MXPA97002410A - Transmission system for audio diffusion digi - Google Patents

Abstract

A digital audio dysfunction system includes an RF transmitter and a corresponding RF receiver. In the RF transmitter, a digitally compressed audio signal is encoded in a symbol stream that is first rotated using a frequency of 150,000 hertz (hz) before transmission to the receiver

Description

TRANSMISSION SYSTEM FOR AUDIO DIFFUSION DIG ^ T ^. CROSS REFERENCE TO RELATED REQUESTS Related subject matter is described in 5 commonly-assigned US patent applications co-pending from Wang and Langberg with the title "A Transmission System for Digital Audio Broadcasting1" (a transmission system for digital audio broadcasting) numbers US series 08/628117, 08/628120, 08/628220, 08/628118 and 08/628119, respectively BACKGROUND OF THE INVENTION The present invention relates to communication systems, and, more particularly to audio broadcasting. Significant advances in source coding make it possible to compress stereo sound by an approximate factor of 10, with no notable loss in quality after decompression.An application that can benefit from this breakthrough is diffusion. quality improvement over AM radio In the last 10 years, many researchers have felt that another stage to further improve the quality of sound transmission. This has resulted in the effort now known as digital audio broadcasting (DAB) or digital audio radio (DAR). However, while it was a significant achievement for source coding researchers, the ability to compress stereo sound of approximately 1.4 REF: 24323 megs / bits / second (Mb / s) at 160 Kilobits / second (Kb / s), it is not a simple task for data communications researchers to design a reliable wireless digital data link in a highly mobile environment as it is presented, as for example by a radio receiver in a moving vehicle. This is because the delivery of a communication system such as DAB is complicated by the fact that the communications channel is highly variant with time and is severely distorted by the effects of multiple paths and Doppler shift. As such, the objective error ratio and the service interruption rate is much more rigid than an application of digital cellular stereo. CQMPEHPSP PE THE INVENTION In accordance with the principles of the invention, a radio frequency (RF) transmitter includes a rotator or rotary positioner, to a low intermediate frequency (IF) signal, which is subsequently modulated for transmission to an RF carrier frequency . The use of a rotator simplifies the design of the receiver by removing ambiguity of phase in a received signal and also allows the use in the receiver of a low-order digital carrier phase recovery circuit, which provides the ability to quickly acquire the RF signal received. This is especially important when the receiver is located in a moving vehicle and is subject to the aforementioned Doppler effects. In one embodiment of the invention, a DAB system includes an RF transmitter and a corresponding RF receiver. In the RF transmitter, a digitally compressed audio signal is encoded in a symbol stream that is then rotated using a frequency of 150,000 hertz (hz) before transmission to the RF receiver. BRIEF DESCRIPTION OF THE DRAWING Figure 1 shows a high-level block diagram, illustrative of a digital audio broadcast communication system incorporating the principles of the invention; Figure 2 shows a more detailed block diagram of a portion of the transmitter of Figure 1; Figure 3 shows a constellation of illustrative signal points for use in the transmitter 100 of Figure 1; Figure 4 shows an illustrative chart format for use in the transmitter 100 of Figure 1; Figure 5 shows a block diagram illustrative of a position of the receiver 300 of Figure 1; Figure 6 shows an illustrative frequency spectrum for a low IF signal generated within the receiver 300 of Figure 1; Figure 7 shows an illustrative block diagram of a phase locked loop circuit, responding to flat fades and frequency fades; Figure 8 shows an illustrative graph of correlator output signal 526; Figure 9 illustrates peak, ignore and silent zones in the correlator output signal of Figure 8; Figure 10 illustrates the concepts of a peak coupling pattern and a silent coupling pattern; Figures 11 12 and 13 show an illustrative synchronization method for use in the receiver 300; Figure 14 shows an illustrative method for calculating compensating coefficients for use in the receiver 300; Figure 15 illustrates the "FFTM threshold formation used to calculate compensating coefficients for use at receiver 300. Figure 16 shows another illustrative method for calculating compensating coefficients for use at receiver 300; and Figure 17 shows an illustrative block diagram. of the symbol recovery element 705 for use in the receiver 300 of Figure 1. nKffRTPCTQN PKT? T-T-? P *. Figure 1 shows a high-level block diagram of a DAB communication system 10 incorporating the principles of the invention. The DAB communication systems 10 comprise a transmitter 100, communications channel 200 and receiver 300. Before describing the details of the inventive concept, a general review of the operation of the DAB communication system will be given 10. Also, the perceptual audio coding It is well known and will not be described in detail. For example see patent of the U.S.A. No. 5,285,498, titled "Method and Apparatus for Coding Audio Signals Based on Perceptual Model" (Method and apparatus for coding audio signals based on perceptual model) granted on February 8, 1994 to Johnson. Other coding techniques are described for example by J.P. Princen and A.B. Bradley, "Analysis / Synthesis Filter Bank Design Based on Time Domain Aliasing Cancellation" (Analysis / synthesis of a filter bank design based on the cancellation of toothing in time domain) IEEE Trans. ASSP, Vol. 34, No. 5, October, 1986.); E.F. Schroeder and J.J. Platte, "MSC: Stereo Audio coding with CD-Quality and 256 kBIT / SEC," (MSCV: audio coding in stereo with CD quality and 256kBIT / SEGM (IEEE Trans. On Consumer Electronics, Vol. CE-33, No. 4, November 1987), Johnston, "Transform Coding of Audio Signals Using Noise Criteria" (Codification of transformation of audio signals using interference criteria), IEEE JSCA, Vol. 6, No. 2, February 1988; US No. 5,341,457, entitled "Perceptual Coding of Audio Signis" (Perceptual coding of audio signals) granted to Hall and collaborators on August 23, 1994. In Figure 1, an analog audio signal 101 is fed to the preprocessor 105 where it is sampled (typically 48 Khz) and converted into a signal with digital pulse code modulation (PCM) (106 typically 16 bits) in standard form.The PCM 106 signal is fed into a perceptual audio encoder (PAC) 110, which compresses the PCM signal and sends out the compressed PAC signal 111. The latter represents a bit stream of 170 kb / s, of which 10 kb / s represents a control channel for auxiliary data and 160 kb / s represents the compressed audio signal. The compressed PAC signal 111 is applied to the error protection encoder 115, which applies a Reed-Solomon code to provide 100% redundancy to the compressed PAC signal 111. The encoder for error protection 115 is also considered to include a buffer and an interleaver (not shown) to additionally combat the effects of communication channel 200. The result is encoded signal 116, which represents an interleaved data stream of 340 kb / s and where the interleaver block contains 320 ms data (1088 kbits) ). The encoded signal 116 is applied to the modulator 150 which, as described further below, develops a signal according to the principles of the invention for transmission on the communications channel 200. Of the communications channel 200, the demodulator 350 of the receiver 300, retrieves a coded signal 351 according to the principles of the invention (described below). The encoded signal 351 is fed to the error protection decoder 315, which operates in a form complementary to the error protection encoder 115, to provide the compressed PAC signal 316 to the perceptual audio decoder 310. The latter decompresses the compressed PAC signal and sends out a PCM signal 311. This signal is fed to the post-processor 305, which creates an analog representation that is, ideally, identical to the analog audio signal 101. Turning now to Figure 2, a block diagram of the modular 150 is illustrated. The decoded signal 116 is applied to the multiplexer (MUX) 155, which aggregates the coded signal 116 with synchronization data (sync) 196 to develop the aggregate data signal 156. The sync data 196 comprises a data stream of 20 kb / s (10 ksymbols / second) representing synchronization and compensation information (described below) generated by the processor 195. The latter is representative it is not self-representative of a digital signal processor. (It will be noted that although the invention is illustrated here as being implemented with discrete functional building blocks, for example mapper 4-PSK 160, etc., the functions of any one or more of those building blocks can be carried out using one or more programmed processors, as represented in processor 195). The aggregate data signal 156 represents a data stream of 360 kb / s that is formatted within a sequence of frames, wherein each frame is 10 milliseconds (ms) in width and divided into two portions. A head portion and a portion of encoded data. In each 10 ms period, the spindle portion represents 200 bits (100 symbols) of synchronization data, while the encoded data portion represents 3400 bits of coded signal 116. The aggregated data signal 156 is applied to a spider mapper. 4-phase offset encryption (PSK) 160 that maps two bits at a time into a two-dimensional complex symbol that has phase and quadrature complements 161 to 162, respectively. Each symbol can be represented equivalently by (a (n) + jb (n).) A constellation of illustrative signal points is illustrated in Figure 3. As can be seen in Figure 3, the constellation of signal points comprises 4"data symbols", two "sound symbols and channel sync" and two "sync symbols of intercalator (symbol clock)." It can be seen from the constellation of signals in Figure 3, that the "sync sound symbols of channel "and the" sync symbols of intercalator (symbol clock) ", are one-dimensional symbols that only obtain values in phase The output signal resulting from the mapper 4-PSK 160 is a sequence of frames, each frame comprises 1800 complete symbols per Each frame An illustrative frame 205 is illustrated in Figure 4. As described above, frame 205 is 10 milliseconds (ms) in width and is divided into two portions: a header portion 206 comprising 100 complex symbols that are restrict pa to be one-dimensional and a portion of encoded data 207 comprising 1700 complex symbols. For the coded data portion of the aggregate data stream, the 4 symbols drawn from the constellation of signal points of Figure 3 are used. The mapping of data symbols is done in accordance with the following rules, wherein each claudátor contains two elements in pair, the input bits and their associated symbol specified by the relative quadrature and in-phase signal strengths in the constellation illustrative of Figure 3: ((0,0), (-1, -1).}., { (0,1), (-1,1).}., { (1,0), (1, -1).}., { (1,1), (1,1)) > With respect to the head portion of each frame, the 100 complex symbols represent a synchronization signal. These 100 complex symbols are intentionally designed as uni-dimensional and have only values in the in-phase component as illustrated in the constellation of points and signals of Figure 3. Among the 100 symbols of synchronization of the head 206, there are 86 one-dimensional symbols used for assist in frame synchronization at receiver 300. These 86 one-dimensional symbols are chosen from the "channel sync sound symbols" of the constellation of signal points of Figure 3 and comprise two sequences "of 31 pseudo-number symbols". random "(31PN) followed by a sequence of 24PN a sequence segment of 1PN). The generation of a sequence of pseudo-random numbers is known in the art. These 86 symbols are also employed in the receiver 300 for channel sound and compensation purposes, including multi-path correction, synchronization phase recovery and carrier phase recovery (described below). The remaining 14 one-dimensional symbols of the head 206 are used for either interleaver synchronization or symbol clock alignment on the receiver 300. These 14 one-dimensional symbols are chosen from the "sync interleave symbols (symbol clock)" of the constellation of signal points of Figure 3. In particular, the 14 symbols are used by the receiver 300 to indicate the start of each interleaver block of 320 ms. This interleaver synchronization is repeated every 32 frames and comprises 2 consecutive sequences of 7PN, as illustrated in Figure 4. In any event, the 14 symbols are always used for data symbol search synchronization in the associated frame. When they are also used to provide interleaver synchronization, the 14 symbols comprise two positive 7PNs. Otherwise, the 14 symbols comprise a positive 7PN followed by a negative 7PN as illustrated in Figure 4. This is required to align the coded data portion of each frame, as long as there is a change of phase time in sample time. Significant or relative intensity of different changes of signal paths and causes a change in the delay seen by the receiver 300. (It should be noted that in this context, a negative PN sequence is simply the opposite of a positive PN sequence. , if a 2PN is represented by the symbol (1,414.0; -1,414.0) illustrated in Figure 3, the corresponding negative 2PN is the sequence of symbols (-1,414.0; 1,414.0) .. Returning to Figure 2, the phase and quadrature components 161 and 162 are applied to the rotator 165, which rotates the phase of each symbol, In particular, each rotated symbol is expressed as follows: A-in a '?) + Jb'ifl (1) where? d = 2pxl50,000 is the rotation frequency, and n represents the time index of the symbol instances spaced by T. The output symbol stream of the rotator 165 is sampled by the sampler speed expander 170 to three. times the symbol speed of 1 / T. The sampling speed of the rotated symbols is expanded by inserting two zero value samples between all the symbols. As a result, the complex rotated symbols are expanded by a factor of and the sampled rate expander 170 provides an expanded complex symbol stream defined as Au (m): A " { M A ' { K / L. For * = 0- ± I. ± 2L ± 3 £ and 0 otherwise, (2) where, m is the time of samples spaced by 3"and L => (T / T ') = 3 is the value of over-sampled.Expanded complex samples are then spectrally shaped by the digital baseband filter 175 which includes the identical band base filter 175-1 and the quadrature base band filter 175-2 (although in theory, the rotated quadrature and phase symbols can be converted to analog signals or filtered by analog filters, it is much more difficult to control the analog filter specification than its digital counterpart). The digital baseband filter 175 has a transfer function defined as h * (m) which only has real values. For purposes of illustration, the digital baseband filter 175 has an excessive 10% bandwidth, using a 66-tap finite impulse response (FIR) filter (with extension of 22 symbols). It should be noted that a digital baseband filter physical equipment implementation 175 can take advantage of the fact that in the expanded samples there are two zero value samples in each symbol range. As a result, this filtering operation can be seen to have three superfilters (not shown), h (3r), h = (3r-l), h = (3r-2), with the same set of symbols, A '(r ), in its feeding where r is a time index to the symbol interval and r = k / L. Each of these three sub-filters individually produces a sample in a cyclic fashion, giving three output samples in each symbol interval indexed by r. Although not necessary for the inventive concept, the use of this sub-filter structure reduces the computational complexity by a factor, as compared to a brute-force implementation of the digital base filter 175. The complex output of digital base filter 175 is: yim) '¿¿¿(mA)? "(k) .or Ca) To the output samples, and () of the digital baseband filter 175, a pilot signal is added, which is digitally generated by the processor 195. The pilot signal is a complex signal expressed as /, r r r which is also three times the symbol rate when adding a sample version of a cosine waveform of 100 Khz to the filtered phase samples and that of a sinusoidal waveform to the filtered quadrature samples. The additional energy that the pilot tone adds to the signal that is provided by the digital baseband filter 175 is approximately 0.3 dB. The complex value signal that is provided by adders 177 and 182 is: Am im) +? E, t '> mT ', (4) where K determines the pilot energy. Alternatively, the in-phase component and the quadrature component, the real and imaginary parts of the above equation can be expressed as: ?, () = U- (rfcos (? DrT) + bir) sin (? ArO \ hh [m - rti +? Cos (? / ND, (5a) yes m) = Mrkos (?? rT) - u (r] sin (,? arT bim - rti +? - en (? p / nr). (5b) wherein equation (9a) is representative of signal 178 and equation (9b) is representative of signal 183. The signal e (m) is then applied to the digital-analog filter (D / A) 185, which comprises the D-A filter in-phase 185-1 and the D / A phase filter in quadrature 185-2. Each D / A filter is considered to comprise a "sample-and-rβten" circuit (not shown) followed by an analog low-pass filter (not shown). The analog signal resulting from the sample-and-catch circuits is denoted as: where H (t) is a rectangular pulse is introduced by the sampling-retention circuits and the D / A filter 185 is defined as follows: and O otherwise (7) Its Fourier transform is a sync function expressed as: Now, it can be seen that the use of a sampled envelope velocity greater than 1 / r 'reduces the spectral shaping effect due to the sampled-and-retention circuits. It also increases the separation between teeth. The output signal of the D / A converter, with toothed repetition in each 1 / 'is then separated by the analog low-pass filter in the filter sample D / A 185. Usually, when designing an analog filter, a distortion of significant envelope delay in the transition region from narrow band to band-stop. However, here the toothing is separated by a large frequency range, such that the analog filter stop band can move from the critical signal spectrum and therefore will not cause significant distortion in the signal. The signal is (t) then converted in an ascending manner to an IF signal, for example 10.7 MHz, followed by further conversion to an RF signal by the RF transmitter 190 as is known in the art. (In the context of this invention, it is considered that the RF signal is within one of the pre-designated frequency channels associated with the FM radio, in addition, we can also use other frequencies that can be assigned for this service). The transmitted signal, at this point can be expressed as:? s (/) - Rd \? e - '* "thh (t - nt) +? e) a' 'gj» (9) where d-, = 2 »r (100,000) and dj = 2tr (150,000) and?,. is the RF carrier frequency. Before going to the receiver section, the reason behind the rotation of symbols in the transmitter will be explained.

As illustrated in (9), the symbol A "is rotated by e -J *, t" T in the transmitter. Considering that channel and modulation are ideal, this low IF signal can be expressed using the previous equation (9) when adjusting? A = dd. As described further below, the circuits in the receiver 300 then produce a low IF signal centered at 150 Khz, dd instead of resulting in a more typical baseband signal. This, excluding the pilot, results in: r (=) where pf j is a channel of pass bands and If a pair of Hilbert filters is used within the receiver 300, they produce an analytical signal: Considering that hb (t) satisfies the Nyquist criterion, z (t), they can be sampled at neighboring intervals to recover the baseband symbols A »> This can be understood from the basic sample theory that illustrates that the spectrum received after the sample is repeated every 1 / T. Therefore, the sample process reconstructs the original "A". This is illustrated in the following equations. t) =? + jb ") h b (t-nT) (cos? t-nT) + sin? t-nT)) (12) n We use the in-phase component as an example.

"« Or thing. / L-nT) -bJ? B (t-nT) sena Jt-nT) (1 a) a t = kT (13b) Re [z (kT)] =? Í- ((* "iffííw» ¿(kn) T) - bjth f (kn) T) sin? J ((kn) T) (14a) μjt l? L-siil? Re ^ kT)] - «t (14b) as hb ((k-n) T) cos? (* - «> 7.}. Hh (fk-n) T) sin? J (k-n) T) ** 0P * ta all k-n (l5b) However, if s (t) is not rotated by e in the transmitter, the result in (14b) and its associated quadrature component would be rotated by and therefore, a rotator would be required in the receiver to compensate for this rotation. In addition to this, this receiver rotator will require compensating for any phase shifts due to Doppler effects, etc. While simple in concept, this leads to a complex implementation of the receiver rotator since it must follow both changes in phase at Δ2rJr Hz and the phase shifts mentioned above. For example, a larger step size may be required in the concurrent control loop to track these phase changes. As a result, the acquisition time and the resulting interference components increase. Therefore and in accordance with the invention, the rotator is placed in the transmitter, as illustrated in Figure 2, to simplify the implementation of the receiver. In particular, the use of rotator 165 removes phase ambiguity simplifies tracking at receiver 300 of the received RF signal. The tracking is simplified since a low-order digital carrier phase recovery circuit can now be employed, for example first order, in the receiver to quickly track any residual phase / frequency changes in the signal received after the RF section . This ability to quickly acquire the received RF signal is especially important when the receiver is located in a moving vehicle and is subject to the aforementioned Doppler effects. More importantly, by using rotation in the transmitter, a simplified coherent carrier frequency offset (or phase) is made possible without major restrictions in the system design, such as the selection of the data block size. As a result of the rotation of the transmitter, the modulation that is provided by the modulator 150 herein is referred to as a QPSK modulation without a carrier as opposed to nominal QPSK modulation. It should be noted that one problem with the baseband approach described above is that the upward conversion of radio requires cosine and sine modulators. It is not trivial to keep the two analog radio modulators exactly 90s apart. If the two modulators do not precisely align in phase, the two signals do not form a perfect Hilbert pair (as is known in the art) or complex analytical function. There are other approaches such as digital passband implementation to avoid this problem. In the bandpass approach, the baseband digital filters are replaced by quadrature and in-phase bandpass filters that form a Hilbert pair. The phased bandpass filter output is subtracted from the quadrature filter output. The resulting signal is then modulated at an intermediate frequency, wherein a selective analog bandpass filter is used to reject the associated image. The filtered bandpass signal is then converted to the radio frequency for air broadcast. This scheme simplifies the need to use two well-balanced mixers (spaced 90ß) at the cost of a bandpass filter with more selective image rejection. (The deviation of phase difference of the two 90a mixers results in performance degradation). To relax the selectivity requirement of this bandpass filter with image rejection, the center frequency of the quadrature / in-phase digital passband filters can be adjusted to a higher frequency which implies the use of a higher sample rate L. Finally, the rotation frequency of the phase rotator before the digital filters should be properly selected, so that the passband and baseband implementations are equivalent. If the center frequency of the passband filters is set to 150 + 180N + X kHz, the rotation frequency should be set to -X kHz (where N is an integer >; 0 and 180 kHz is the symbol speed). In total the baseband approach described above may result in a small performance penalty, given the implementation of the receiver 300 which will now be described. Also, it should be noted that the pilot in the case of passband must be offset from the center frequency of the passband.

At the receiver 300, a received RF signal is applied to the demodulator 350, which is illustrated in block diagram form in Figure 5. The demodulator 350 comprises the RF 505 down converter, which is an "IF" filter that converts into a descending the RF signal received at an IF frequency, for example 10.7 MHz, as is known in the art. The resulting IF output signal is then applied to the down converter IF 510, which provides a low IF signal of step band 511, centered at 150 kHz and which includes the pilot signal described above at 250 kHz, which is used in the receiver 300 as a reference for carrier recovery and as a source for generating other clock signals for the receiver (described below). An illustrative spectrum of the magnitude of the low band IF signal of step 511 is illustrated in Figure 6. Before proceeding with a discussion of the remainder of the demodulator 350, it should be noted that to compensate for any difference in carrier frequency between the RF signal transmitted and the received RF signal, an analog bearer phase-locked loop (PLL) is usually included within the RF 505 down converter. (Although the analog carrier PLL can be implemented in other parts of the demodulator 350, it is better to implement this function in the RF). However, in this type of mobile environment, the received RF signal may be subject to both selective frequency fading and amplitude fading (here referred to as "flat fading"). Therefore the PLL circuits of the RF 505 down converter are modified as illustrated in Figure 7 to be responsive to both a flat fading and a selective frequency fading. In particular, a received RF signal is applied to the mixer 605 which also receives a local oscillator signal (LO signal) 631. It is considered for simplicity that the mixer 605 includes all the circuits required to provide a recovered IF signal 606 illustratively at 10.7 MHz as it is known in the specialty. This recovered IF signal is applied to the automatic gain control 610, which is used to adjust the amplitude of the recovered IF signal, to provide the aforementioned IF output signal 506. Analog PLL 630 is a phase locked loop and it is considered that includes a crystal for generating the required LO signal 631. As is known in the art, the analog PLL 630 adjusts the phase of the LO signal in response to an IF signal, here represented by the IF 506 output signal. However, the Analog PLL 630 is operated in either tracking mode or an interlocking mode depending on the state of the "hold / proceed" signal 636. If the latter is representative of a logical "ONE" (ONE), the analog PLL 630 stops the tracking and interlocking phase of the LO 631 signal.

On the other hand, if the "catch / proceed" signal 636 is representative of a logical "ZERO", then the analog PLL 630 continues to adjust the phase of the LO 631 tape. In other words, the feedback loop of the analog PLL 630 is they keep open to avoid bad adjustment when there is a severe flat fading or a frequency selective fading at the pilot frequency. In particular, a flat fading of the received RF signal is detected by a flat fading detector 615 which compares an output signal of the AGC 610 which is representative of the fitness of the recovered IF signal 606. When the amplitude of the IF signal after recovery 606 is less than a predetermined value such as -110 dBm, plane fading detector 615 applies a logical "ONE" to gate 635. The latter applies a logical "ONE" signal to analogue PLL 630 which inhibits tracking. Similarly, the pilot fading detector 620 is used to detect a selective frequency fading around 250 Khz. The aforementioned low level IF signal 511 is applied to the narrow band filter 625 which is centered at the pilot signal frequency, for example 250 Khz. The narrow band filter 625 provides the recovered pilot signal 626 to the pilot fading detector 620. The latter compares the recovered pilot signal 626 against a reference thres. Whenever the pilot fading detector 620 detects a recovered pilot signal 626, it is considered that there is no frequency fading and the fading detector 620 provides a logical "ZERO" to the OR gate 635. However, when the fading detector of pilot 620 does not detect the recovered pilot signal 626, pilot fading detector 620 applies a logical "ONE" to gate O 635, which then inhibits analogue PLL 630 from adjusting the phase of signal LO 631. It will be noted that the recovered pilot signal may be generated by other methods, for example by using the channel characterization or compensator assembly described below. However, if other approaches are taken, those in the art should be aware of any significant processing delays (time) in general the recovered pilot signal. It is also possible to make the flat fading detector and the fading detector generate their outputs according to the quality of the pilot instead of logical "one" or "zero" for example, the flat fading detector can produce an output that is proportional to the difference between the received signal and a predetermined value such as -110 dBm, to adjust the PLL loop bandwidth. In the extreme case, the PLL is in the "" state by reducing the loop bandwidth to zero (opening the loop). However, another alternative could be to use different weighting coefficients as a function of the feed signal level instead of just a one and zero binary. A decision to either "" or "proceed" is based on comparing the thres value to a value calculated instantaneously or cumulatively over a predefined time interval using the weighting coefficients. Returning to Figure 5, as described above, the down converter IF 510 produces a low IF signal 511 centered at 150 Khz, Δd, instead of giving a more typical baseband signal. As noted above, Figure 6 shows an illustrative frequency spectrum for low IF signal 511, which occupies the frequency range from 50 Khz to 250 Khz. The generation of the low IF signal 511 as opposed to a more typical baseband signal is chosen in recognition of the fact that any interference rejection of the upper adjacent RF channel can be further enhanced by a digital low pass analog filter, while any Lower adjacent RF channel interference should be reduced only by an IF filter. However, a highly selective filter is more difficult to design at IF frequencies. Undoubtedly, lower adjacent RF interference, after passing through a typical RF descending converter, will present toothing in the main signal through the final IF frequency conversion. However, to reduce this tooth formation, a low bandpass IF signal is generated, such that the low IF signal 511 is supported between 50 and 250 Khz with a "free" region of 0 to 50 Hz. Free region guarantees that the first Khz of the adjacent lower RF channel interference remaining not toothed in the main signal. In other words, if the low IF signal 511 has been set to 100 Khz and no free region is allowed, the IF 10.7 MHz signal will have to be designed such that the attenuation of the stop band at 10.6 MHz is the same as that of the case previous to 10.55 Khz. However, the selection of the low IF signal 511 centered at 150 Khz relaxes the stop band attenuation requirement for 10.7 MHz IF filter. Also, it should be noted that while in traditional RF design it is important that the RF down converter 505 does not introduce undulations and distortion with envelope delay in the band of interest, that requirement can be relaxed in the implementation of Figure 5, since a digital system with a compensator follows (described below). In fact, any IF filter imperfection can be compensated by the receiver equalizer at a minimum cost of interference improvement if any. However, it is important that the signal energy is adjusted before the IF filter and in this way any interference from the internal back system is negligible. In fact, the IF filter response can be shifted towards the lower frequency to obtain more band-stop attenuation to the lower adjacent channel interference.

The low IF 511 signal is applied to the filter 590 which is designed to include both a baseband analog low pass filter and a high pass filter in series. The low pass filter (not shown) is designed with a significant band-stop attenuation between 250 and 270 Khz to further reject the upper adjacent channel interference. This baseband analog low pass filter prevents formation of toothing in the sampling process caused by possible insufficient IF filter rejection of 10.7 Mhz. The high pass filter (shown) is designed to further reduce the lower adjacent channel interference which may adversely affect the synchronization detection. (Note that it does not eliminate lower adjacent channel interference that has already formed with serrations in the main signal due to insufficient rejection in RF frequency conversions). The low IF output signal of the filter 590 is sampled by the analog-to-digital converter (A / D) 515 at three times the symbol rate, here illustratively 540 Khz (the Nyquist frequency is 270 Khz). The digital sample stream 516 of the A / D converter 515 is applied to the digital gain control (DGC) 520, which develops a stream of received digital samples 521. The latter applies to the correlator 525 and delay line (or separator) 530 .

The delay line 530 is designed to take into account the processing delay to process the channel pulse, the calculation of the compensating coefficient and the delay that is required to implement the ambient average compensation (all of which is described below). In a common conventional receiver design, a pair of Hilbert filters is used to generate quadrature-phase signal components and a complex (crossover-coupled) compensator is used to recover the transmitted baseband signal. The complex compensator (coupled by diaphragm) comprises four filters arranged in such a way that the quadrature and phase output signals, each generated by two filters. For example, the output signal in phase is the result of a first filter processing of the phase feed signal and a second filter processing of the quadrature feed signal. The quadrature output signal is constructed in a similar way with a pair of different filters. Despite this obvious suggestion, the aforementioned generation of a band-IF low signal as opposed to a more conventional baseband signal - it allows the use of the uncoupled compensator for crosstalk 570. The compensator 570 is composed of two filters, one for phase (I-EQ 570-1) and the other for quadrature (Q-EQ 570-2). Both of these filters have a common power signal at a 3 / T speed, here referred to as fractionally spaced samples. Each filter produces output signals recovered in phase and quadrature at a speed 1 / T. of samplers 575-1 and 575-2, respectively. Although not described here, it can be shown mathematically that the compensator that is not crosstalk-coupled 570 not only retrieves a two-dimensional signal but also forms a Hilbert pair. This result is important for designing both quadrature compensating coefficients and in-phase of the sounding signal by single-in-phase channel substitution (described below). It should be noted that the complexity of the physical compensating equipment is reduced by using a compensator that is not crosstalk coupling 3 / T compared to that of the aforementioned 2 / T crosstalk coupling compensator. For example, although the compensator 570 operates at a sampling rate of 3 / T, only two filters are required. This is a 25% saving in physical compensating equipment, compared to the compensator of four crosstalk coupling filters 2 / T in addition to the savings in the pair of Hilbert filters required in front of the crossover coupling compensator. Phase and quadrature output signals (which are streams of digital samples at symbol speed 1 / T) are applied to the carrier recovery loop 580, which compensates for any phase shift fk in the received signal. As described above, since a rotator is present in the transmitter 100 to compensate for the generation of the passband of the low IF signal 511, a low order phase lock loop can be employed in the receiver 300 (as described above). previously) to quickly track any frequency / phase changes in the received RF signal. This ability to quickly acquire the received RF signal is especially important when the receiver is located in a moving vehicle and is subject to the aforementioned Doppler effects. However, in this illustrative embodiment, the bearer recovery loop is represented by the digital bearer recovery loop 580 comprising the rotation estimator 580-2 and phase rotator 580-1. For purposes of illustration, it is considered that the phase rotation estimator 580-2 is implemented in a digital signal processor (DSP) (not shown). The DSP measures the difference in angle between the output of the compensator and its ideal position and feeds this information back to the phase rotator 580-1 for counter rotation of the angle difference. The phase difference is obtained by averaging over 32 symbols and uses the resulting calculation for the next 300 symbols. As a result, only six estimates are made during a whole table or data block (as illustrated by table 205 in Figure 4). This is a procedure with block-based anticipation correction instead of a loop phase feedback implementation in typical phase, where the error is taken from the difference between the output of the rotator and the slice portion is filtered and used to direct a circuit of phase locking to give the estimate. This procedure with early correction is simple and agile to follow up especially for a fast moving vehicle. However, its tracking range with frequency offset is very limited, for example in the order of ± 18 Hz. Furthermore, the carrier phase change in every 300 symbols can be significant enough to cause degradation. It should be noted that the implementation of a symbol-based second-order loop-in-phase or an additional frequency tracking loop in the carrier recovery loop system can also improve tracking. Before proceeding further with a description of the phase corrected symbol stream processing that is provided by the digital carrier recovery loop 580, attention is now directed to correlator 525, pattern coupler 540, decision device 565 and response element 535. These elements provide the functions of frame synchronization and channel characterization.

With respect to frame synchronization, an algorithm for synchronization detection is generally designed to correspond to a predetermined signal pattern (also referred to as the training signal) in the receiver to the input signal. If there is correspondence, synchronization is declared. The devices used to produce the information in determining whether or not there is correspondence are called a correlator. A counter at the output of the correlator is increased or decreased according to whether a particular symbol in the synchronization signal is reciprocated or not. The result of the counter indicates the similarity between the input signal and the signal pattern stored in the receiver. This accountant is referred to as a trusted accountant. This simple synchronization mechanism uses only very limited information available from the correlator. The regular data signal unless it is restricted may have a pattern similar to the synchronization signal and may cause a false detection. However, it has been considered that a synchronization detection device can be improved if the synchronized signal is designed in such a way that the correlator, when adjusted to the synchronization signal correspondence, results in distinctive signal characteristics that can be used to differentiate the synchronization signal from the data signal in an environment with interference. In this case, a particular training signal with a certain property is required. It is also convenient that the information obtained in the synchronization process from this particular training signal is used to determine the characteristics of the channel (channel sounding). Therefore, a synchronization pattern with distinctive properties of the autocorrelation function is used to improve synchronization reliability. In illustrative form, that is a reason for the use of a binary pseudorandom sequence in the head 206. When this sequence is repeated in the transmitter and correlates in the receiver with a copy of the non-repeated pattern, the correlator produces a high value (peak ) when there is a correspondence and low values (silent zone), in another way. Since the sequence is repeated, the information known a priori and regarding the periodicity and the width of peaks and silent zones can be used to improve the reliability of detection. When this sequence of Ns with symbols that acquire values 1 and 0, is repeated in the transmitter and correlated by a stored copy of the sequence with value 1 and -1 (-1 substitutes 0) in the receiver, the output of the correlator results in a high amplitude peak (n + l) / 2, when the sequence is adjusted to a low value, otherwise. It is also possible to use sequences of values 1 and -1 in the transmitter and to correlate with a stored copy of the sequence of values 1 and 0. The received signal can be physically sliced to a 1 or 0 for simple processing. However, this method alone is not robust in the case of severe channel distortion, such as a channel sequence. Therefore, as described below, it is considered that any correlator feed retains full precision given by the A / D converter. That is, the output signal of the correlator is representative of real numbers and is not limited to a simple logic 1 or 0. Returning to FIG. 5, the stream of received digital samples 521 is applied to the correlator 525. The latter is implemented using a sub-correlator structure. Examples of sub-correlator structures can be found in U.S. Pat. No. 5,260,972, titled "Technique for Determining Signal Dispersion Characteristics in Communication Systems" (Technique for determining signal dispersion characteristics in communications systems) granted on November 9, 1993 to Wang; and the US patent. No. 5,406,586 entitled "Signal Correlation Technique" granted on April 11, 1995 to Wang. It will be noted that if synchronization symbols are no longer restricted to one-dimensional, additional sets of correlators are required. The synchronization process is designed to detect the start of each data block of 10 ms, composed of two consecutive 31PN sequences, followed by a partial 24PN segment (the aforementioned 86 symbols of head 206). Using a copy of the 31PN sequence as its coefficients and receiving a feed signal with full precision, the correlator 525 produces a correlator output signal 526. An example of correlator output signal 526 is illustrated in Figure 8, which shows the output signal of the correlator 526, while the head 206 of a frame is received. For example, region 11 corresponds to the tail end of the data portion of the previous frame, region 12 generally corresponds to head 206 of the current frame, and region 13 corresponds to the beginning of the data portion of the current frame. (The two remaining synchronization symbols 7PN in spindle 206 are compensated and used to synchronize the deinterleaver and to adjust the symbol location in each received data block described below). The output signal from correlator 526 is applied to confidence counter 540, which comprises the following circuits: high slice 545, low slice 555, high pattern matching 550 and low pattern matching 560. High slice circuitry 545 slices the signal output of correlator 526, to provide a one or zero depending on whether its absolute value exceeds a predetermined high threshold. Similarly, low slice circuits 555 slice the output signal of correlator 526, to provide one or zero, depending on whether its absolute value is less than a predetermined low threshold. It is also possible to represent the aforementioned one or zero with real numbers, to reflect the quality of the signal, when compared with high or low threshold. These two sliced outputs are then applied to high pattern matching circuits 550 and low pattern matching circuits 560, respectively. The high pattern matching circuits 550 and the low pattern matching circuits 560 are also referred to herein as the secondary correlators. The latter produce information to indicate how similar the self-correlation of the power signal is compared to that of the pre-stored signal. The information of the secondary high and low correlators is then weighted and summed for synchronization decision making by the decision device 565, which provides a synchronization (sync) signal. It should be noted that a correlator that has a hard slice feed has well-defined peaks and silent zones at the output - if the slice feed is correct. As noted above, it is considered that in the presence of channel imbalances, a correlator that accepts full precision is employed. The output of a full precision correlator is the convolution of that response of the hard slice correlator and the channel impulse response. Since the received signal is band-bound and distorted by multiple paths, silent zones will not exist if the channel extension is longer than the length of the transmitted PN sequence. Therefore, to avoid the silent zone that is completely corrupted, a PN sequence of 31 symbols is used because the 31PN sequence is much longer than the worst case channel extension. This ensures that there will be areas at the output of the correlator that are silent and can be used for reliable synchronization detection. A high threshold pattern (peak) is used to detect periodic peaks separated by 31 symbols. The width of each peak detection zone in the high threshold pattern is L samples, where typically L = »3 or a symbol. A low threshold pattern is used to detect periodic silent zones. The width of each silent zone detection is N samples typically of 18 or six symbols. The region where there may be a response triggered by multipath is defined as the "ignore zone". Its contribution is adjusted to zero by setting the associated coefficients in the secondary correlators to zero values. (It should be noted that in region 12 of Figure 8, the first expected peak of the first 31PN sequence is in some proportion, correlated with data from the end of the previous frame, however, the second received 31PN sequence, absent of corruption , has clear peak areas and silent zones, since the second 31PN sequence is theoretically correlated with the first 31PN sequence.The final PN sequence may have less than one peak since it is only a 24PN sequence). This is illustrated in Figure 9, which is identical to Figure 8 except for the illustrative labeling of a "peak detection zone" (P), "ignoring detection zone" (I), silent detection zone "(Q) During the peak detection zone, a peak matching pattern is searched After the peak detection zone, the correlator output is ignored for a period of time as represented by the detection ignore zone (also illustrated as n). After the last one, a silent correspondence pattern is searched during the silence detection zone The illustrative peak mapping and silent correspondence patterns are shown in Figure 10.

Essentially, this synchronization process looks for the highs and lows and the periodicity of the framing signal. Despite a high correspondence during the peak detection zone, a low correspondence during the silent detection zone is used to verify the start of a frame. The ignore zone compensates reflections, delays, etc. to the received signal. This general method for use in the decision device 565 is illustrated in Figures 11, 12 and 13. Initially, when the receiver 300 is first tuned to a respective frequency, the decision device 565 starts in an "acquisition" mode as it is illustrated in Figure 11, where synchronization is not declared. A correspondence counter, which may illustratively be a variable or a register, is initially set to zero in step 60. In step 61, the synchronization process attempts to detect a spindle. Once a spindle is detected, the correspondence counter is incremented in step 62, a plurality of symbols n,, is skipped in step 63, and an attempt to detect a spindle is again performed in step 64. The plurality of symbols n3 is related to the frame length (described below). if no spindle is detected, the decision device 565 returns to step 61. However, if a spindle is detected, the matching counter is incremented in step 65. If the matching counter is equal to a pre-defined number, Ml f then decision device 565 switches to "steady state mode" in step 67 and operates in accordance with Figure 13 (described below). For example if Mx is equal to 3, then once three consecutive heads are detected the transition to steady state mode occurs and a synchronization (sync) signal can be estimated. This requirement of sequential matching to a plurality of headers establishes a confidence level before declaring synchronization and switching to a "steady state mode". However, if the mapping counter is not equal to Mx the decision device 565 jumps n3 symbols and goes to step 63. It will be noted that the value of the pre-defined number Mx can be static or variable. For example, when the receiver is initially activated (or when a station first tunes in) the value of Mi may be greater than a value for Mi when the return to the acquisition mode was the result of a loss of synchronization. This will enforce a higher confidence level when the signal is first acquired. However, a lower confidence level can be tolerated to handle re-synchronization. The difference between steps 61 and 64 is that of the assumption regarding the position within each received frame, when an attempt is made to detect a head. In the context of step 61, spindle detection is initiated at any point within the received signal. That is, the received signal is "searched" for a peak zone in step 61 of FIG. 11. Once a spindle is initially detected, if it is an actual spindle, a similar detection will occur at the beginning of the next frame, which is a later fixed time interval, as represented by the n3 symbol jump. In this context, the spindle detection of step 63"searches" the spindle at the appropriate later time interval (since the periodicity of the signal is known a priori, the data signal portion of a frame is easily skipped). As a result, although illustrated separately, steps 61 and 64 essentially detect the spindle in the same way. This is illustrated in Figure 12, which shows a flow diagram of an illustrative head detection method. In step 50, the synchronization process searches for a peak zone, i.e. a decision zone 565 waits for the detection of any indication of a peak zone from a confidence counter 540. When detecting a peak zone, the synchronization process jumps to symbol intervals in stage 51, to compensate for any signal reflection, etc. (This is an area to ignore). In the next step 52 the synchronization process searches for a silent zone. If no silent zone is detected within a predefined time interval, the synchronization process returns to step 50 to search for a peak zone. However, when detecting a silent zone, then the decision device 565 searches for a peak area in step 54 within a predetermined time interval. If no peak zone is detected, the synchronization process returns to stage 50 to scan a peak zone. However, upon detecting a peak zone in step 54, the synchronization process jumps -pa symbol intervals in step 55, to compensate for any signal reflections, etc., (this is another ignore zone) where n2 = n .. The values of nx and na are determined experimentally based on the knowledge of the worst case channel extension. In the next step 56, the synchronization process searches for a silent zone. If no silent zone is detected within a pre-defined time interval, the synchronization process returns to step 50 to search for a peak zone within a pre-determined time interval. However, when detecting a silent zone, then the decision device 565 searches for a peak zone in step 58. If no peak zone is detected, the synchronization process returns to stage 50, to search for a peak zone. When detecting a peak zone, the synchronization process has detected a spindle. It should be noted that the sum of the intervals of time elapsed in skipping the ignore zones, and searching for peaks and silent zones is designed in such a way that the periodicity of 31PN is used.

Once in a stable state mode, the above described method of Figure 12 can also be used for each frame. Alternatively, other methods may be employed, one of which is illustrated in Figure 13. In the latter, a fault counter, which illustratively may be a variable or a register, is initially set to zero. In step 72, the decision device 565 searches for a peak zone within the following frame, which in this example occurs n4 later symbols. Here, nA > n. since as will be described, only the first peak and the silence zone are detected, therefore more than the table requires to be skipped. If a peak zone is detected, the decision device 565 skips x symbol intervals in step 73 and searches for a silent zone in step 74. However, if no peak zone is detected, a fault is declared in step 75 and the fault counter increases. The value of the fault counter is verified in step 76. If the fault counter value is greater than a pre-determined number M2, then the decision device 565 changes back to the acquisition mode, ie synchronization is lost and decision device 565 operates in accordance with Figure 11, described above. Otherwise, the decision device 565 proceeds to step 81 and therefore to step 72 as described above. The value of i-, is greater than n < since more symbols require to be skipped (as represented by the time to process stages 73, 74, 78 and 79). At step 74, if a silence zone is detected, decision device 565 provides a sync signal in step 70, jumps n4 symbols in step 71 and searches for a peak zone at the start of the next frame in step 72. However, if no silent zone is detected, a fault is declared in step 78 and the fault counter is incremented. The value of the fault counter is verified in step 79. (It is considered here that a fault is represented by an invalid sync signal, for example a logical ZERO against a logical UNO, or if the sync signal is a multi-bit signal, In the alternative, a separate signal can be provided by the decision device 565). If the value of the fault counter is greater than a predetermined number M2, then the decision device 565 changes back to the acquisition mode, ie the synchronization is lost and the decision device 565 operates in accordance with Figure 11, as It was described earlier. Otherwise, the decision device 565 proceeds to step 71 and the process continues. In this example, the fault counter is allowed to increase until it reaches the aforementioned threshold. However, variations can be used to readjust the fault counter. For example, the fault counter may be readjusted periodically, if no fault is detected within a predetermined time period. 0, the fault counter can be readjusted each time a synchronization signal is provided in step 70. Also, it should be noted that a failure in detection or lack of synchronization, can be advantageously employed to indicate the signal-to-interference condition channel. For example, in the case of a failure detection or a lack of synchronization, a signal (for example the sync signal) is sent to the Reed-Solomon decoder. The latter then ignores the current received frame for the purpose of correcting errors. This is more efficient than accumulating error information in a received signal over a period of time as is typical in prior art systems referred to as a deletion in Reed-Solomon decoder. For example, it is known to develop error statistics based on slicing an output signal from a compensator. When the error accumulates over a certain threshold, circuits such as a Reed-Solomon decoder subsequently ignore the received signal for a period of time. Nevertheless, simply using the aforementioned sinc signal eliminates these circuits. As described above, the polling signal is transmitted every 10 ms. Based on theoretical and experimental results, it has been determined that this limits the vehicle's top speed to between 135 and 200 km per hour, depending on the severity of the communications channel. The ability to deal with vehicle speed increases linearly with an increase in the repetition rate of the sounding signal. For example, if the sounding signal is transmitted every 5ms, the top speed of the vehicle will be in the range of 270 to 400 Km / Hour. It should be noted that other methods can also be used to determine synchronization using the circuits of Figure 5. For example, in contrast to the previous serial approach for evaluating the output signals of the confidence counter 540, as described in Figures 11 and 12 , a type of joint decision of analysis can be used. Once frame synchronization is achieved, the impulse response to the channel embedded in the output signal of the correlator 526 is processed to obtain compensating coefficients for the compensator 570. In particular, once synchronization is declared, the predefined head, that is, the training signal is identified and used to represent the channel impulse response. This is particularly useful because a mobile wireless channel is characterized by the presence of multiple reflection paths. As such, the received signal can be seen as composed of a main signal and a significant amount of indirect signals caused by reflection. Reception difficulty is also increased by continuous change in the channel and becomes even more difficult when transmitting data symbols at a speed exceeding a few hundred Khz. For example, a delay of a few micro seconds between paths causes inter-symbol interference between many data symbols. As a result, to recover data in this type of communications environment, channel characteristics are required to correct the deterioration caused by the channel. The characterization of the channel can be described as follows. At the transmitter, a known training signal is sent through an unknown channel. In the receiver, the observed signal received is used to characterize the channel. We define A (f) as the training signal frequency spectrum, H *. (F) as the transmitter frequency response, Hc (f) as the channel frequency response, Hu (f) as the frequency response of tuner, B (f) as the frequency response of channel characteristic on the receiver and T (f) as the total frequency response. (For now, it is considered that the system does not have interference). T (f) = A (f) Ht (f) Ha (f) Hu (f) B (f) (12) Yes A (f) B (f) is equal to a constant k over the transmission band, a referred to below, as the "processing gain", then: T (f) = kHt (f) Ha (f) HJf), or (13a) T (f) = kH (f) (13b) where H (f) ) is the total transfer function and the channel information is obtained. Using this information, compensator derivation coefficients can be obtained to correct channel distortion. The following described techniques precisely characterize a multipath channel and provide various types of information for synchronization and estimation for carrier phase shift as described above. If the channel is corrupted by interference, then: T { f) ~ kH (f) + kÑ (f),. { ! ) wherein N (f) is the channel interference power spectrum and k is the processing gain as described above. The ratio k \ H (f) \ 2 / N (f), integrated over the transmission band defines the estimated signal-to-interference ratio of channel. The greater the processing gain, the better protection will be given to the estimated channel characteristics of interference. In general, the harder the training signal, the better the communication channel characterization will be. In this design, the training signal corresponds to the aforementioned head. Consequently, there is a compromise between the factors that reduce the transmission performance and the amount of time dedicated to characterize the communication channel. This compensation is represented here by the selection of 86 symbols in the head 206, such as the length of the training signal for synchronization and channel polling purposes. As previously noted, the compensator 570 forms a Hilbert pair, such that the coefficients for the quadrature and phase compensators of the compensator 570 can be obtained from the phase channel response only. As such, the channel assignment response element 535 first finds the phase channel response embedded in a correlator output signal 526. The new quadrature compensation coefficients are obtained through a Hilbert transform. An illustrative process to obtain linear compensating coefficients is illustrated in Figure 14. (It is possible to use variations in a compensator with decision feedback, reference to the US patent application by Gadot et al., Serial No. 08/322877 , presented on October 13, 1994 and granted on November 15, 1995). In step 30, the response element for assigning channel 535 receives the sync signal from the detection device 565, noting that the output signal from correlator 526 is representative of the channel impulse response. In step 31, the response element for assigning channel 535, transforms the output signal of correlator 526 from the time domain, into a frequency domain representation according to a "Fast Fourier Transform" (FFT) technique. or "discrete Fourier transform" (DFT) (the FFT and DFT processes are known in the art). Normally the compensating coefficients can simply be determined by taking the reciprocal of the FFT output (for frequency domain compensation) and the inverse FFT and the IFFT (to return to the time domain) (for time domain compensation). However, excessive improvement of compensator interference may be due to the presence of multipath reflections. This is particularly true for large reflections, for example identical force reflections. As such, only the compensating coefficients generated by the FFT response may not provide convergence and on the contrary make the intersymbol interference (ISI) recovery difficult. Therefore, it has been considered that by introducing a small distortion in the channel impulse response, multiple reflections can be handled with only a slight degradation in the overall performance. In particular, the channel impulse response is trimmed in the frequency domain if the received signal is too high or too low, hereinafter referred to as "FFT threshold formation". In other words, a simple threshold is applied to the magnitude of the FFT of the output signal of the correlator 526 as illustrated in Figure 15. If the FFT of the output signal of the correlator 526 exceeds these predetermined thresholds -rh and tl t the Signal is simply cropped. For example, if the magnitude is greater than rTh, the magnitude is set equal to rh. Similarly, if the magnitude is less than l t the magnitude is set equal to Ti. The determination of the thresholds is a compromise between the magnitude of the expected reflections and the degree of ISI that is acceptable and must be determined empirically. This FFT threshold formation approach avoids improvement of excessive compensating interference in compensation due to multipath environment. In an analogous way, can be seen as compensation using a good type of minimum quadratic mean as opposed to forced zero compensation. It is also important that in order to avoid the effect of circular convolution in the digital frequency time transformations, the length of FFT and IFFT must exceed the sum of the worst case channel and compensator extensions to avoid cyclical toothings when performing FFT and IFF operations. Finally, the out-of-band response is set to zero in the frequency domain, to designate the compensator with a bandpass characteristic of 180 Khz between 60 to 240 Khz.

Returning to Figure 14, the element for channel assignment response 535 applies threshold formation FFT in step 32. The reciprocal of the FFT threshold is taken in step 33. The resulting frequency domain response is then processed using a reverse FFT. (IFFT) as is known in the art to obtain the phase compensator coefficients in step 34. Finally, the response element for assigning channel 535 processes the phase compensating coefficients by means of a Hilbert transform in time domain, to obtain the quadrature compensating coefficients in step 35. Once the compensating coefficients are obtained, the element for response assign channel 535 updates or discharges the compensator coefficients 570 in step 36. As noted above, delay line 530 must regulate the sample flow such that the compensator 570 searches for the appropriate data stream. The length of this data separator takes into account the time to process the channel pulse, the calculation of compensating coefficients and the delay that is required to implement the compensation. It is important that the sampling phase relationship is adequately maintained across all signal processing and delay circuits. After the delay separator, the compensator processes two partial blocks, half data block before and one half after the synchronization pattern of 100 symbols. This is the so-called average ambient compensation. It should be noted that the compensator is a passband compensator. The same data stream is fed to the in-phase portion of the compensator 570 as well as the quadrature portion of the compensator 570. The output of the compensator 570 is resampled at the symbol rate and provided to the carrier recovery loop 580 previously described. The output of the carrier recovery circuits is then sliced to retrieve transmitted symbols. A variation of the method of Figure 14 for calculating compensating coefficients is illustrated in Figure 16. The latter is identical to Figure 14, except for replacing step 35 with step 45 and moving step 36. In step 35, phase compensating coefficient in the frequency domain is processed by a -jsgn (f) (the frequency domain representation of a Hilbert transform) that generates the phase quadrature compensator coefficient in the frequency domain, from which the time domain coefficients for quadrature are generated by taking IFFT in step 34. Returning to Figure 5, the stream of phase-corrected symbols that is provided by the loop for recovery of digital bearer 580, is provided to the recovery element of symbols 705, which provides coded signal 351, previously described. The symbol recovery element 705 is illustrated in block diagram form in Figure 17 and comprises the correlator 710 and counter-separator 715. As a reminder, the remaining 14 one-dimensional symbols of the head 206 are used either for synchronization interleaver or clock alignment of symbols on the receiver 300. These 14 one-dimensional symbols are chosen from the sync symbols of interleaver (symbol clock) "of the constellation of signal points of Figure 3. In particular, the 14 symbols are used by the receiver 300 to indicate the start of each 320 ms interleaver block This interleaver synchronization is repeated every 32 frames and comprises two consecutive sequences of 7PN as illustrated in Figure 4. (It should be noted that the interleaver depth is say the interleaver block size affects the ability to recover the signal from the clogged terrain and it is also a function of the lower limit of the speed of the vehicle. Consequently, other variables of intercalary depth can be used, depending on particular characteristics of the system. For example, if a higher incidence of clogged terrain is present in the diffusion area, the interleaver depth can be adjusted to 640 ms).

When the 14 symbols for interleaver synchronization in the receiver 300 are not used, they are used for synchronization of data symbols in the associated frame. In this instance, the 14 symbols comprise a negative 7PN followed by a positive 7PN as illustrated in Figure 4. This is required to align the coded data portion of each frame, as long as there is a phase change in significant sample time or the relative strength of different signal path changes and causes a change in the delay seen by receiver 300. It should be noted that in this context, a negative PN sequence is simply the opposite of a positive PN sequence. For example, if 2PN are represented by the symbols (1,414.0; -1,414.0), the corresponding negative 2PN is the sequence of symbols (-1,414.0; 1,414.0). As such, the correlator 710 is similar in function to the previously described correlator 525, confidence counter 540, and the decision device 565 except that it has an additional information signal - the sync signal, which establishes frame synchronization for the frame received current (As a result, the correlator 710 can be a simpler binary correlation). During a valid frame (as represented by a valid sync signal), the correlator 710 provides interleaver synchronization signal 352 for subsequent use by the error protection decoder 315, illustrated in FIG. 1, to de-interleave symbol blocks before detection of two consecutive 7PN sequences of the same sign. Similarly, the correlator 710 provides a synchronization signal of data symbols 712 upon detection of two consecutive 7PN sequences of opposite sign or upon detection of the interleaver synchronization sequence. The last condition ensures symbol synchronization, even during that frame indicating the start of a new interleaver block. It should be noted that when using two 7PN sequences of the same sign for interleaver synchronization and two 7PNs of the opposite sign (one positive and one negative) for symbol synchronization, the receiver decoding is designed to be phase rotation invariant. This provides additional protection to this important date stamp information in case of a severe channel. Counter-separator 715 responds to the data symbol synchronization signal 712 and separates the data-only portion of the current frame. The implementation of the 715 intermediate meter-separator can be done in any number of ways. For example, as a linear separator, or a circular separator, additions and deletions are made using a pointer and a counter. When storing the current received symbols, the counter-separator 715 is considered to perform the following functions. First, the counter-separator 715 rigidly slices the received symbol stream. (For simplicity - the slicer - an element known in the art - is not illustrated). The symbol stream received slice is then stored. Ideally, the number of sliced symbols stored should be equal to or greater than the predefined size of a data block, ie 1700 data symbols. However, a synchronization shift may already cause more or fewer data symbols to be associated with the received frame current. This synchronization shift is due to misalignment of the transmitter and transmitter and receiver clocks and the multipath aspects of the communication channel itself. With respect to the transmitter and receiver clocks, the above-described estimated channel impulse response obtained from the channel poll, has synchronization phase shift information between the transmitter and receiver symbol clocks. A fractional spaced compensator that uses coefficients derived from the estimated channel response, it can compensate the synchronization phase shift to a limited extent, since the compensator can be kept frozen until the next synchronization arrives. If between the transmitter and receiver clocks have a frequency difference, a synchronization phase shift is generally increased from zero to a certain value before the new estimate arrives. It is well known that the sensitivity to this problem depends on the excessive bandwidth of the transmitter filter. For example, if a transmission system uses a transmitter filter with excessive bandwidth of zero percent (sin x / x), a synchronization phase shift of four percent (15 °) will introduce a noise of - 23 dB per second. below the signal. When this noise is added to the interference, it causes a degradation of 0.3 dB to the sensitivity of the receiver, if a transmitter filter with approximately 10% of excessive bandwidth is used, this is comparable to a synchronization phase shift of 11. %. The inaccuracy of the maximum allowed receiver symbol clock of the following equation: i praised them to a block of D data ¿doßßssppl_.aazzßßa «i-ßßnnttoa daee,. , T 1 aiicroni-icio-paraitido. { '- >;) where D is the synchronization shift for the maximum allowed receiver symbol of the transmitter and a division by two is due to amble mean compensation. For a synchronization shift of 11% of a data block size of 1800 symbols, D = 120 ppm (parts per million). As long as the receiver symbol clock is within ± 120 ppm of the transmitter symbol clock, the phase in synchronization will not change significantly against the data block to cause significant performance degradation.

Another case that may cause deletion or addition of symbols is when a multi-path channel varies. For example, when a relative intensity of all trajectories changes, the compensator always takes that with the strongest power as the main signal and produces symbols recovered in accordance. In any case, the end result is that if there is a displacement in time in the alignment of symbols, more or less data symbols may exist in the current received frame. This displacement in symbol time will probably occur when the compensator coefficients are changed. Therefore, the counter-separator 715 measures the number of data symbols between any two consecutive double sequences 7PN, as represented by the data symbol synchronization signal 712. When there is an additional symbol, that in the middle of the block It is eliminated. If the number of symbols is less than 1700, the average symbol is repeated. In this situation, an error may occur. While the simple format design can take this problem into account, for simplicity of implementation this condition can be ignored and on the contrary, the Reed-Solomon decoder (not shown) with the decoder for error protection 315 can be used to correct this problem. This causes less degradation in total system performance. It should be noted that since a half-ambient compensator is used, two consecutive data blocks are placed in the separator to perform this symbol realignment. The foregoing merely illustrates the principles of the invention and in this way it will be appreciated that those skilled in the art will be able to design numerous alternate arrangements which, although not explicitly described herein, incorporate the principles of the invention and are within their spirit and scope. For example, although the invention is illustrated herein as being implemented with discrete functional building blocks, for example a perceptual audio encoder, response element for assigning channel, etc., the functions of any one or more of those building blocks can be carried performed using one or more appropriate programmed processors, for example a digital signal processor. It is noted that in relation to this date, the best method known to the applicant to carry out the aforementioned invention, is that which is clear from the present description of the invention. Having described the invention as above, property is claimed as contained in the following:

Claims (17)

1. CLAIMS 1. Receiving apparatus, characterized in that it comprises: a radio frequency down converter, operative in a received radio frequency signal, to provide an intermediate frequency signal; and an intermediate frequency descending converter for providing a passband signal from the intermediate frequency signal, such that the passband signal is centered with respect to a first selected frequency to be set at a rotational frequency of a rotator inside a corresponding transmitter of the received radio frequency signal.
2. 2. The apparatus in accordance with the claim 1, characterized in that it further comprises: a compensator operative in the passband signal, to provide a compensated signal; and a carrier recovery circuit that responds to the compensated signal, to provide compensation only for phase shifts due to effects of a communications channel that couples the receiving apparatus to the transmitter.
3. 3. The apparatus in accordance with the claim 2, characterized in that the circuit for carrier recovery is a low order phase locking loop.
4. 4. - The apparatus in accordance with the claim 2, characterized in that the circuit for carrier recovery is a circuit for digital carrier phase recovery.
5. 5. The apparatus in accordance with the claim 4, characterized in that the compensated signal represents a sequence of frames and wherein the digital carrier phase recovery circuit is a signal processor that performs phase correction with block-based correction in a portion of each frame of the compensated signal .
6. 6.- The apparatus in accordance with the claim 5, characterized in that the signal processor determines an average phase difference on xl symbols of each frame by comparing one phase of each xl symbols to a reference phase and wherein the signal processor uses the average phase difference to correct the phase of the x2 immediately following symbols.
7. 7.- The device in accordance with the claim 6, characterized in that only those symbols comprising the data portion of the frame are used by the signal processor.
8. 8. The apparatus according to claim 6, characterized in that x2 > xl and x2 is less than the total number of symbols that comprise the data portion of each frame.
9. 9. - The apparatus according to claim 6, characterized in that it further comprises a rotator that is controlled by the digital signal processor to counter-rotate the x2 symbols by the determined average phase difference.
10. 10. The apparatus in accordance with the claim 2, characterized in that the circuit for carrier recovery further comprises: an averaging circuit for operating on the compensated signal, which represents data formatted in a sequence of frames, each frame has a head portion and a data portion comprising an amount of symbols, wherein the averaging circuit averages a phase difference between each xl symbols and a reference signal to provide an average phase difference; and a rotator that rotates the immediately following x2 symbols by the average phase difference determined where xl < x2.
11. 11. The apparatus according to claim 10, characterized in that the averaging circuit is a signal processor.
12. 12. The apparatus in accordance with the claim 10, characterized in that the symbols xl and x2 are taken from the data portion of each frame.
13. 13. The apparatus according to claim 2, characterized in that the compensator is a compensator not coupled by crosstalk.
14. 14. - A method for transmitting a communication signal, characterized comprises the steps of: mapping symbols of one. data signal to generate a first signal representative of a stream of N dimensional symbols; rotating a phase of each N-dimensional symbol by a rotation frequency, to provide a second signal in such a manner that the rotation frequency falls within a range of band frequencies of a passband signal to recover the signal of communications by a receiver; and processing the second signal to provide the communication signal for transmission over a communication channel.
15. 15. The method according to claim 14, characterized in that the processing step further includes the steps of: converting the second signal in ascending order to an intermediate frequency signal; and upconverting the intermediate frequency signal to a radio frequency signal which is the communication signal.
16. 16. The method according to claim 15, characterized in that the processing step further includes the step of adding a pilot signal to the second signal before upconverting the second signal to the intermediate frequency signal.
17. 17. - The method according to claim, characterized in that the turning stage provides a carrier signal.