GB2334861A

GB2334861A - Testing digital modulation systems

Info

Publication number: GB2334861A
Application number: GB9804248A
Authority: GB
Inventors: Christopher Ryan Nokes
Original assignee: British Broadcasting Corp
Current assignee: British Broadcasting Corp
Priority date: 1998-02-27
Filing date: 1998-02-27
Publication date: 1999-09-01
Anticipated expiration: 2018-02-27
Also published as: GB9804248D0; GB2334861B

Abstract

The noise floor of e.g. a digital broadcast television receiver is determined, by connecting it in an artificial difficult channel. A modulator 10 supplies a signal to a low noise floor transmitter 42, the output of which is combined with the output of a variable noise source 46 in a combiner 44. The resultant signal is applied to the receiver under test 50, the output of which is monitored by a bit-error ratio meter 48. The amount of noise which can be added with the meter showing a predetermined bit-error ratio is noted. A first measurement is made with the modulator operating normally. The modulator is then changed so as to operate on every nth bit to cause the generation of valid but incorrect transmitted values, by forcing every nth bit to take the value 1. A second measurement of the amount of noise that can be added is then made. The noise floor Px can be determined from the difference between the two measurements. The method can be used to determine other errors and artefacts, and can be used to determine the transmitter noise floor instead of the receiver noise floor.

Description

APPARATUS AND METHOD FOR TESTING DIGITAL MODULATION SYSTEMS Background of the Invention This invention relates to an apparatus and method for testing digital modulation systems, and particularly, though not exclusively, for measuring the 'noise floor' in a digital modulation system.

The invention is applicable to the measurement of receiver noise floor (PX(rx)) with digital terrestrial television signals, such as those to be transmitted under the so-called DVB-T transmission system. It can also be of use with other signals such as those in the DVB-S satellite transmission system and the American 8-VSB digital terrestrial television transmission system, for example.

The 'noise floor' of a receiver or indeed of any part of the digital transmission system may be defined as the sum of excess noise contributions which are independent of carrier level, and is usually denoted by Px.

Considering the example of the DVB-T system, a multicarrier Coded Orthogonal Frequency Division Multiplex (COFDM) modulation system is employed. It uses concatenated coding with an outer Reed-Solomon coder and an inner convolutional coder. A block diagram of a modulator is shown in Figure 1, which is taken from European Telecommunication Standard ETS 300 744 of September 1997, published by European Telecommunications Standards Institute (ETSI), F-06921 Sophia Antipolis CEDEX France, to which reference should be made br further description thereof. The inner convolutional encoder takes the form shown in Figure 2. Various inner convolutional code rates are specified, all derived from a mother code of 1/2. Higher code rates, i.e. 2/3, 3/4, 5/6, 7/8, are derived from the mother code by puncturing.

The convolutional code is usually decoded in a receiver using a Viterbi or soft-decision decoder. A soft-decision decoder looks at an input stream of l's and 0's and identifies not just whether any given bit is closest to a 1 or 0, but how confident a decision that is. That is, 0.9 means a fairly confident 1 whereas 0.6 means a not very confident 1, which might in fact be a 0 with a large error.

It has been appreciated that an important parameter in DVB-T receiver design is the receiver noise floor, Pxrx). It is not however clear how this parameter can be measured. There are difficulties in that there are no direct methods of measuring the overall value of PX(rx).

'Difficult' channels exist which include multi-path distortion, co-channel interference (CCI) and/or adjacent channel interference (ACI). The carrier to noise (C/N) ratio for such a channel has to be higher than for a normal, Gaussian, channel. Existing methods for measuring Px include: (i) direct measurement of the components of Px. These can not easily be measured non-intrusively, and components arising from the receiver demodulator circuit can not be measured at all.

(ii) indirect measurement of the effect of Px in difficult channels. This will vary depending on the precise type of channel chosen, and limitations of demodulator algorithms may affect the performance in difficult channels and may mask the true value of Px.

(iii) a substitution method. This may not be appropriate for all demodulator circuits, and may therefore lead to inconsistent results.

There are other features or artefacts of digital modulation systems which are also difficult to measure.

Sununary of the Invention The invention in its various aspects is defined in the independent claims below, to which reference should now be made. Advantageous features are set forth in the appendant claims.

A preferred embodiment of the invention will be described in more detail below. In the preferred embodiment, the noise floor of a digital broadcast television receiver is determined by connecting it in an artificial difficult channel. A modulator supplies a signal to a low noise floor transmitter, the output of which is combined with the output of a variable noise source in a combiner The resultant signal is applied to the receiver under test, the output of which is monitored by a bit-error ratio meter. The amount of noise which can be added with the meter showing a predetermined bit-error ratio is noted. A first measurement is made with the modulator operating normally. The modulator is then changed so as to operate on every nth bit to cause the generation of valid but incorrect transmitted values, by forcing every nth bit to take the value 1. A second measurement of the amount of noise that can be added is then made. The noise floor Px can be determined from the difference between the two measurements. The method can be used to determine other errors and artifacts, and can be used to determine the transmitter noise floor instead of the receiver noise floor.

Brief Description of the Drawings The invention will now be described in more detail, by way of example, with reference to the drawings, in which: Figure 1 is a block schematic diagram of a known DVB-T modulator design; Figure 2 shows the construction of the inner coder in the modulator of Figure 1; Figure 3 shows the construction of the inner coder of a modulator used in a preferred embodiment of the invention; Figure 4 shows an alternative form for the inner coder of Figure 3; Figure 5 shows a modified form of the inner coder of Figure 3; Figure 6 is a block diagram of a test system embodying the invention; and Figure 7 is a graph of a calibration curve relating the difference between carrier to noise ratio measurements obtained in the system of Figure 6 to the noise floor.

Detailed Description of the Preferred Embodiment Brief reference will first be made to relevant parts of Figure 1, which is a block schematic diagram of a DVB-T modulator 10 taken from ETS 300 744, mentioned above. The modulator 10 uses concatenated coding, with an outer Reed-Solomon coder 12 and an inner convolutional coder 14, and employs quadrature amplitude modulation (QAM) or quaternary phase shift keying (QPSK). After the outer coder 12 there is an outer interleaver 16, and after the inner coder there is an inner interleaver 18. The other elements are described in the ETS standard and a description thereof is not therefore repeated here. The modulation system is an example of a modulation system using forward error correction and soft-decision decoding.

Figure 2, also taken from ETS 300 744, shows in a little more detail the constitution of the inner coder 14. The inner coder comprises a convolutional coder 20 providing X and Y outputs, and a puncturing circuit 22 providing a serial output to the inner interleaver 18.

In accordance with this invention, the inner coder in the circuit of Figure 2 is replaced by the circuit of Figure 3, to provide an artificial difficult channel. The inner coder of Figure 3 includes a convolutional coder 20, which is connected to the output of the outer interleaver of Figure 1. It also includes a puncturing circuit 22 providing a serial output to the inner interleaver 18. In between the convolutional coder 20 and the puncturing circuit 22 are a parallel to serial converter 24 and a serial to parallel converter 26, with an OR-gate 28 between them. The output of the parallel to serial converter is applied through the OR-gate to the serial to parallel converter, but bits received from a switch 30 in Figure 3 are combined by the OR-gate 28 with the output of the parallel to serial converter 24 to modify the bits from the parallel to serial converter before application to the serial to parallel converter 26.

The modulator is in fact arranged to be switchable between the Figure 2 arrangement and the Figure 3 arrangement, so that the modification can be switched in or switched out as desired.

The bits I applied from switch 30 constitute deliberate errors. In the example of Figure 3, the deliberate errors are introduced by setting 1 bit in n bits to be always a '1'. That is, it is set at a predetermined value. Thus, whenever this causes a change in the nth bit, a valid, but incorrect, constellation point is transmitted. This fools the receiver's Viterbi decoder into assigning inappropriately high confidence to the received value. This significantly degrades the performance of the receiver, and hence increases the required carrier to noise ratio for a given output error ratio. The degradation introduced is random, so it should be independent of the design of the demodulator. This allows the value of the noise floor Px to be gauged, by measuring the required carrier to noise ratio (C/N) both with and without the Figure 3 modification to the modulator 10.

In the alternative shown in Figure 4, the OR-gate 28 of Figure 3 is replaced by an AND-gate 28a. In this case the deliberate errors are introduced by setting 1 bit in n bits to be always a '0'. The switch 30 goes to the 0 position for 1 bit in every n.

In a modified arrangement, instead of forcing every nth bit to a specified value, every nth bit can be inverted. This is shown in Figure 5, where the OR-gate of Figure 3 is now replaced by an Exclusive-OR gate 28b. The switch again goes to the 1 position for 1 bit in n. If the bits are forced to a value, as illustrated in Figure 3 or Figure 4, some bits are modified and some are not.

Thus the errors are created at a random set of locations, although only some locations may be modified. With inversion as in Figure 5, an error is always created every 1 in n bits. With 'forcing' the error rate is on average half that for 'inversion', and the locations are to some extent randomised by the data. The calibration curves (see Figure 7) which would be used are different for the two approaches.

The ratio n of erroneously transmitted bits can be chosen to determine the increase in required carrier to noise ratio above the requirement for a Gaussian channel.

In one example, a simulation has indicated that for 64-QAM modulation with a code rate of 2/3, a value of n of 69 might be appropriate, though a lower value may be preferred in practice. Typically, n may desirably lie in the broad range 10 to 1000, and preferably is in the narrower range 30 to 100.

The value of n should be chosen so as not to have undesired effects. Nearby values may have very different effects. For example, since the modification is being made to the bits prior to the puncturer 22, a value of n which is equal to a multiple of 4 (for code rate 2/3) could lead to no detected increase in the required carrier to noise ratio, since the affected bits will simply be punctured.

To make the required measurements, the modified modulator is included in a test arrangement 40 illustrated in Figure 6. The test arrangement includes the modified modulator 10 the output of which is connected to a high quality transmitter 42, indicated as the 'front end' in Figure 1, which itself has a low noise floor and thus introduces insignificant degradation. The output of the transmitter 42 is combined in a combining circuit 44 with the output of a noise source 46, and the resultant applied to a broadcast television receiver 50 under test, the noise floor of which is to be determined. The noise source 46 is capable of proving a variable amount of noise to be added to the signal. The output of the Viterbi decoder in the receiver is applied to a bit-error ratio (BER) meter 48.

The steps involved in making the required measurement of noise floor will now be described.

The system works by causing the operating point of the transmission system in a Gaussian channel to be increased substantially, and to bring it much closer to the system noise floor being measured. It also allows two measurements to be made with one test system, so that the difference can be calculated. Thus it relies on the difference between two values, and not on the absolute magnitude of one value. It is this reliance on the difference between two values which makes it largely independent of the receiver implementation algorithms.

First, the modulator is switched to adopt the Figure 2 arrangement, with none of the bits being modified. The carrier to noise ratio requirement is then determined with the modulator operating normally. This is achieved by increasing the level of noise added by the noise source 46, until the bit-error ratio (BER), after Viterbi decoding in the receiver, and as shown by the BER meter 48, is at the ratio of 2x10-4, which is defined in the ETSI standard referred to above as corresponding to quasi-error-free conditions at the output of the receiver.

The level of added noise required to produce this error ratio is then noted. This then is the carrier to noise requirement assuming that the channel approximates to a Gaussian channel.

The next step is to switch the modulator to its 'test' mode, in which it adopts the arrangement of Figure 3 and inserts deliberate errors in the manner described above. The level of noise inserted by the noise source is adjusted until the quasi-error-free state is again achieved, that is the BER again equals 2x19-4. The level of added noise required to produce this error ratio is again noted.

In the final step the difference between the two levels of added noise is determined. This is the measured increase in carrier to noise ratio for the system under test. This increase will be greater than would have been the case for a very low noise floor system. Hence the measured difference can be used to estimate the noise floor of the system under test, for example by using a calibrated curve. An example of such a curve is given in Figure 7. In the curve of Figure 7, the measured difference A in dB in carrier to noise ratio is plotted against the noise floor value Px. The curve of Figure 7 has been plotted assuming the required carrier to noise ratio for a low noise floor system with an unmodified modulator is 18.5 dB and for the modified modulator is 25 dB. The dotted lines show an example of a measurement where the difference in measured carrier to noise ratio was 11.1 dB. This implies that the noise floor for the system under test is -26.5dBc.

The curve of Figure 7 is derived by computer simulation, or by taking measurements with a low noise floor system. If the modulator is capable of operating with several different values of the deliberate error ratio n, then a series of curves of the type shown in Figure 7 is needed. It may also be necessary to have different curves for Viterbi decoders with different parameters, for example the number of bits of soft decisions, and the maximum path length.

It will be seen that the system operates by transmitting a 'one' when the sequence demands a 'zero' (or vice versa), and the receiver decoder receives this as a 'one' with high confidence.

In a practical arrangement, receivers to be tested will be connected as desired to the system via terminals 52 and 54.

In practice what is actually measured is the overall system noise floor of the system from the modulator to the receiver. If a high quality modulator and transmitter are used, the main degradation is in the receiver, and thus effectively it is the receiver's performance which is measured. If the receiver under test is replaced by a high quality receiver, and the high quality transmitter is replaced by a transmitter to be tested, then the system will effectively measure the transmitterls performance. Similarly the performance of the modulator itself can be checked by using a high quality transmitter and receiver.

The method described produces an artificial difficult channel. This has the advantage that it should be relatively independent of the particular design of the demodulator chip, although it does not precisely reflect any realistic receiving conditions. It does however provide an adequate measure of the receiver noise floor Px(rx) within the range of interest. As noted above it can be extended to measuring the transmitter noise floor PX(tx).

Although in the example described the selective modification to the bits to introduce the deliberate errors takes place at the output of the convolutional coder, there are other places where the transmitted bits could be modified to achieve a similar effect, for example at the output of the puncturer 22 or the output of the interleaver 18. Indeed, it can be arranged that only certain bits of the transmitted constellation points are modified, for instance the most significant bits (MSBs), the lowest significant bits (LSBs), or the central significant bits (CSBs), or some combination of these.

That is to say, only bits of a predetermined level or levels of significance are modified. The effects on performance may be somewhat different. Similarly, if modifications are directly chosen at rates directly related to OFDM symbol or frame rates, other features or artefacts of the system can be tested instead of the noise floor. For example, modifications at the symbol rate would repeatedly affect one or several carriers. This could be used to explore problems associated with transmission or reception of individual carriers or groups of carriers.

The method has been described in the context of the measurement of receiver noise floor in the DVB-T system. It could however be used for the measurement of the noise floor of any part of the DVB-T transmission system, including both the transmitter and the receiver.

It could also be used for measurement of the noise floor in the DVB-S satellite transmission system, and, with appropriate modification of the precise details of the error insertion mechanism, and of the insertion rates and calibration curves, for other digital modulation systems, including the American 8-VSB digital terrestrial television transmission system (ATSC).

The principles can also be applied to other digital modulation systems using forward error correction and soft-decision decoding. That is, valid but erroneous data points are transmitted, i.e. points at valid constellation sites are transmitted, and are subsequently assigned an inappropriately high confidence by the 'demapper', that is the soft decision assigner. This will seriously degrade the performance of the Viterbi or like decoder, increasing the required carrier to noise ratio, and creating what is termed above an artificial difficult channel.

Claims

CLAIMS 1. Apparatus for testing a digital modulation system of the type using forward error correction and soft-decision decoding, the apparatus comprising a digital modulator; a transmitter connected to the output of the digital modulator; a variable noise source; a combiner connected to combine the outputs of the transmitter and the noise source; a receiver connected to the output of the combiner; and a measuring circuit connected to the output of the receiver to determine the bit-error ratio; in which the modulator selectively operates on every nth bit to cause the generation of valid but incorrect transmitted values, and in which one of the modulator. the transmitter and the receiver is under test and the others introduce insignificant degradation.
2. Apparatus according to claim 1, in which the modulator includes concatenated coding circuitry comprising an outer coder and an inner coder.
3. Apparatus according to claim 2, in which the outer coder is a Reed-Solomon coder and the inner coder is a convolutional coder.
4. Apparatus according to claim 3, in which the convolutional coder is a punctured convolutional coder.
5. Apparatus according to claim 4, in which punctured convolutional coder includes a convolutional coding circuit followed by a puncturer, and in which the modulator operates on every nth bit to cause the generation of valid but incorrect transmitted values by operating on the bits at the output of the puncturer.
6. Apparatus according to claim 2, in which the modulator operates on every nth bit to cause the generation of valid but incorrect transmitted values by operating on the bits at the output of the inner coder.
7. Apparatus according to claim 2, in which the inner coder is followed by an interleaver, and in which the modulator operates on every nth bit to cause the generation of valid but incorrect transmitted values by operating on the bits at the output of the interleaver.
8. Apparatus according to claim 1, in which the modulator operates on every nth bit to cause the generation of valid but incorrect transmitted values by forcing every nth bit to take a predetermined value.
9. Apparatus according to claim 1, in which the modulator operates on every nth bit to cause the generation of valid but incorrect transmitted values by inverting every nth bit.
10. Apparatus according to claim 1, in which the modulator modifies only bits of a predetermined level or levels of significance.
11. Apparatus according to claim 1, in which n lies between 10 and 1000.
12. Apparatus according to claim 1, in which n lies between 30 and 100.
13. A method of determining the noise floor of a receiver for use in a digital modulation system of the type using forward error correction and soft-decision decoding, the method comprising the steps of: providing a digital modulator, a transmitter connected to the output of the digital modulator, a variable noise source, a combiner connected to combine the outputs of the transmitter and the noise source, and a measuring circuit for determining bit-error ratio; connecting a receiver to be tested with its input coupled to the output of the combiner and its output coupled to the input of the measuring circuit; applying a signal from the modulator through the transmitter to the receiver with the noise source set at a predetermined value and determining the bit-error ratio; commanding the modulator to operate selectively on every nth bit to cause the generation of valid but incorrect values; again applying a signal from the modulator through the transmitter to the receiver and adjusting the noise source to achieve the same bit-error ratio; determining the difference between the two noise levels in the two applying steps; and determining from the difference the noise floor of the receiver.
14. A method according to claim 13, in which the modulator includes concatenated coding circuitry comprising an outer coder and an inner coder.
15. A method according to claim 14, in which the outer coder is a Reed-Solomon coder and the inner coder is a convolutional coder.
16. A method according to claim 15, in which the convolutional coder is a punctured convolutional coder.
17. A method according to claim 16, in which punctured convolutional coder includes a convolutional coding circuit followed by a puncturer, and in which the modulator operates on every nth bit to cause the generation of valid but incorrect transmitted values by operating on the bits at the output of the puncturer.
18. A method according to claim 14, in which the modulator operates on every nth bit to cause the generation of valid but incorrect transmitted values by operating on the bits at the output of the inner coder.
19. A method according to claim 14, in which the inner coder is followed by an interleaver, and in which the modulator operates on every nth bit to cause the generation of valid but incorrect transmitted values by operating on the bits at the output of the interleaver.
20. A method according to claim 13, in which the modulator operates on every nth bit to cause the generation of valid but incorrect transmitted values by forcing every nth bit to take a predetermined value.
21. A method according to claim 13, in which the modulator operates on every nth bit to cause the generation of valid but incorrect transmitted values by inverting every nth bit.
22. A method according to claim 13, in which the modulator modifies only bits of a predetermined level or levels of significance.
23. A method according to claim 13, in which n lies between 10 and 1000.
24. A method according to claim 13, in which n lies between 30 and 100.
25. A method according to claim 13, in which the modulator uses an OFDM system with a defined symbol rate, and the modifications are made at a rate directly related to the symbol rate.
26. A method according to claim 13, in which the receiver is a digital broadcast television receiver.
27. Apparatus for testing a receiver for a digital modulation system of the type using forward error correction and soft-decision decoding, the apparatus comprising a digital modulator; a transmitter connected to the output of the digital modulator; a variable noise source; a combiner connected to combine the outputs of the transmitter and the noise source; a measuring circuit connected to the output of the receiver to determine the bit-error ratio; and terminals connected to the output of the combiner and to the input of the measuring circuit for connection to a receiver to be tested; and in which the modulator selectively operates on every nth bit to cause the generation of valid but incorrect transmitted values.
28. Apparatus for testing a digital modulation system, substantially as herein described with reference to Figure 3 et seq. of the drawings.
29. A method of determining the noise floor of a receiver, substantially as herein described with reference to Figure 3 et seq. of the drawings.