4 stars based on
Kind code of ref document: Country of ref document: Date duobinary signaling and decoding words ref document: Year of fee payment: The invention relates to decoding of encoded signals transmitted duobinary burst and it is a particularly important application in the television program broadcast facilities in the form of a multiplex of analogue image signals and digital signals, coded in duobinarysound and data.
The vector decoding mode duobinary signals is to compare the signal level to thresholds. A more efficient decoding mode described in the patent application FR-A-2,, is to use a maximum likelihood duobinary signaling and decoding words and to implement the Viterbi algorithm. This solution allows, in the case of a transmission of the duobinary signal in full time that is to say continuously to achieve performance reducing the value of the ratio minimum acceptable signal to noise.
But the straightforward application of the Viterbi decoding signals transmitted by bursts significantly degrades performance. This is particularly due to the fact that during the intervals between bursts, duobinary signaling and decoding words decoder receives a signal which is not encoded duobinary e.
MAC signal and submit it to the decoding while presenting violations the coding rule. In the most common case, part of the digital form of information messages is fixed format and another part is in variable format.
These signals are generated by the same encoder duobinary as that used for coding the sounds and the data transmitted in the line blanking intervals, the allocation of lines used by the teletext being defined by information on the time-multiplexed structure information said TDM CTL described in part I, chapter of the Standard mentioned above.
This information is contained in the line of the frame television. It is necessary that the decoding method provided in the reception can accommodate the second type format, which further complicates the problem. In addition, two cases are possible, depending on the operating mode of the broadcast transmitter. Dans le premier cas, le fonctionnement du codeur duobinaire est interrompu entre les salves successives.
In the first case, the operation of the duo-binary encoder is interrupted between successive bursts. During the transmission of MAC image signal and during the alignment time, the data sequence applied to the input of the duobinary encoder is simply forced to zero.
The invention aims to provide a duobinary signal decoding method issued by bursts to approach the performance obtained for continuous data stream, without leading to undue complications. For this purpose, the invention proposes in particular a data decoding process issued in the form of duobinary coded signals in and transmitted in bursts in a multiplex, characterized in that digitizes the multiplexed baseband and subjected to a Viterbi decoding by masking parts of the multiplex which would cause a violation of the duobinary coding rule due to interruptions between bursts.
A first embodiment of the invention that achieves optimal results, requires the known information messages encoded duobinary locations to treat. In this case, the non coded portions are masked by interrupting the duobinary decoder clock signal. It is also necessary to respect the parity coding rule. A simpler, but fails to achieve the most favorable results solution is to extract information transmitted known fixed format of the receiver, ignoring the additional information that may be transmitted in variable format in the intervals blanking: This solution eliminates the need to process the data in line It is possible to achieve optimal results without having to interpret the information in lineprovided that the number of bits at 1 contained in each line blanking is even.
This condition can be always filled by adding a parity bit in each block corresponding to a data line during the frame blanking period. In this case, the last bit of the block transmitted during a line blanking interval before the first line synchronization bit is the parity bit of the bit block that precedes it.
This bit "pad" at the end of each burst duobinary signaling and decoding words used to collect the error in the case where the Viterbi decoder operates continuously no stop at the end of each burst. The invention also provides a device for implementing the above defined method.
The invention will be better understood from reading the following description of specific embodiments of the invention, given as non-limiting examples. The description refers to the accompanying drawings, wherein: Figures 8a and 8b show a possible construction of the two constituent parts of the Viterbi decoder of an implantable unit in full decoding device.
This multiplex has simplified manner shown in Figure 1 structure where legends L1, The multiplex sound-data digital occupies bursts per frame, the line having spare bits and an alignment marker, while the full line is assigned to the transmission of a clock synchronization word, a frame synchronization word and service data.
Chaque ligne comporte un mot de synchronisation de six bits. Each line has a six-bit synchronization word. Sound and data are transmitted in fixed-length packets each having bits, a header of twenty-three bits and ninety-one bytes of payload data. Each line contains ninety-nine useful bits, the first beginner package immediately after the line sync word line 1 and the package 82 ending in the line The number of rows affected teletext may be implied fixed size or left to the free choice of the diffuser variable size and, in this case, the number of lines and their numbers are shown in the line Under these conditions, for optimum operation, the decoding duobinary signaling and decoding words must satisfy two conditions: Before Viterbi decoding itself, the decoded signal should be processed to reconstruct the input of the decoder a duobinary sequence in which there is no violation of the duobinary coding rule that is to say no transition prohibitedeliminating all parts of the signal is not representative of actual broadcast messages.
It is necessary to eliminate all sequences of "zero" known to the receiver during the implementation of the Viterbi algorithm by interrupting the clock signal, so as to avoid overflow in memory registers of the decoding circuit Viterbi. Duobinary signaling and decoding words the number of "zeros" is even eliminated, simply juxtaposing the samples received burst to burst.
If it is odd, the received samples will be juxtaposed with burst burst inversion to avoid rape of the coding rule. The application of those rules requires that it be possible to determine the receiving location data introduced when coding the issue. The filtered signal is applied to a circuit 12 for automatic gain control and restoration of the DC component.
The signal is then applied, first, the MAC decoder 14 for providing the image signal and on the other hand, the circuits that process the digital component. These circuits comprise a complementary filter 16 for limiting the band duobinary signaling and decoding words 5 MHz approximately. It will be seen that if the receiver is of mixed type and is adapted to also receive D-MAC programs, a switch may be provided to eliminate this will filter. The signal from filter 16 is applied to two parallel processing paths.
The first channel comprises a duobinary decoder thresholds 18, a circuit 20 for recovery of the clock signal at Another solution is to generate the clock signal at The first route also includes the circuit 30 processing line to know the assignment of the blanking lines used for teletext and locate bursts are not transmitted to fixed format.
The constitution thus far described is conventional and, therefore, does not require detailed description. Quantification over four to six bits is usually sufficient. The converter 32 attack the Viterbi decoder Indeed, the Viterbi decoding causes a delay in the decision: Finally, the duobinary signaling and decoding words channel comprises a circuit 38 for sound and data processing, which can have a conventional structure and is not concerned by the invention.
This circuit 38 provides deshuffling and interlacing frames and plays audio and data. The Viterbi decoder itself can have the structure described in patent application FR 86 For decoding to be optimal, the operation of the Viterbi decoder must be stopped outside the useful burst reception periods, especially when the broadcast signal comes from a encoder which is not interrupted between useful bursts.
In the embodiment illustrated in Figure 2, this is achieved by interrupting the signal from the clock 20 applied to the Duobinary signaling and decoding words decoder after receipt of the sample that corresponds to the "zero" level at the input of the decoder and up to sample, that is to say during periods of the clock signal at The interruption is performed by a circuit 40 controlled from the time base 28 and the circuit 30 processing line This latter circuit program based on time to ensure the required interrupts.
The device of Figure 3 differs essentially from that of Figure 2 by the absence of the complementary filter 5 MHz, since analog and digital signal components occupy frequency bands of the same order and can duobinary signaling and decoding words processed by the same filter 10 to 8.
In addition, the clock signal is recovered at a frequency equal to A time base 40 is further provided to stop the application of the clock signal to the Viterbi decoder 34 after receipt of the sample to the sample of each line, that is to say during periods of the clock signal at We see that a versatile receiver is easily achievable, the only changes necessary to spend D2 duobinary signaling and decoding words D being a switch on the filter 16 and the doubler 22 of Figure 2 and a change on the time base The decoding device shown in Figures 2 and 3 comprises a processing circuit of the line which identifies the multiplex structure.
Insofar as we are only interested in decoding its information-data transmitted in the line blanking intervals, the apparatus can be simplified duobinary signaling and decoding words performing the Viterbi decoding ignoring the supplementary information transmitted in the blanking interval, which in particular implies that ignores the data in line In addition, the metric should be re-initialized at the beginning of each television line may contain digital information.
This re-initialization is to assign the same value to the metric corresponding to the two nodes 0 and 1 of the trellis in the sense of the Viterbi algorithm, as defined in the patent application FR 86 15 already mentioned before applying the decoder the first duobinary sample of the burst.
This resetting can be done by forcing a predetermined level of logic elements contained in the internal circuits of the decoder. This solution can be adopted that the last bit of each burst D sample is not assigned to the user data transmission. But this degradation is duobinary signaling and decoding words a real downside, given the existence of a specific parallel way synchronization. Figure 4 shows, by duobinary signaling and decoding words of example, a possible construction of the decoding channel-sound data for implementing the embodiment which has just been described.
A duobinary signaling and decoding words 44 for resetting metric is also controlled by the time base The latter is intended to provide a pulse at the last clock period before the first sample of each burst and in response, the circuit 44 assigns the same value to the metric corresponding to the nodes 0 and 1 of the trellis.
In the embodiment shown in Figure 5, the metrics are re-initialized from the outside, by applying appropriate signals to the input of the Viterbi decoder The control input 48 receives a signal supplied by the time base 28 during a time window corresponding to the bit transmission duration of each line in the case of a D2 MAC signal PACK. The time window signal and the clock signal are also applied to a counter 52 which enables the generator 50 only during intervals not duobinary signaling and decoding words the sound signal and data.
Pendant chacune de ces lignes: During each of these lines: In yet another alternative embodiment, the device has no metric reset circuit, which represents a further simplification. Degradation of decoding is increased in consideration duobinary signaling and decoding words in particular systematic errors occur in the detection of the line synchronization word, which is of no consequence insofar as line synchronization occurs only from the frame sync word. Before describing in detail with reference to Figures 8a and 8b, the constitution of a Viterbi decoder for use with a quantification of six bits will define how the Viterbi algorithm can be adapted to the duobinary coding.
Assume that the transmitted signal c k is of the form: The sequences to be transmitted are of finite length and can be written: The evolution of the state of the encoder over time can be represented by the mesh of Figure 6, duobinary signaling and decoding words nodes correspond to successive states of the system. Due to the presence of noise, we can write the result of the samples obtained in reception, after passing through the support 16 of Duobinary signaling and decoding words 1, in the form: Due to the independence of the noise samples n k, we can write: This formula shows that the factors add Metric, due to the independence of the noise samples.
The Viterbi algorithm takes into account, among the 2 N possible sequences, the maximum metric sequence: Convergence towards the survivors is statically ensured, even with an arbitrary choice of initial conditions. Once convergence is achieved, the decision can be taken as the common portion up to the maximum path metric D. Almost certain convergence is achieved with a predetermined memory capacity.
Dans la pratique, il suffira, dans le cas du codage duobinaire, d'adopter un treillis de 22 cellules.