The invention moreover relates to a system as defined in the introductory portion of claim 5 for the transmission of WDM data signals, said system comprising at least one WDM multiplexer and at least one WDM demultiplexer con- nected thereto via an optical signal path, the optical inputs of the WDM multiplexer being fed by E/O convert- ers, such as lasers, which each generate a channel-spe- cific wavelength, said E/O converters being modulated by data signals.

Finally, the invention also relates to an AM modulation unit for a WDM system as defined in claim 9.

In connection with the ever increasing use of optical transmission fibres for data communication/ telecommunication there is an ever greater need for the development of an increased transmission capacity. This development means that it becomes increasingly difficult for the existing cable connections to offer the capacity which the market can consume. The development has not become less pronounced with the growing popularity and use of e. g. the internet. Another example is the increased transmission need in video transfer via data signals on the telecommunications network.

A consequence of this development is of course that the transmission capacity in the existing cable networks has to be extended and/or made more effective. As extension of the optical cable network is an extremely expensive operation, it is attempted to optimize the use of the ex- isting cable network to the greatest extent possible without laying further cables, when this can be avoided.

This may e. g. be done by modifying or replacing the transmission hardware used. In spite of the implementa- tion costs involved by this, the costs will only be a fraction of the costs which supplementary laying of new cable connections would involve.

Two concrete examples of this modification are known from TDM, Time Division Multiplexing, and WDM, Wavelength Di- vision Multiplexing, whereby the data capacity may be in- creased considerably. The most important difference be- tween the two types of transmission of TDM may be consid- ered as a purely sequential transmission protocol, where a specific range of the data sequence is allocated to each transmission channel, while WDM may rather be re- garded as a parallel and mutually independent transmis- sion of several optical channels in the same fibre, a specific wavelength being allocated to each channel.

By nature, a drawback of TDM is that it basically serves to increase the data channels at disposa, but does not increase the overall transmission capacity without in- creased transmission rate. WDM, however, provides both an increase in the data channels at disposa and a corre- sponding increase in the transmission capacity. In prac- tice, the transmission wavelengths used will be located within a relatively small optical window with small wave- length intervals, which together are within the opera- tional range of the optical amplifiers. Thus, it is abun- dantly clear that the data capacity of the overall

transmission system is increased very strongly for each added WDM channel.

The number of these channels thus increases greatly, and the 4-16 channels used at present are expected to be ex- tended currently in the years to come in step with the development of the hardware components necessary for the purpose. This particularly applies to the necessary opti- cal filters and laser components.

However, a problem involved by the use of WDM is that the individual channels at the receiver side are to be branche again to separate signals after transmission.

This problem is particularly pronounced at the receiver side where the individual demultiplexed signals, although they might be found, cannot readily be identifie unambi- guously under all conditions, as they usually do not have channel identification in the data frame for WDM channel identification. Further, it is difficult to decide where a channel has dropped out or is just passive for a period of time.

The invention When each data signal (D1-Dn) at the transmitter side is AM modulated with at least one LF pilot signal (fol, f2,.. fn) which is specific for the data signal concerne, it is possible to perform a simple unambiguous on/off de- tection of the WDM channels at the receiver side.

A further avantage of the invention is that the overall transmission system may be varied dynamically, so that given transmission channels do not necessarily have to be located to a specific filter in the multiplexer. A poss- ible change in wavelengths in the WDM channels at the transmitter side will thus only cause the tuneable fil-

ters at the receiver side to tune into the new or changed WDM channels automatically or under control.

When the data signals (D1-Dn) are AM modulated with the pilot signal (Ps-Pn) before the E/O conversion, an advan- tageous embodiment of the invention is achieved, as the electrical modulation of the data signal may be obtained in a simple manner by modulation with electrical pilot signal generators.

When each data signal is AM modulated with a combination of at least two LF pilot tone signals, a particularly ad- vantageous embodiment of the invention is achieved, as each data signal may be coded unambiguously in such a manner that the total number of employed pilot signals of different frequency is reduced. An essential avantage of this coding is thus that the multiplexed optical WDM sig- nal has a limited number of different pilot signal compo- nents. This circumstance becomes extremely pronounced with an increase in the channels of the WDM system, mak- ing it necessary to extend the bandwidth of the WDM sig- nal at disposa for the desired LF components. As pilot signals should be kept below a limit frequency of about 50 kHz to the greatest possible extent in order not to turn into noise in the LF components of the WDM signal, the present embodiment will thus allow marking of more optical WDM channels than would otherwise have been poss- ible. It should be noted in this connection that the more closely spaced the individual pilot frequencies are, the more narrow-banded the filters must be at the receiver side. It should additionally be mentioned that the inven- tion is also particularly advantageous in connection with the identification of the pilot signals at the receiver side, as the bandwidth of the filter to be used in order to filter the pilot signals must be relatively great.

This means in turn that the time constant of the filters

may be minimized, thereby allowing the detection response time to be minimized at the receiver side.

It will thus be appreciated that the avantages of the invention become more pronounced with a growing number of WDM channels.

When the LF pilot tone combination comprises r+1 mutually different pilot signals for a 2r channel WDM system, in which all the frequencies of the pilot signals used are below 50 kHz, a particularly advantageous embodiment of the invention is achieved, as n+1 different pilot signals are sufficient for unambiguously modulating all the chan- nels in a 2-channe WDM system. It will likewise be appreciated that the invention is particularly advan- tageous with an increasing number of channels. For example, according to the invention, a 256 channel WDM system may be coded with just nine different pilot sig- nals, all of which can be detected using relatively inex- pensive electrical filters with a small time constant.

Scanning in the demultiplexer over a given optical domain may thus be performed extremely quickly according to the invention.

It should moreover be noted that r+1 coding of the WDM channels ensures that all WDM channels are marked and can therefore be recognized positively.

When the demultiplexer is provided with tuneable BP fil- ters, which each are controlled by a control unit in re- sponse to the emitted pilot tone content on the outputs of the O/E converters, so that each tuneable filter is tuned into a well-defined and identifie WDM channe, a particularly effective control and monitoring of the in- dividual WDM channels is achieved.

If some channels might have dropped out, either because of a technical error or because the system is e. g. in a low load period, the control unit can positively detect how many and specifically which channels are actually present on the output of the demultiplexer, and the con- trol unit can record on the basis of this information how many and which channels are not present on the demulti- plexer.

Accordingly, it will be possible to allow the tuneable filters in the demultiplexer to scan the fed light spec- trum on the input of the demultiplexer under control of the control unit. This scanning is thus possible with a specific marking of each channe, as, otherwise, the con- trol unit cannot record the WDM channel into which the filter has actually been tuned.

Drawing The invention will be described below with reference to the drawing, in which fig. 1 shows an example of a WDM system according to the invention, fig. 2 shows an example of the spectral distribution of the pilot tones used according to the invention, fig. 3 shows an example of how WDM channels are distrib- uted relative to wavelength, fig. 4 shows an example of a preferred embodiment of the invention, and fig. 5 schematically shows how a 16 channel WDM may be AM modulated by just five pilot tones.

Example Fig. 1 shows an illustration of an embodiment of the in- vention.

The four channel WDM system shown comprises a WDM multi- plexer 1 which is optically connecte to a WDM demulti- plexer 3 by means of an optical fibre 2.

The system comprises four summation units 14,15,16 and 17 which sum up four data signals D1, D2, D3 and D4 and a corresponding channel-specific pilot signal pl, p2, p3 and p4.

Subsequently, the amplitude modulated electrical signals are fed to the E/O converters 10,11,12 and 13, i. e. electrically modulated lasers, which are electrically connecte to the summation units 14,15,16 and 17. Each laser is thus modulated with signals D1+p1, D2+p2, D3+p3 and D4+p4.

Each data channel D1, D2, D3 and D4 is modulated by the E/O converters at channel-specific wavelengths. The wave- length X for each channel will typically be within a win- dow of 1550 nm 10 nm for standardized systems. A chan- nel position may e. g. be ch1-1544 nm, ch2-2547 nm, ch3 = 1553 nm, ch4 = 1556 nm. The pilot signal frequency may e. g. be selected to be 10 kHz, 12.5 kHz and 17.5 kHz for each channel.

The multiplexed pilot signal modulated signals are then transmitted to the WDM demultiplexer 3 via the optical transmission path 2 in which the signal is split into four separate optical signals and fed to four 0/E con- verters 20,22,24 and 26 by means of tuneable optical BP

filters not shown tuneable in the demultiplexer 3. The O/E converters 20,22,24 and 26 will typically be formed byphotodiodes.

The electrical output from the O/E converters 20,22,24 and 26 is subsequently split into an HF and an LF compo- nent by means of an electrical, tuneable LP/BP filter ar- rangement 21,23,25 and 27 arrange for each channel.

The LF component, which contains the channel-specific modulation signal on the data signal, is subsequently fed back electrically to a control unit 4, CU, for each channel, said control unit controlling the tuneable fil- ters in the demultiplexer 3 on the basis of the informa- tion achieved. The HF components remaining after the LP filtering then constitute the original data signals D1, D2, D3 and D4.

In practice, the splitting of the LF and HF components may take place in several ways known per se within the scope of the invention.

The HF signals wavelength filtered in the demultiplexer 3, i. e. D1, D2, D3 and D4 each having a channel-specific identifier pl, p2, p3, p4 modulated thereon, can thus be recognized unambiguously by means of the identifier pl, p2, p3 and p4, and the tuneable filters in the demulti- plexer 3 can therefore scan the added light spectrum on the input of the demultiplexer under control of the con- trol unit 4. This scanning is thus just possible with a specific marking of each channe, as the control unit cannot record either into which WDM channel the filter has actually been tuned.

If a tuneable filter has thus been tuned into an already found WDM channel, the control unit 4 will record this coincidence and ensure that the filter continues to scan

until a channel not previously"locked"has been found.

Further, the control unit 4 will be able to record when all the searched channels have been found, and then the scanning is interrupted.

Conversely, the coding technique according to the inven- tion ensures that the output of the demultiplexer, or more specifically the output of the O/E converters has full information on which channels are present and on which output D1, D2, D3, D4 these are placed.

It will be appreciated that D1, D2, D3 and D4 will not necessarily or do not have to emerge from the O/E con- verters 20,22,24 and 26 in the well-arranged order which is outlined in the figure. If a concrete output layout should be desired according to the invention, it is possible according to the invention to adapt the con- trol unit 4 for this purpose.

A distinct avantage of marking each wavelength with an LF signal is that each signal may be identifie unambigu- ously using relatively inexpensive and non-bulky LF fil- ters.

Finally, it should be noted that the example shown is monodirectional. It will be appreciated, however, that the invention may also be implemented in a bidirectional WDM system.

Fig. 2 shows an illustration of how the components for the coding of a data channel may be constructed.

For one thing, LF components are used, which should typi- cally be selected to be below an upper limit frequency of 50 kHz, thereby ensuring that the pilot signals do not

interfere with the HF signal and thereby impair the transmission quality.

The example shown uses five different pilot signals Psl, Ps2, Ps3, Ps4 and Ps5, as the signals shown may be used for complete multi pilot signal coding of a sixteen chan- nel WDM system. The coding technique used will be de- scribed more fully with reference to fig. 5. It will be appreciated that more or fewer mutually different pilot signals may be used depending on the coding technique and the number of WDM channels.

Psl, Ps2, Ps3, Ps4 and Ps5 may e. g. be selected to be 10 kHz, 12.5 kHz, 15 kHz, 17.5 kHz and 20 kHz, respectively.

If the pilot signal laying was to correspond to the example shown in fig. 1, the selected pilot signals would e. g. be Psl, Ps2, Ps3 and Ps4.

It should be stresse that the figure just serves to illustrate the invention, which applies e. g. to the size of amplitude ratios. It will moreover be appreciated that the illustrated frequency intervals are just to serve to illustrate the mutual spacing between pilot signals and the frequency content of the data signal, respectively.

It will be seen from fig. 3 how the transmission channels may be distributed over the wavelengths #1, #2, #3, #4 and up to kn. In practice, all the channels will be dis- tributed over a window of 1550 nm 10 nm. In systems us- ing a relatively large number of WDM transmission chan- nels, the wavelength intervals are thus very small.

Fig. 4 shows an additional and preferred embodiment of the invention in which a plurality of pilot channels are coded on the data channel.

As appears from fig. 2, the number of the necessary pilot tones according to fig. 1 will quickly become very large corresponding to the current number of channels. This should be seen in relation to the circumstance that the pilot signals should be below 50 kHz in order not to in- terfere with the data traffic in the WDM channel.

Such a problem will be eliminated in practice in fig. 4, since, according to this embodiment, each channel is coded with a combination of pilot signals Sp, which are subsequently split and identifie unambiguously as a channel marking at the receiver side.

The shown example comprises n-channe WDM data signals which each are modulated in the electrical domain with a combination of pilot signals Sp. This combination should be regarde as a purely binary coding of each data signal by means of LF pilot signals.

The system thus inclues n summation units 44,45,46 and 47 which sum up n data signals Dl-Dn and a corresponding channel-specific pilot signal of combinations Of P1-Pi, where"i"corresponds to the number of mutually different LF pilot signals.

The actual coding of the data signal will be described more fully with reference to fig. 5.

Subsequently, the summed electrical signals are fed to E/0 converters 40,41,42 and 43, i. e. electrically modu- lated lasers, which are electrically connecte to the summation units 44,45,46 and 47.

The wavelength X for each channel will typically be within a window of 1550 nm 10 nm in standardized sys- tems.

The multiplexed pilot signal modulated signals are then transmitted to the WDM demultiplexer 33 via the optical transmission path 32, in which the signal is split to n separate optical signals and is fed to n O/E converters 50,52,54 and 56 by means of tuneable optical BP filters not shown in the demultiplexer 33. The O/E converters 50, 52,54 and 56 will typically be formed by photodiodes.

The electrical output from the O/E converters 50,52,54 and 56 is subsequently split into an HF component and an LF component by means of an electrical, tuneable LP/BP filter arrangement 51,53,55 and 57 provided for each channel. The LF component containing the channel-specific modulation signal on the data signal is then fed back electrically for each channel to a control unit 4, CU, which controls the tuneable filters in the demultiplexer 3 on the basis of the information achieved. The HF compo- nents remaining after the LP filtering then constitute the original data signals D1-Dn.

The HF signals BP filtered in the demultiplexer 33, i. e.

Du-don each having a channel-specific identifier pl, p2, p3,... pi modulated thereon, can thus be recognized un- ambiguously by means of the identifier formed by the pi- lot signals, and the tuneable filters in the demulti- plexer 33 can therefore scan the added light spectrum on the input of the demultiplexer under control of the con- trol unit 34. This scanning is thus only possible with a specific marking of each channe, as, otherwise, the con- trol unit will not be able to record into which WDM chan- nel the filter has actually been tuned. Thus, the control unit 34 will ensure with simple control algorithms that

the tuneable filters in-té demultiplexer 33 track all the various channels which may be found on the input of the multiplexer, i. e. up to n channels.

If some channels might have dropped out, either because of a technical error or because the system is e. g. in a low load period, the control unit can positively detect how many and specifically which channels are actually present on the output of the demultiplexer 33, and, on the basis of this information, the control unit 34 can also record how many and which channels are not present on the demultiplexer.

Finally, it should be mentioned that the invention is not restricted to monodirectional WDM signals.

Fig. 5 shows a table of an example of a coding of a 16- channel WDM system.

The example shown uses five pilot tones with the frequen- cies fl, f2, f3, f4 and fr which together may be modulated on the individual data channels at the transmitter side so that these may subsequently be found and recognized at the receiver side.