched to the transmitted signal spectrum. Two, its combined transmission and reception impulse response should exhibit equally spaced zero crossings so there is no inter-symbol interference at adjacent symbol detection points.
- the averaging and reduced sampling rate realized with the pre-averager of the invention provides a mechanism by which the frequency response can be substantially matched to the frequency spectrum of the transmitted signal.
- a data filter is realized with the pre-averager dis ⁇ closed herein when averaging occurs over a single symbol and the averaging is effected to produce one output sample per symbol in the detection path.
- the pre-averager output bandwidth closely approximates that of the transmitted signal. That is, when the pre-averager output is one sample per symbol, the sin x/ ⁇ aperture response emulates the receive data filtering operation.
- the pre-averager data filter of the invention has a filter response slightly different from that of a conventional data filter.
- predistortion that is equalization, may be applied at the transmit end.
- the transmit end equalization must compensate for a 0.9 dB excess loss at the Nyquist frequency and a softer overall response.
- FIG. 5 An embodiment of the data filter according to the second embodiment of the invention is illustrated in Fig. 5.
- the data filter implemented to detect asynchronous data, includes two parallel paths.
- An even or detection sampling point path is for data detection, carrier phase recovery, and AGC estimation.
- An odd or zero crossing path is for clock recovery.
- Implementation of the data filter using parallel arranged pre-averagers includes at least the following two novel concepts. First, in averaging down to one sample per symbol and concomit- tently reducing the bandwidth to the range of a data filter, input samples must be reused. Second, the sampling points of the incoming data must be offset to provide properly centered samples to avoid attendant performance loss resulting from distortion in the averaged output.
- the pre- averager data filter includes an A/D converter 26 receiving baseband signal i(t) and outputting the digitized samples, I k , representing input signal, i(t) .
- the digitized samples are simultaneously applied to the even output sample accumulator and odd output sample accumulator.
- the even output sample accumula ⁇ tor includes accumulator 28 E in the form a summing circuit, a one sample delay 30 ⁇ in the form of flip— flop circuits, a binary point shifter 32 E in the form of a barrel shifter, and latching circuit 34 E in the form of flip-flop circuits.
- the odd output sample accumulator is similarly constituted.
- each of the parallel paths is the same as the operation of the circuitry of Fig. 3.
- sampling points are offset from the ideal sample points at the peaks and zero crossings and samples are reused. This concept of sample reuse will be discussed with reference to Fig. 6 which illustrates sample reuse together with the sampling point offset feature of the invention.
- the input symbols represent a 1/0 symbol pattern, with Fig. 6, for simplicity, illustrating two input samples per symbol which are to be averaged down to one output sample per symbol in each of the even and odd output paths.
- Input samples are taken at offset sampling points a, b, C, d, and e.
- samples b and c are averaged, as are samples d and e.
- the average A of samples b and c are shown superposed on an imaginary waveform labelled Even Outputs. It is of course understood that average sample A occurs after both samples b and c have occurred and thus the even and odd outputs do not have the phase relationship illustrated in Fig. 6. This figure simply depicts the sampling point offset and sample reuse features of the invention.
- the average sample B is an average of input samples d and e.
- the even path provides one output sample per symbol which uniquely defines the symbol value, 1 or 0.
- a first zero crossing at output average sample C is generated by averaging input samples a and b. It is to be noted that sample b is used both for the data detection path and the clock recovery path, thus the sample reuse feature of the invention may now be appreciated.
- a second zero crossing at output average sample D is generated by averaging input samples c and d. In this case, input sample d is reused for the same reason input sample b is reused. In fact, note that all samples are used twice.
- the invention is not limited to the case of only two input samples per symbol, nor is it limited to the use of two parallel paths, as discussed in greater detail below.
- Fig. 7 is a timing diagram showing the opera ⁇ tion of the even and odd accumulator paths where four samples per symbol are taken on the input signal i(t) .
- This signal is a typical preamble with alternating 1/0 symbols. That is, in the Fig. 7 example, the input sampling rate R SAMP equals 4 samples per symbol while the output sampling rate R s equals 1 sample per symbol. R s is the complement of R s .
- the digital samples I k represent the input signal values at points 1-4 of a first symbol, at points 5-8 of a second symbol and 9-12 of a third symbol.
- timing diagram E/O Ace Input The timing of the four samples per symbol are illustrated in timing diagram E/O Ace Input as timing blocks 1-12.
- samples 1, 1+2 and 1+2+3 are fed back to the summing circuit through delay 30 E as shown in timing diagram Even Ace Feedback In.
- the summing circuit sums samples 1, 2, 3, and 4 to produce a summed output as shown in timing diagram Even Ace.
- an even reset pulse sets the delay 30 E to zero to begin the summing process again, this time with samples 5-8.
- an output sample pulse R s rises to clock the summed samples 1-4, scaled in the binary point shifter 32 E , to the parallel array of latching flip-flips represented by flip-flops 34 E .
- the outputs from the latching flip-flops is the signal I E representing the data value for the symbol corresponding to samples 1-4.
- the odd path operates in the manner of the even path except that as a result of the timing differences between R s and R s , and the even reset.
- Reset, and odd reset, Reset different input samples are averaged.
- the odd path like the even path, receives the samples 1-12.
- the odd accumulator feedback input receives samples (-2) , (-1) , and 1 which are summed with sample 2 before the delay 30 0 is reset to zero by a Reset' pulse.
- the output sampling pulse R s rises to clock the sum of samples (-2) , (-1) , and 2 into latching cir- cuits 34 0 after being scaled by binary point shifter 32 0 .
- This process continues with samples 3-6, and then samples 7-10 as can be appreciated from Fig. 7.
- the odd path averages samples 3, 4 from a first symbol with samples 5, 6 of the next symbol so that the zero crossing sample between symbols is generated.
- averaging of samples 7, 8 of the second symbol with samples 9, 10 of the third symbol generates the zero crossing sample between the second and third symbols.
- the band ⁇ width of the pre-averager advantageously can be set greater than or on the order of that of the incoming signal, such that the noise variance is reduced by the averaging factor while the signal is passed. It will be noted that when the pre-averager bandwidth is set nearly equal to the signal bandwidth, the pre-averager acts as the demodulator's receive detection filter.
- two parallel accumulators average digitally quantized samples coming from the output of A/D converter 26. That is, the average value of every N incoming samples is represented by an even and an odd sample whose repeti- tion rate is 1 /N times that of the input signal i(t) .
- one of these streams of samples is processed for data detection, carrier phase recovery, and level control while the other stream is used for symbol timing recovery.
- Each stream advanta- geously has a rate of one sample/incoming data symbol.
- Fig. 5 has been demon ⁇ strated as being able to reduce the incoming noise bandwidth
- this embodiment has been found to provide less than optimum performance in the presence of tightly packed adjacent channels. This degradation is apparently due to the relatively soft rolloff filter ⁇ ing characteristics of the circuit, which effectively has a time domain impulse response that is only one data symbol wide when a digital data stream is trans ⁇ mitted.
- Another preferred embodiment of the present invention will now be described while referring to Figs. 8-10-.
- two fundamental improve- ents have been incorporated in the embodiment illus ⁇ trated in Fig. 8: 1) multiplying the input samples by a weighting function and 2) extending the length of the averaging interval from one to two symbols.
- another embodiment of the pre-averager circuit according to the present inven ⁇ tion comprises an A/D converter 26 receiving an analog baseband signal i(t) and providing a digitized samples I k is operatively connected to odd and even binary point shifters, generally denoted 32, via two pairs of multiplier accumulators (MACs) , generally denoted 50, and a pair of multiplexers, generally denoted 29.
- each multiplexer 29 is operatively coupled to a corresponding pair of MACs 50 on its input side and to a corresponding shifter 32 on its output side.
- the output signals produced by the MACs 50 are advan ⁇ tageously combined to yield even and odd sample outputs I e and I 0 .
- the output I e of I 0 of multiplexers 29 E or 29 0 respectively, advantageously is an inter- laced signal produced from the output of the pair of MACs 50 coupled to the corresponding multiplexer 29, as discussed in detail below.
- MAC 50a comprises a multiplier circuit, generally denoted 27, which is coupled to A/D converter 26 to receive digitized samples I k , and which outputs a weighted I k , discussed in greater detail below, to an accumulator circuit, generally denoted 28.
- the output of, for example, accumulator 28 a which corresponds to the output of MAC 50a, is provided to both multiplexer 29 E and a one sample delay, generally denoted 30.
- Each of the MACs 50 is connected to one output of a conventional first-in/first-out buffer circuit 40 receiving weighting factors from a memory 42 in response to a control signal provided by an address sequence generator 44.
- buffer 40, memory 42 and generator 44 are conventional devices and will appre- ciate their use without further detailed description of these components.
- the input signal i(t) is a signal providing non-return-to-zero (NRZ) digital data transmission.
- N non-return-to-zero
- the incoming signal i(t) is sinusoidal and is sampled at a rate N-times the data symbol rate (where N > 2) , thereby producing the digitized samples I k shown in Figure 9.
- These incoming samples are multiplied by samples corresponding to four time-staggered versions of the same impulse response weighting function, shown as h 0 , h,, h 2 and h 3 in Fig. 9.
- the signals h 0 - h 3 in Fig. 9 illustrate time-staggered versions of a raised cosine impulse response function.
- the samples of the raised cosine response are periodic over 2 symbols, and advantageously are staggered in half-symbol (N/2) increments.
- the samples h 0 - h 3 are the weighting factors applied to MACs 50, thus provid ⁇ ing each of the MACs 50 with an amplitude weighting factor. From the discussion above, it will be apparent that the two MAC pairs must be interlaced as was shown in Fig. 8 to provide even and odd outputs at a one sample/symbol rate.
- Each output sample is then a weighted sum of contiguous groups of 2N input samples, where the accumulator 28 outputs are sampled and reset to zero after each input group of 2N samples is processed.
- the timing is such that the interlaced even samples I e occur at their peak values, whereas the odd samples I 0 are zero.
- the accumulators 28 are reset to zero in the time slot when their respective weighting coeffi ⁇ cients are zero. In this situation, no multiplication is necessary.
- Fig. 10 The reset and interlace timing for the pre- averager circuit illustrated in Fig. 8 is shown in Fig. 10, in which the input digitized sample I k is again illustrated as a sinusoid and the sampling rate, NR SAMP , is illustrated immediately below the I k curve. From the pre-averager circuit of Fig. 8 it will be appreciated that four reset signals, Reset 0 - Reset 3 , advantageously are provided for resetting delays 30 and each of these reset signals are shown in Fig. 10.
- each reset is conducted when the associated one of the weighting factors h 0 - h 3 is zero, and, as can clearly seen from comparing Figs 9 and 10, each reset Reset 0 - Reset 3 is applied when the corresponding weighting factor h 0 - h 3 output by buffer 40 is zero.
- the output from each of the multiplexers 29 provides an interlaced resultant signal, I e or I 0 , from the weighted accumulated output signals produced by its corresponding pair of MACs 50.
- the output of MAC 50a is interlaced with the output of MAC 50b by multiplexer 29 E to provide the output signal I e .
- multiplexer 29 E interlaces the outputs of MACs 50a and 50b in response to clock signals R s /2 and R s /2 , respectively, which are derived for a sample clock (not shown) operating at a rate R s .
- Clock signals R s /2 and R s /2 which are time-staggered to provide complementary sampling periods, are advan ⁇ tageously applied to multiplexer 29 E it control the interlace function provided by that component.
- Fig. 10 illustrates that each accumulator 28 accumulates weighted digital signal samples over a period of 2N samples and that the output of each accumulator 29 is provided to the corresponding multiplexer 29 in response to the leading edge of a clock signal, generally denoted R s /2. It will be apparent that the clock signal controlling the ulti- plexers 29 occurs immediately proceeding the corre ⁇ sponding reset pulse applied to the delays 30.
- the MACs 50 advantageously perform four parallel time-staggered finite impulse response (FIR) filtering operations. It will also be apparent that the output sample rate of the even and odd MAC pairs has been decimated by l / N, so that only one averaged sample point from each filter convolution is available for further signal process ⁇ ing. Moreover, the pre-averager filter bandwidth will be on the order of half the symbol frequency, which is a direct result of the length of the accumulator 30's averaging interval. It will also be appreciated that the shifters 32 advantageously are provided to proper ⁇ ly scale the most significant output bits, since the accumulator outputs will vary depending on the number of samples in the averaging interval.
- FIR finite impulse response
- the digital processing circuitry downstream of the A/D converter 26 advantageously can be implemented with a single digital signal processing (DSP) chip, which, in conventional configurations comprises a single MAC.
- DSP digital signal processing
- the DSP chip would be reaccessed four times in order to provide the output signals I e and I 0 provided by the embodiment of the invention shown in Fig.. 8. It will be further appreciated that reaccessing the DSP chip would reduce the maximum processing rate by a factor inversely proportional to the number of time the DSP chip is reaccessed.
Abstract
Description
Claims
Priority Applications (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP5502890A JPH06509219A (en) | 1991-07-16 | 1992-07-16 | Programmable noise bandwidth reduction through digital averaging |
CA002102500A CA2102500C (en) | 1991-07-16 | 1992-07-16 | Programmable noise bandwidth reduction by means of digital averaging |
EP92915452A EP0669058A1 (en) | 1991-07-16 | 1992-07-16 | Programmable noise bandwidth reduction by means of digital averaging |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US07/730,426 US5216696A (en) | 1989-12-22 | 1991-07-16 | Programmable noise bandwidth reduction by means of digital averaging |
US730,426 | 1991-07-16 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO1993002506A1 true WO1993002506A1 (en) | 1993-02-04 |
Family
ID=24935316
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/US1992/005849 WO1993002506A1 (en) | 1991-07-16 | 1992-07-16 | Programmable noise bandwidth reduction by means of digital averaging |
Country Status (6)
Country | Link |
---|---|
US (1) | US5216696A (en) |
EP (1) | EP0669058A1 (en) |
JP (1) | JPH06509219A (en) |
CA (1) | CA2102500C (en) |
MX (1) | MX9204204A (en) |
WO (1) | WO1993002506A1 (en) |
Families Citing this family (23)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5216696A (en) * | 1989-12-22 | 1993-06-01 | Comsat Laboratories | Programmable noise bandwidth reduction by means of digital averaging |
US5319679A (en) * | 1992-12-09 | 1994-06-07 | Datum Systems | Method and apparatus for recovering data from a radio signal |
US5485395A (en) * | 1994-02-14 | 1996-01-16 | Brigham Young University | Method for processing sampled data signals |
US6334219B1 (en) | 1994-09-26 | 2001-12-25 | Adc Telecommunications Inc. | Channel selection for a hybrid fiber coax network |
US5692014A (en) * | 1995-02-03 | 1997-11-25 | Trw Inc. | Subsampled carrier recovery for high data rate demodulators |
US7280564B1 (en) | 1995-02-06 | 2007-10-09 | Adc Telecommunications, Inc. | Synchronization techniques in multipoint-to-point communication using orthgonal frequency division multiplexing |
USRE42236E1 (en) | 1995-02-06 | 2011-03-22 | Adc Telecommunications, Inc. | Multiuse subcarriers in multipoint-to-point communication using orthogonal frequency division multiplexing |
US6026129A (en) * | 1996-03-27 | 2000-02-15 | Matsushita Electric Industrial Co., Ltd. | Radio receiving apparatus for receiving communication signals of different bandwidths |
US6628699B2 (en) | 1997-06-23 | 2003-09-30 | Schlumberger Resource Management Systems, Inc. | Receiving a spread spectrum signal |
US6178197B1 (en) | 1997-06-23 | 2001-01-23 | Cellnet Data Systems, Inc. | Frequency discrimination in a spread spectrum signal processing system |
US6741638B2 (en) | 1997-06-23 | 2004-05-25 | Schlumbergersema Inc. | Bandpass processing of a spread spectrum signal |
US6047016A (en) * | 1997-06-23 | 2000-04-04 | Cellnet Data Systems, Inc. | Processing a spread spectrum signal in a frequency adjustable system |
US6683919B1 (en) | 1999-06-16 | 2004-01-27 | National Semiconductor Corporation | Method and apparatus for noise bandwidth reduction in wireless communication signal reception |
US6801605B1 (en) * | 2000-09-13 | 2004-10-05 | 3Com Corporation | Telephone network signal conversion system |
US7113562B1 (en) * | 2000-12-27 | 2006-09-26 | Intel Corporation | Method and apparatus for receiving data based on tracking zero crossings |
US6674401B2 (en) * | 2002-02-19 | 2004-01-06 | Eride, Inc. | High sensitivity GPS receiver and reception |
EP1482672A1 (en) * | 2003-05-26 | 2004-12-01 | Delphi Technologies, Inc. | Sampling for quadrature demodulation |
US7053808B2 (en) * | 2003-11-26 | 2006-05-30 | Texas Instruments Incorporated | Suppressing digital-to-analog converter (DAC) error |
KR100644868B1 (en) | 2004-10-06 | 2006-11-15 | 광주과학기술원 | Method and Apparatus for Bandpass Sampling with Normal Placement for Multiple RF Signals |
US7283792B2 (en) * | 2004-10-15 | 2007-10-16 | Nokia Corporation | Method and apparatus for providing limiting power adjustment in a wireless communication system |
US8855254B2 (en) * | 2010-03-05 | 2014-10-07 | The Aerospace Corporation | Systems and methods for pre-averaged staggered convolution decimating filters |
US8489662B2 (en) * | 2010-03-05 | 2013-07-16 | The Aerospace Corporation | Systems and methods for sliding convolution interpolating filters |
US9680307B2 (en) * | 2012-12-21 | 2017-06-13 | General Electric Company | System and method for voltage regulation of a renewable energy plant |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4563637A (en) * | 1982-07-19 | 1986-01-07 | Cselt Centro Studi E Laboratori Telecomunicazioni S.P.A. | System for measuring amplitude of noise-contaminated periodic signal |
US4583184A (en) * | 1982-01-13 | 1986-04-15 | Terumo Kabushiki Kaisha | Ultrasonic image processor |
US4714929A (en) * | 1986-09-04 | 1987-12-22 | Davidson Eldon F | Digital averaging filter particularly suited for use with air navigation receivers |
US4727504A (en) * | 1984-07-05 | 1988-02-23 | The Charles Stark Draper Laboratory, Inc. | Interference canceller and signal quantizer |
Family Cites Families (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4958361A (en) * | 1988-04-22 | 1990-09-18 | Hughes Aircraft Company | Edge effect reduction by smoothing in digital receivers |
FR2645373A1 (en) * | 1989-03-28 | 1990-10-05 | Js Telecommunications | METHOD AND DEVICE FOR REDUCING NOISE ON A CODABLE SIGNAL AT SEVERAL PREDETERMINED LEVELS |
US5216696A (en) * | 1989-12-22 | 1993-06-01 | Comsat Laboratories | Programmable noise bandwidth reduction by means of digital averaging |
US5052027A (en) * | 1989-12-22 | 1991-09-24 | Comsat Laboratories | Programmable noise bandwidth reduction by means of digital averaging |
-
1991
- 1991-07-16 US US07/730,426 patent/US5216696A/en not_active Expired - Fee Related
-
1992
- 1992-07-16 EP EP92915452A patent/EP0669058A1/en not_active Withdrawn
- 1992-07-16 WO PCT/US1992/005849 patent/WO1993002506A1/en not_active Application Discontinuation
- 1992-07-16 JP JP5502890A patent/JPH06509219A/en active Pending
- 1992-07-16 CA CA002102500A patent/CA2102500C/en not_active Expired - Fee Related
- 1992-07-16 MX MX9204204A patent/MX9204204A/en unknown
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4583184A (en) * | 1982-01-13 | 1986-04-15 | Terumo Kabushiki Kaisha | Ultrasonic image processor |
US4563637A (en) * | 1982-07-19 | 1986-01-07 | Cselt Centro Studi E Laboratori Telecomunicazioni S.P.A. | System for measuring amplitude of noise-contaminated periodic signal |
US4727504A (en) * | 1984-07-05 | 1988-02-23 | The Charles Stark Draper Laboratory, Inc. | Interference canceller and signal quantizer |
US4714929A (en) * | 1986-09-04 | 1987-12-22 | Davidson Eldon F | Digital averaging filter particularly suited for use with air navigation receivers |
Also Published As
Publication number | Publication date |
---|---|
JPH06509219A (en) | 1994-10-13 |
US5216696A (en) | 1993-06-01 |
CA2102500A1 (en) | 1993-01-17 |
EP0669058A1 (en) | 1995-08-30 |
MX9204204A (en) | 1993-04-01 |
CA2102500C (en) | 1999-10-26 |
EP0669058A4 (en) | 1995-04-26 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US5216696A (en) | Programmable noise bandwidth reduction by means of digital averaging | |
US5052027A (en) | Programmable noise bandwidth reduction by means of digital averaging | |
KR0143116B1 (en) | Radio receiver for receiving both vsb and qam digital hdtv signals | |
USRE38456E1 (en) | Decimation of baseband DTV signals prior to channel equalization in digital television signal receivers | |
CA2181843C (en) | Demodulator, modulation and demodulation system, and demodulation method | |
JP3649735B2 (en) | Carrier independent timing recovery system for vestigial sideband modulation signal | |
US5172070A (en) | Apparatus for digitally demodulating a narrow band modulated signal | |
US6333767B1 (en) | Radio receivers for receiving both VSB and QAM digital television signals with carriers offset by 2.69 MHz | |
US5040192A (en) | Method and apparatus for optimally autocorrelating an FSK signal | |
CA2242294C (en) | Bandpass phase tracker with hilbert transformation before plural-phase analog-to-digital conversion | |
US6545728B1 (en) | Digital television receivers that digitize final I-F signals resulting from triple-conversion | |
US6493398B1 (en) | Burst mode digital communications system | |
US5479453A (en) | Wireless telecommunication digital receiver | |
US6512555B1 (en) | Radio receiver for vestigal-sideband amplitude-modulation digital television signals | |
RU2145776C1 (en) | Method and device for recovering clock synchronization of characters from common-mode and broad-band quadrature signals being received | |
KR0176643B1 (en) | Digital vsb detector with bandpass phase tracker using radear filters, as for use in an hdtv receiver | |
WO1997006604A1 (en) | Universal rf receiver | |
US6496229B1 (en) | TV receiver using read-only memory shared during VSB and QAM reception for synchrodyning I-F signal to baseband | |
EP1693954B1 (en) | Demodulation circuit for use in receiver using IF sampling scheme | |
KR100276772B1 (en) | Digital television signal receiver decimates baseband digital television signals before channel lighting | |
US10320365B2 (en) | Filter that minimizes in-band noise and maximizes detection sensitivity of exponentially-modulated signals | |
EP0265131B1 (en) | Simplified recovery of data signals from quadrature-related carrier signals | |
EP0643511B1 (en) | Synchronization circuit for subcarrier signal | |
US6546237B1 (en) | Differential FM detector for radio receivers | |
US6999499B1 (en) | Demodulation apparatus of a base station in a CDMA mobile communication system |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
AK | Designated states |
Kind code of ref document: A1 Designated state(s): CA JP KR |
|
AL | Designated countries for regional patents |
Kind code of ref document: A1 Designated state(s): AT BE CH DE DK ES FR GB GR IT LU MC NL SE |
|
DFPE | Request for preliminary examination filed prior to expiration of 19th month from priority date (pct application filed before 20040101) | ||
WWE | Wipo information: entry into national phase |
Ref document number: 2102500 Country of ref document: CA |
|
WWE | Wipo information: entry into national phase |
Ref document number: 1992915452 Country of ref document: EP |
|
WWP | Wipo information: published in national office |
Ref document number: 1992915452 Country of ref document: EP |
|
WWW | Wipo information: withdrawn in national office |
Ref document number: 1992915452 Country of ref document: EP |