|Publication number||USH2015 H1|
|Application number||US 08/414,824|
|Publication date||Mar 5, 2002|
|Filing date||Mar 31, 1995|
|Priority date||Mar 31, 1995|
|Publication number||08414824, 414824, US H2015 H1, US H2015H1, US-H1-H2015, USH2015 H1, USH2015H1|
|Inventors||William M. Waters|
|Original Assignee||The United States Of America As Represented By The Secretary Of The Navy|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (6), Non-Patent Citations (2), Referenced by (3), Classifications (4), Legal Events (1)|
|External Links: USPTO, USPTO Assignment, Espacenet|
1. Field of the Invention
This invention in general describes a communications system, specifically a communications system having sharply bandlimited waveforms for sampled data communications.
2. Description of the Related Art
Proliferation of communications, radar, and other systems making use of electromagnetic wave propagation has resulted in overcrowding in the useful portions of the electromagnetic spectrum. Even line-of-sight communications links seldom utilize frequencies above the X-band because of the absorption and scattering by rain and other constituents of the troposphere. Consequently, it becomes more important for transmitters to radiate signals with well confined spectra so limited spectral space is used more efficiently.
In order to communicate a sequence of sampled digital data, r(tn), over a band limited communications channel, the sequence of values must be transformed into an analog signal which is spectrally confined to the band available. Traditionally, confinement of radiated signals to assigned bands has been accomplished by frequency or phase modulation (FSK or PSK) followed by analog filtering operations at intermediate frequency and final amplifier stages of the transmitter. At the receiver, this signal is filtered from adjacent channels after appropriate down-conversion by utilizing analog filters to select the desired signal for subsequent conversion to digital form.
It is well known to those practicing the art that the spectrum of a signal formed from appropriate combinations of cardinal functions will have zero energy outside the designated frequency band width. For this to be exact, it is required that the cardinal functions be of infinite length. Theoretically, this causes no intersymbol interference since cardinal functions are zero at all sample points except one, t=0. Thus the analog/digital converter, appropriately synchronized and in the absence of noise, reads values of the transmitted sequence.
Functions of infinite length are not physically realizable. However, studies have shown that, by appropriately modifying the cardinal function, one can achieve excellent interpolation accuracies and resultant phase accuracies in direct sampling applications even if the function is truncated to just a few sample intervals.
In the field of electromagnetic transmission of digital data there is a need for an alternative to FSK and various forms of PSK signals which would improve the confinement of the signal spectrum to an assigned frequency band in transmitters and receivers. It is well known that bi-phase code modulation consists of a sequence of equal-length intervals, often called “chips”. Within each chip a sinusoid at the carrier frequency is bi-phase modulated; more specifically, the carrier phase within a chip is constant but may jump from chip-to-chip by 180 degrees in accordance with modulating binary data. A drawback to the application of this waveform is the 180° phase “jump” which causes spreading of its spectrum. However, after a bi-phase signal is modified by the process which is the subject of this invention, its spectrum is well-defined to the desired band.
Further, it is important that adjacent communication channel interference and mutual interference among radar, navigation, communications and other systems which depend upon electromagnetic radiation be prevented.
It is the object of this invention to transform a communications sequence in a communications system into an analog signal which is spectrally confined to the band available for the transmission within a band limited communications channel without compromising communication accuracy.
Another object of this invention is to improve spectral confinement thereby allowing closer channel spacing and better electromagnetic compatibility among radar, navigation, and communications systems.
These and other objects are achieved by a communication system using sharply bandlimited signals for sampled data communications wherein a sharply bandlimited waveform is computed for each sample value in a sequence of data samples. Each waveform is centered at a sample point, is weighted by the corresponding data value, and is truncated outside an appropriate time interval. Resulting data-weighted waveforms are summed in the computer for all data samples of the sequence to be communicated, and the result is converted to analog form, up-converted to the desired carrier frequency, and transmitted. Because the signal spectrum is sharply bandlimited, communication channels may be spaced more closely, permitting more channels in the same frequency band.
At the receiver, signal-plus-noise is down-converted, data is sampled and converted to digital form, de-multiplexed (if necessary), and delivered to corresponding users.
FIG. 1 shows the error for low-pass interpolation using cardinal functions and self-truncating cardinal functions for B/W=¾ and ½ (33% and 100% over-sampling, respectively) as a function of the truncation length as measured by N, the number of samples within the truncated interval.
FIG. 2a shows the output of the waveform generator, the sharply bandlimited keying waveform.
FIG. 2b shows the spectral confinement of the sharply bandlimited keying waveform.
FIG. 3 is a schematic of a transmission system utilizing a sharply bandlimited communications filter.
FIG. 4 depicts a test configuration for testing a sharply bandlimited communications system.
FIG. 5 shows (a) the measured spectrum of the output waveform of the waveform generator with interpolation and (b) the measured spectrum of the output waveform of the waveform generator without interpolation.
FIG. 6a shows the I component before processing.
FIG. 6b shows the Q component before processing.
FIG. 6c shows the I component after processing.
FIG. 6d shows the Q component after processing.
FIG. 6e shows the correlated magnitude (I2+Q2)½ of the I and Q components.
In order to communicate a sequence of sampled digital data over a bandlimited communications channel in a format other than the conventional analog format, a bi-phase code modulation utilizing a sharply bandlimited keying waveform may used.
A sharply bandlimited keying waveform is computed for each sample value in a sequence of data samples and centered at a sample point, weighted by a corresponding data value, and truncated outside an appropriate time interval. For a predetermined carrier frequency, fo, it is well known in the art that sampling theory requires that the signal must be sampled at a rate fs=4fo/(2M−1) which is greater than or equal to twice the signal bandwidth, where M is an integer. Defining the bandwidth to be 1/tc, where tc is the spacing between “chips” (elements of a phase code consisting of a sequence of equal-length intervals) or chip width, then the minimum sampling rate is fs=2/tc.
A signal with no spectral energy outside the band from −W to +W is completely specified by samples taken at uniform intervals less than or equal to ½ W. The original signal may be exactly reconstructed from these sampled values, r(tn), by the interpolation
where tn=n/2W, n being the number of the samples, and the interpolation function, or sharply bandlimited keying waveform, sometimes called an interpolation waveform or sampling function, is
A bandpass signal with no spectral energy outside the bands −MW to −(M−1)W and (M−1)W to MW may also be completely specified by samples taken at ½ W (or shorter) intervals. The original function may be reconstructed exactly from these sampled values by the interpolation defined by Eq.(1).
Eq.(1) holds exactly only if s(t) as defined by Eq. (2) for the bandpass case is defined for all values of t from −∞ to +∞. The interpolation error (magnitude of the difference between right and left sides of Eq. (1)) grows as s(t) is truncated to shorter lengths, and as the width of the spectrum of the sampled signal, B, approaches ½ the sampling rate fs+2W.
FIG. 1 shows the error for low-pass interpolation (Eqs. (1) and (2)) for B/W=¾ and ½ (33% and 100% over-sampling, respectively) as a function of the truncation length as measured by N, the number of samples within the truncated interval. FIG. 1 also shows how the interpolation error shrinks dramatically (for large N) as a self-truncating “taper” is applied to the interpolation function. This tapered function is given for the low-pass case where q=1−B/W and m depends on both q and N.
Since interpolation error is dramatically reduced using the “self-truncating” interpolation function, it is logically concluded that the spectrum of s(t) from Eq. (3) is better confined to the band −W to +W than of s(t) from Eq. (2). The same logic applies to the bandpass case since the taper of Eq. (3) (the bracketed factor raised to the power, m) may be used in the bandpass case with similar results.
The interpolation waveforms s(t) have the property that the zeros are at the right points so as not to cause intersymbol interference. An example of s(t) is plotted in FIG. 2a; the spectrum of s(t) is shown in FIG. 2b. Normally the keying waveform is truncated to a shorter time interval resulting in poor spectral confinement.
In the preferred embodiment, referring to FIG. 5, sampled digital data in a sequence of data samples 12, r(tn) input to the system 10 where computer 14 and waveform generator 16 generate a weighted interpolation waveform r(tn)s(t−tn) for each data sample 12. The sampled input data 12 can be voice, television signals, pagers, or any other signal sampled as a function of time.
The sequence of data-weighted waveforms is summed in a computer 14 for all data samples of the sequence to be communicated, and the result is converted to analog form by a digital-to-analog (D/A) converter (not shown), filtered in a low-pass filter 22, up-converted to the desired carrier frequency in up-converter 18, and the signal is transmitted by a conventional method for transmission of electromagnetic waves, a transmitter 24.
The computer 14 receives a data bit stream from the sampled input data 12 and performs three functions. First, the computer 14 controls a waveform generator 16, secondly it weighs each waveform with the data and, thirdly, it forms the sum of weighted waveforms in digital form. The waveform generator 16 takes the digital data bits form the output signal and converts it to analog form by the use of an internal digital-to-analog (D/A) converter (not shown). The mathematical mechanism by which a waveform generator 16 synthesizes the bandlimited signal is as described above.
The output of the waveform generator 16 (the sharply bandlimited keying waveform shown in FIG. 2a) is passed through an integral or external low-pass filter 22 which filters out any harmonics previously generated. The filtered output of the waveform generator 16 is applied to an up-converter where the signal is up-converted to the desired carrier frequency by way of intermediate frequency (IF) stage to the transmitter 24 for generation of a sampled data signal output as radio frequency (RF) energy. A number of data channels may be multiplexed consistent with the bandwidth, if desired, by the application of the filtered analog signal from the low-pass filter 22 to a multiplexer 21.
A number of data channels may be grouped as sub-bands and a digital-to-analog (D/A) converter (not shown) would convert the data to an analog form at a rate consistent with the group bandwidth. Final separation of data channels may then be accomplished by digital filtering.
At a receiving location, the transmitted signal, plus any induced noise, is preamplified by a low-noise amplifier 28, demultiplexed by a demultiplexer 28 and down-converted by a low-noise amplifier filter down-converter 32, data is sampled and converted to digital form by an analog-to-digital (A/D) converter 34 , separated into separate channels by a digital channel separation filter 36, and applied to an audio or digital display device 38.
In the formation and transmittal of the sharply bandlimited signals for sampled data communications, the waveform generator 16 receiving the sample data delivers accurate data to the receiver 28 unless the remainder of the transmitter 24 distorts the radiated waveform. The up-converter 18 and transmitter 24 power amplifier (not shown) must exhibit a flat amplitude and linear phase characteristic across the signal band. Since the signal is bandlimited, this is more easily accomplished by analog low-pass filters 22. The better spectral confinement achieved through the use of the device set forth above leads to closer channel spacing and better electromagnetic compatibility among radar, navigation, and communications channels.
Operational data interpolation computer 14 may be performed by a finite impulse response (FIR) filter (not shown) implemented by a digital signal processor (DSP) microchip, such as a Part no. DSP96002, manufactured by Motorola Corp. of Phoenix, Ariz. The preferred DSP should have a speed and digital precision to provide the same processing as computer 14, but requiring only a single microchip plus requisite drivers and power supplies. D/A conversion may also be performed by a much smaller unit than a waveform generator 16, such as a Model 1010-60-860, manufactured by KOR Industries of Garden Grove, Calif. digital-to-analog converter.
The better spectral confinement provided by this invention leads to closer channel spacing and better electromagnetic compatibility among radar, navigation, and communications systems.
FIG. 4 depicts a sharply bandlimited keying waveform test configuration 40 of the preferred embodiment under laboratory conditions to verify that the output data is identical to the input data prior to operational data link interpolation. See, Sharply Bandlimited Signals for Sampled Data Communications, NRL/MR/5330—94—7481, Jun 94, which is hereby incorporated by reference.
Directly sampled and digitized data 42 from a pseudo-random bi-phase coded sinusoid (not shown) are selected by a computer 43 and entered into a waveform generator 52 synthesizer memory module 44. These data are used to weight the interpolation function previously described incorporating the computer program set forth in Table 1. Table 1 specifies mathematically the mechanism by which the waveform generator synthesizes the bandlimited signal from a sequence of data signals. Samples 42 are taken from the sinusoid at intervals, ts, of 0.144 μs and have a width of biphase modulation chips, tc, equal to 0.288 μs. Weighted and summed interpolation functions are computed at 0.008 μs intervals corresponding to a 125 MHz rate of the first memory 44 of the waveform generator 52 and read into a fast digital-to-analog (D/A) converter 46.
Within the D/A converter 46, the digital values from first memory 44 are converted to analog form and low-pass filtered by a low-pass filter 48 (0 to 50 MHz) to form the bandlimited analog signal centered at fo output of the waveform generator 52, FIG. 3. The waveform generator 52 in the test installation functioned not only as a waveform generator but also as a D/A converter 46 and a low-pass filter 48. The spectrum, as shown in FIG. 5a, of the bandlimited analog signal of the waveform generator 52 is measured by a spectrum analyzer 54. FIG. 5b shows the spectrum of the same phase coded signal but without the weighting of Eq. (3).
To analyze the performance of the sharply bandlimited keying waveform system (ie., to determine fidelity with the digital sample 42 from the bi-phase phase modulated sinusoid), the I and Q components 66 and 64, respectively, of the output from the waveform generator 52 with the measured spectrum from the spectrum analyzer 54 are split out in an I/Q detector 56 when a reference signal 58 is heterodyned against the signal having the sharply bandlimited keying waveform in phase quadrature in order to obtain phase and amplitude information. Phase accuracy, where the phase, φ, is defined as Tan−1Q/I, is dependent upon careful matching of these separate quadrature channels. Accordingly, phase errors limit the performance achievable from subsequent digital processing.
The digital outputs of the A/D converters 66 and 68 are applied to a computer 74 where the I and Q components are correlated with the pseudo random code used to generate the original input data samples 42 in the correlator 76, and an envelope detector 78 computes the square root of the sum of the squares which is applied to a plotter 82 for display of the test data.
In the test configuration 40, the particular values of ts and tc were chosen to be integer multiples of the waveform generators 52 fast memory read interval of 0.008 μs. The cycle period of the output signal carrier frequency was also chosen to be an integer multiple of 0.008 μs (1/15.625=0.064 and 1/31.25=0.032). The results of this test are shown in FIG. 6a through 6 c. The curves in FIG. 6a and 6 b depicts I and Q components of the received signal; FIGS. 6c and 6 d show the I and Q components after correlator processing and FIG. 6e depicts the correlated magnitude (I2+Q2)½ from FIGS. 6c and 6 d.
It will be understood by those skilled in the art that still other variations and modifications are possible and can be affected without detracting from the scope of the invention as defined by the claims.
PROGRAM TO GENERATE A DIRECT SAMPLED MAXIMAL
LENGTH BIPHASE WAVEFORM WITH WEIGHTED CHIPS
FOR OUTPUT TO THE HP8770 WAVEFORM SYNTHESIZER
PRINTER IS CRT
! GENERATE MAXIMAL LENGTH SHIFT REGISTER
OPTION BASE 1
SEPARATE ALPHA FROM GRAPHICS
INTEGER X(8), Xs(255)
DIM Λa(25), La(25), Ra(25)
Ix=67 !INITIAL CONDITION
N=8 !NUMBER OF STAGES
FOR I=N TO 1 STEP −1 ! SET INITIAL CONDITION
IF Ic>=(Pn) THEN X(I)=1
IF Ic>=(Pn) THEN Ic=Ic−Pn
Pn=2“N−1 ! NUMBER OF ELEMENTS IN SEQUENCE
FOR I=1 TO Pn
Xn=(X(8)+X(4)+X(3)+X(2)) MOD 2! FEEDBACK
FOR K=1 TO N
PRINT USING ““(D),3X,DDD”;X(8),X(7),X(6),
FOR J=N TO 2 STEP −1
DISP “ENTER CHIP WIDTH IN MICROSECONDS”;
DISP “ENTER THE CARRIER FREQUENCY”;
DISP “ENTER THE SAMPLING RATE OF THE HP8770”;
DISP “ENTER NAME OF FILE WHERE DATA IS TO BE
DISP “ ENTER A COMMENT DESCRIBING WAVEFORM”;
Nc=T*W!NUMBER OF SAMPLES IN CHIP
Print “NUMBER OF SAMPLES IS ”;Ns
PRINT “LENGTH OF PULSE IS ”;Ns/W
PRINT Nz;“ ZEROS WILL BE ADDED TO MAKE SAMPLES A
MULTIPLE OF 8”
PRINT “ TOTAL PULSE WILL BE ”;Tp
PRINT “ +− 1 12-BIT WEIGHT”
! INTERPOLATION FUNCTION PARAMETERS
M1=13 ! FOR Λ .4 CHIP LENGTH !5 ! FOR Λ
.144 CHIP LENGTH
FOR T=0 TO (Ns/W) STEP 1/W ! STEP THROUGH THE
! TAPER START AND END
IF Chip<=5 THEN Chs=5
IF Chip>=Pn−4 THEN Che=Pn−4
FOR N=Chs TO Che
FOR M−1 TO 2
IF Y=0 THEN
! LIMIT AMPLITUDE TO 1 V PEAK
IF Mag>=1 THEN V=V/Mag*.99
1040!PRINT USING “DDDDD,4X,DD.DDDD,4X,DDDDDD,4X,
PRINT Nz;“ ZEROS WILL BE ADDED TO MAKE SAMPLES A
MULTIPLE of 8”
FOR I=1 TO Nz
PRINT “TOTAL SAMPLES=”;Ic
CREATE BDAT Nam$,1,L+200
ASSIGN @Wf TO Nam$
ASSIGN @Wf TO *
PRINT “=****** DONE *******”
PROGRAM TO READ WAVEFORMS AND OUTPUT IN AN
CONVERTED TO SERIES 300 3/30/88 MOD 4/14/88
PRINTER IS CRT
PRINT “ PROGRAM TO SEND WAVEFORMS TO
PRINT “ AS A MINIMUM, FUNCTION KEYS 1 AND 2
NEED TO BE USED
PRINT “ 1 TO LOAD WAVEFORMS FROM DISC AND
STORE IN THE 8770 MEMORY”
PRINT “ 2 TO SET UP THE SEQUENCING OF THE
PRINT “A MODE EXISTS VIA KEY 5 TO CHANGE TIME
DELAY BY ENTRY OF TIME IN MICROSECONDS”
ON KEY 1 LABEL “LOAD WAV” GOSUB 1480
ON KEY 2 LABEL “SEQ SET” GOSUB 1820
ON KEY 3 LABEL “SEQ CHG” GOSUB2180
ON KEY 4 LABEL “STATUS” GOSUB 2030
ON KEY 5 LABEL “DEL TIME” GOSUB 2250
ON KEY 6 LABEL “CLOCK” GOSUB 2570
ON KEY 7 LABEL “CAT” GOSUB 2690
ON KEY 8 LABEL “RESTART” GOTO 1280
KEY LABELS ON
OUTPUT 719;“PURGE BOTH”
OUTPUT 719;“FORMAT SIGN”
GOSUB 2570 !SELECT CLOCK SOURCE
S$=VAL$(A)! START FIRST WAVEFORM SEGMENT AND
PACKET AT ZERO
DISP “ENTER THE NUMBER OF WAVEFORM FILES”;
FOR I=1 TO J
DISP “ENTER FILE NAME”;
PRINT “ WAVEFORM MEMORY CONTENTS”
PRINT “ NAME LENGTH IN SAMPLES ADDRESS
PULSE W at CLOCK BANDWIDTH START FREQ”
FOR I=1 TO J
ASSIGN @W1 TO F$
ENTER @W1,1;L,W$,Cdir,Fs,B,T,W,Comm$! READ
NUMBER OF BYTES IN, AND DATA
N$=VAL$(Nn(I))! NUMBER OF WAVE SEGMENTS
PRINT USING “10A,4X,DDDDD,11X,DDDDD,6X,DDD.D,
4A,DDD,5X,DDD.DDD,6X,DD.DDD”;F$,Nn(I),Λ,T,” at “,W,B,Fs
OUTPUT 719;“WMEM “&S$&”,#Λ“z&U$&W$|1,L|
S$=VAL$(Λ) NEXT START ADDRESS
DISP “ENTER NUMBER OF WAVEFORM SEGMENTS”;
FOR I=Y TO J1
DISP “ENTER SEGMENT STARTING ADDRESS”;
DISP “ENTER LENGTH OF SEGMENT”;
DISP “ENTER NUMBER OF TIMES TO REPEAT
OUTPUT 719;“PURGE SEQ”
FOR I=1 TO J2
OUTPUT 719;“PACLIT “&S$&”,“&N$&”,“&R$&”, AUTO”
PRINT “ SEQUENCE STATUS”
PRINT “ADDRESS LENGTH REPEATS TIME”
FOR I=1 TO J2
PRINT USING “DDDDD,3X,DDDDD,5X,DDDDD,4X,
PRINT “ WAVEFORM MEMORY CONTENTS”
PRINT “ NAME LENGTH IN SAMPLES ADDRESS′
FOR I=1 to J
DISP “TO CHANGE ONE SEGMENT ENTER SEGMENT
GENERATING WAVEFORM SEQUENCES OF DESIRED
PRINT “AT START THERE MUST BE ONLY ONE ZERO
SEGMENT IN SEQUENCE”
DISP “ENTER ADDRESS OF ZERO WAVEFORM SEGMENT”;
DIPS “ENTER SEQUENCE POSITION OF ZEROS”;
FOR I=J2 TO I2+2 STEP −1
DISP “ENTER DESIRED TIME”;
IF T<6.1 THEN T=6.1
L=INT(F/8+.5)*8! SEQUENCE MUST BE MULTIPLE OF 8
R1=INT((T−3)/L*F)! REPETITIONS OF SEQUENCE L
DISP “TO EXIT, ENTER AN E -- TO CONTINUE,
IF E$=“E” THEN RETURN
ROUTINE TO MODIFY CLOCK SELECTION
DISP “enter an EXT for external clock
otherwise the internal 125MHz is assumed”;
IF Cls$=“EXT”THEN 2650
PRINT “INTERNAL CLOCK SELECTED FREQUENCY
OUTPUT 719;“CLKSEL INT”
PRINT “EXTERNAL CLOCK SELECTED FREQUENCY
OUTPUT 719;“CLKSEL EXT”
GET A CATALOG OF DISC
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US4227250||Nov 9, 1978||Oct 7, 1980||Bell Telephone Laboratories, Incorporated||Minimization of excess bandwidth in pulse amplitude modulated data transmission|
|US4507794||May 4, 1983||Mar 26, 1985||Jones Stephen K||Filtered Manchester Coded PSK transmission system|
|US4550415||Nov 16, 1983||Oct 29, 1985||At&T Bell Laboratories||Fractionally spaced equalization using nyquist-rate coefficient updating|
|US4601045||Aug 3, 1984||Jul 15, 1986||Larse Corporation||Modulator-demodulator method and apparatus with efficient bandwidth utilization|
|US4782324||May 6, 1987||Nov 1, 1988||Genrad, Inc.||Digital signal synthesizer|
|US5063574||Mar 6, 1990||Nov 5, 1991||Moose Paul H||Multi-frequency differentially encoded digital communication for high data rate transmission through unequalized channels|
|1||Waters et al., "Phase Accuracy Experiments With a Direct-Sampling Coherent Detector," NRL Report 9182, Mar. 10, 1989.|
|2||Waters et al., "Sharply Bandlimited Signals for Sampled Data Communications," NRL Report NRL/MR/5330-94-7481, Jun. 1994.|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US8805301 *||Dec 9, 2011||Aug 12, 2014||Bae Systems Information And Electronic Systems Integration Inc.||Drop zone rally point beacon|
|US20070186678 *||Jan 23, 2007||Aug 16, 2007||Karl Griessbaum||Paired ZF Sampling for Pulse Running Time Filling Level Sensor|
|US20130149960 *||Dec 9, 2011||Jun 13, 2013||Bae Systems Information & Electronic Systems Integration Inc.||Drop zone rally point beacon|
|Jun 2, 1995||AS||Assignment|
Owner name: NAVY, UNITED STATES OF AMERICA, THE, AS REPRESENT
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:WATERS, WILLIAM M.;REEL/FRAME:007504/0870
Effective date: 19950330