WO2012051668A1

WO2012051668A1 - Bispectrum method and apparatus for recovery of optically transmitted signals

Info

Publication number: WO2012051668A1
Application number: PCT/AU2011/001345
Authority: WO
Inventors: Le Nguyen Binh; Steve Mutabazi; Nhan Duc Nguyen
Original assignee: Ausanda Communications Pty Ltd
Priority date: 2010-10-22
Filing date: 2011-10-21
Publication date: 2012-04-26

Abstract

The current invention is in relation to optical fibre wavelength division multiplexing (WDM) communication systems and describes a method and apparatus to enable highly sensitive recovery of signals with embedded detection of transmission impairments for signals received over channels of wavelength division multiplexing (WDM) communication systems in which the signals are carried in the form of modulated optical carrier signals comprising N-bits per symbol, where N=1, 2, 3, 4 or greater.

Description

BISPECTRUM METHOD AND APPARATUS FOR RECOVERY OF OPTICALLY TRANSMITTED SIGNALS

FIELD OF THE INVENTION

The present invention relates to optical fibre wavelength division multiplexing (WDM) communication systems and more particularly it relates to a method and apparatus to enable highly sensitive recovery of signals with embedded detection of transmission impairments for signals received over channels of wavelength division multiplexing (WDM) communication systems in which the signals are carried in the form of modulated optical carrier signals comprising N-bits per symbol, where N=1, 2, 3, 4 or greater.

DESCRIPTION OF PRIOR ART

The use of optical transmission, in which an information signal is modulated onto an optical carrier, is widely employed in modern communication systems. In particular, many long-haul transmission networks employ single-mode optical fibres for the transmission of digital information at high bit rates, using one or more optical carriers, or wavelengths, over each fibre. There have been some recent advances in optical technologies, however the rate at which optical information can be received and decoded is seen as one of the significant limiting factors in the development of efficient, effective high-speed data streams. Additionally, due to the complexity of fibre optical networks and the increasing reliance on their functioning at optimal levels, at present there is not an easy way of detection and recovery of optically transmitted signals wherein the signal provides data on the health of the optical transmission path thereby assessing the health of any given WDM communication system. OBJECTIVE OF THE INVENTION

It is an object of the present invention to provide a method and apparatus for highly sensitive detection and recovery of optically transmitted signals. A further object of the invention is to provide a method and apparatus for highly sensitive detection and recovery of optically transmitted signals wherein the received signal provides data on the health of the optical transmission path.

Other objects and advantages of the present invention will become apparent from the following description, taking in connection with the accompanying drawings, wherein, by way of illustration and example, an embodiment of the present invention is disclosed.

SUMMARY OF THE INVENTION

According to the present invention, although this should not be seen as limiting the invention in any way, there is provided a method and apparatus for highly sensitive detection and recovery of optically transmitted signals, the method including the step of: a) Receiving an optically transmitted signal in the form of a modulated

optical carrier signal comprising N-bits per symbol, where N≥1 [that is 1 , 2, 3, 4 or greater] and where each of the two polarization modes of the carrier is used to carry its own set of symbols of the same bjt capacity; b) Separating the received signal via a polarization de-multiplexer (4500) into at least a first separated signal (4030) and a second separated signal (4040);

c) Separating the first separated signal (4030) and a second separated signal (4040) into a set of parallel optical signals [via a optical preprocessor] Si, S2, ... Sj, where J = 2^N [comprising the first signal Si 5410, the second signal S₂ 5420 up to the J^,h signal Sj 5499];

d) Passing each parallel optical signal Si, S2, ... Sj through a coherent receiver (4032), each parallel optical signal Si, S₂, ... Sj thus generating a pair of output electrical signals being the in-phase signal (I) and a quadrature signal (Q) (Sn/Si_Q, S₂|/S_2Q, .. . SJI/SJQ) (5210, 5220, ... 5299); e) Digitizing each of the pair of output electrical signals, being the in-phase signal (I) and an quadrature signal (Q) (S11/S1Q, S₂I/S₂Q, ... SVSJQ), [analogue to digital converter] into a stream of digital time sequenced sample sets (4033)

f) Passing the time sequenced sample sets (4033) into a bi-spectrum

processing engine (7000) wherein the time sequenced sample sets (4033) are separated into:

i) Symbol region data (9220); and

ii) Optical health transmission data (9250);

g) Wherein the Optical health transmission data is data outside of the

Symbol region data. BRIEF DESCRIPTION OF DRAWINGS

Figure 1(a) is a block diagram depicting the functions of the Receiver 9200 as an exemplary embodiment of the present invention.

Figure 1(b) is a block diagram depicting the functions of a transceiver 9000 in which an appropriate transmitter 9100 is deployed as a sub-system in conjunction with the receiver 9200 according to the present invention to form the transceiver system 9000.

Figure 2(a) is a phase state constellation diagram showing symbols encoded with 1 -bits per symbol.

Figure 2(b) is a phase state constellation diagram showing symbols encoded with 2-bits per symbol.

Figure 2(c) is a phase state constellation diagram showing symbols encoded with 3-bits per symbol.

Figure 2(d) is a phase state constellation diagram showing symbols encoded with 4-bits per symbol in which symbols are generated as a result of application of one bias voltage.

Figure 2(e) is a phase state constellation diagram showing symbols encoded with 4-bits per symbol which includes alternating symbols as a result of the alternating application of two bias voltages where one symbol is generated by application of the first bias voltage and the next neighbouring symbol is generated by application of the second bias voltage.

Figure 3 is a block diagram depicting the functions of the receiver structure employed as an exemplary embodiment of the present invention Figure 4 is a block diagram depicting in more detail the key sub-structures necessary in the construction of the receiver 9200 as an exemplary embodiment of the present invention. Depicted are the functions for handling one of the two signals carried over the two polarization states of the optical carrier. Figure 5 is a block diagram depicting the processing sequences of the bispectrum processing engine 7000.

Figure 6(a) is an illustration of how according to the present invention the optical pre-processor 5100 may be constructed in its first form.

Figure 6(b) is an illustration of how according to the present invention the optical pre-processor 5100 may be constructed in its second form.

Figure 7 is a graphical illustration of the use of progressively delayed versions of the same signal to achieve multiple times more sample per sample period that otherwise possible at the available sampling speed of the ADC.

Figure 8 illustrates the comparative performance of the bispectrum receiver 9200 against the standard receiver both handling QPSK modulated signals operating at 20 G symbols per second.

TABLE 1 illustrates the generating of the triple correlation terms used in the construction of the bispectrum data set. It is based on an example where 4 samples per symbol are used and shows how the triple products are calculated and how the triple correlation terms are obtained as the summations of the row values of the triple product terms.

DETAILED DESCRIPTION OF THE INVENTION Figure 1(a) is a block diagram depicting the functions of the Receiver 9200 as an exemplary embodiment of the present invention.

In this general embodiment, the client side interface 9220 of the receiver 9200 comprises a number of parallel lanes over which data is delivered including but not limited to Y parallel lanes 9220, where Y is an integer number equal to or greater than 1 , and where these Y lanes 9220 deliver data bits in parallel according to a common clock controlling the flow of data bits across the Y lanes 9220. In addition the client side interface comprises other parallel lanes 9250 over which the receiver delivers its assessment of the health of the optical transmission path for the signal prior to its reception by the receiver.

Also in this general embodiment of present invention the optical line side interface of the receiver 9200 comprises K wavelengths 9230 where K is an integer number equal to or greater than 1.

Figure 1 (b) is a block diagram depicting the functions of a transceiver 9000 in which an appropriate transmitter 9100 is deployed as a sub-system in conjunction with the receiver 9200 according to the present invention to form the transceiver system 9000.

The transceiver 9000 supports an electrical interface such as set out in the Optical Interworking Forum (OFI) multi-supplier agreement (MSA) interface definition for 10 Gbps, 40 Gbps and 100 Gbps This electrical interface accepts data over Y lanes 9120 in accordance with the MSA specifications. Subsequent to processing by the transmitter 9110 the aggregated input data is then transmitted over the host WDM system using the first wavelength 9131 and as required the second wavelength 9132, up to K wavelengths. The electrical interface of the transceiver 9000 also delivers data over Y lanes 9220 of the receiver sub-system 9200 in accordance with the MSA specifications, having received the data from the host WDM system using the first wavelength 9230 and as required the second wavelength 9232, up to K wavelengths. During the processing in accordance with the present invention of the data symbols received over in-coming wavelengths 9230 the receiver sub-system 9200 according to the present invention compiles information characterizing the level and trend of impairment noise observed during a prescribed period of transmission thereby providing continuous transmission performance information 9250.

Figure 2(a) is a phase state constellation diagram 2502 for a signal received over a channel of a wavelength division multiplexing (WDM) communication system in which the signal has been carried in the form of modulated optical carrier signals comprising symbols encoded with 1 -bits per symbol. Depicted are the symbol states So 2200 and Si 2210 representing the two possible values of the one bit received per symbol.

Figure 2(b) is a phase state constellation diagram 2503 for a signal received over a channel of a wavelength division multiplexing (WDM) communication system in which the signal has been carried in the form of modulated optical carrier signals comprising symbols encoded with 2-bits per symbol. Depicted are the symbol states S₀ 2300, Si 2310, S₂ 2320, S₃ 2330 representing the four possible combinations of the two bits received per symbol and where all the symbols are of the same amplitude 2701 achieved with bias voltages of the same amplitude 2700.

Figure 2(c) is a phase state constellation diagram 2504 for a signal received over a channel of a wavelength division multiplexing (WDM) communication system in which the signal has been carried in the form of modulated optical carrier signals comprising symbols encoded with 3-bits per symbol. Depicted are the symbol states So 2400, Si 2410, S₂ 2420, S₃ 2430 S₄ 2440, S₅ 2450, S₆ 2460, S₇ 2470 representing the eight possible combinations of the three bits received per symbol and where all the symbols are of the same amplitude 2701 achieved with bias voltages of the same amplitude 2700.

Figure 2(d) is a phase state constellation diagram 2505 for a signal received over a channel of a wavelength division multiplexing (WDM) communication system in which the signal has been carried in the form of modulated optical carrier signals comprising symbols encoded with 4-bits per symbol. Depicted are the symbol states S₀ 2600, Si 2605, S₂ 2610, S₃ 2615 and so on to Si₅ 2675 representing the sixteen possible combinations of the four bits received per symbol and where all the symbols are of the same amplitude 2701 achieved with bias voltages of the same amplitude 2700.

Figure 2(e) is a phase state constellation diagram 2506 for a signal received over a channel of a wavelength division multiplexing (WDM) communication system in which the signal has been carried in the form of modulated optical carrier signals comprising symbols encoded with 4-bits per symbol. Depicted are the symbol states So 2600, Si 2605, S₂ 2610, S₃ 2615 and so on to Si₅ 2675 representing the sixteen possible combinations of the four bits received per symbol and where alternating symbols are of one amplitude 2701 achieved with bias voltages of the same amplitude 2700 while the alternate symbols are of a higher amplitude 2703 achieved with correspondingly higher bias voltage amplitude 2702.

Figure 3 is a block diagram depicting the functions of the receiver structure employed as an exemplary embodiment of the present invention, in the detection and recovery of a signal 9231 received over one channel of a wavelength division multiplexing (WDM) communication system in which the signal Θ231 may have been carried in the form of a modulated optical carrier signal comprising N-bits per symbol, where N=1 , 2, 3, 4 or greater, and where each of the two polarization modes of the carrier is used to carry its own set of symbols of the same bit capacity.

In accord with general practice such a signal 9231 may in the first instance be amplified through the optical amplifier 4010 to generate the amplified signal 4020 suitable for subsequent detection and recovery. The amplified signal 4020 is then optically separated by the polarization de-multiplexer 4500 into components 4030 and 4040 each carrying its own set of symbols. According to the present invention each of the components 4030 and 4040 are first processed through dedicated but otherwise identical sub-structures 4600 and 4610 to produce a parallel set of output signals 4032 and 4042 respectively. Each of the sub-structures 4600 and 4610 is equipped with an optical pre- processor 5100 constructed in accordance with the present invention. The optical pre-processor 5100 processes its respective in-coming optical signal component 4030 or 4040 and in accordance with the present invention generates a set of parallel optical signals 4031 or 4041. This set of parallel optical signals 4031 or 4041 is then optically demodulated and converted into corresponding streams of electrical signals 4032 and 4042 by the respective coherent receiver structures 5200 and 52 0.

The parallel streams of electrical signals 4032 and 4042 are then digitized into digital streams 4033 and 4043 through the respective analogue-to-digital converter (ADC) structures 5400 and 5410. The digital streams 4033 and 4043 are passed through processing steps constructed and operated in accordance with the present invention and herein referred to as the bi-spectrum processing engine 7000, to deliver the continuous flow of received data 9220 and optical transmission impairments data generated by the bi-spectrum processing engine 7000 in accordance with present invention.

Figure 4 is a block diagram depicting in more detail the key sub-structures necessary in the construction of the receiver 9200 as an exemplary embodiment of the present invention. For clarity other functions commonly included in the construction of a receiver system are not shown. Upon receiving the in-coming optical signal 4030 the optical pre-processor 5100 generates a set of output optical signals 4031 comprising the first signal Si 5410, the second signal S2 5420 up to the J^th signal Sj 5499. These signals then become the inputs to respective coherent receivers within the coherent receiver structure 5200. In accordance with the well established operating processes of coherent receivers each coherent receiver generates a pair of output electrical signals comprising in-phase signals such as Sn 5210 with its corresponding quadrature signal SIQ 5220, phasor combinations of which provide the magnitude and phase information characterizing the optical signals Si 5410, S₂5420 up to Sj 5499. All the pairs of output electrical signals from the receiver structure 4600 are analogue. They are therefore subsequently digitized by the analogue-to-digital converter (ADC) structure 5400 into streams of time sequenced sample sets, one for each of the incoming electrical signal. All the time sequenced sample sets 4033 and 4043 corresponding to the original optical signals 4030 and 4040 respectively form the complete input to the bi-spectrum generator 5600, being the first processing stage within the bi-spectrum processing engine 7000.

The operation of the bi-spectrum processing engine 7000 will now be described with reference to Figure 5. In the first instance the bispectrum generator 5600 acts as the receptor for the time sequenced sample sets 4033 and 4043 emanating from the ADC structure 5400. In this description reference will be confined to sample sets 4033 however sample sets 4043 are treated similarly according to the present invention.

Ordinarily in accord with established practices sample sets 4033 may be directly processed for instance using the standard symbol recovery 5603 which may employ a range well known processing techniques such as least squares estimation to recover the symbols carried within the received signal. However in order to achieve a significantly higher level of accuracy in recovering symbols encoded with multiple bits (2-bits, 3-bits, 4-bits and so on) and transmitted at symbol rates equal to or greater than 10 Giga symbols per second, then innovative steps as set out in accordance with the present invention are required.

The innovative steps according to the present invention include but are not limited to: a) Passing of the sample sets 4033 through a fast Fourier transform generator 5630 to generate one-dimensional frequency domain spectrum matrices 5655 corresponding to the sample sets 4033;

b) Optionally passing the spectrum matrices 5655 through the spectrum symbol recovery step 5701 within the symbol recovery function 5700 to recover the symbols 9220;

c) Optionally passing the spectrum matrices 5655 through the bispectrum converter 5640 and thereby generating the two dimensional bispectrum matrices 5660; d) Optionally passing the sample sets 4033 through the triple correlation step 5610 to generate the triple product matrices 5650 and then converting the triple product matrices 5650 through the two dimensional fast Fourier transform 5620 into the bispectrum matrices 5660;

e) Passing the bispectrum matrices 5660 through the bispectrum symbol recovery step to obtain the received symbols 9220;

0 Passing the bispectrum matrices 5660 through the optical transmission impairments detection step 5800 to recover a profile over time detected impairments 9250. Upon adopting the option to use the bispectrum matrices 5660 in the processing of the sample sets 4033 a multiplicity of processing sequences may be applied to achieve this objective. Only two such processes are illustrated here.

The first processing sequence 5601 comprises the triple product generation processing sequence 5610 which transforms the time sequenced sample sets 4033 and 4043 into the structured triple product matrix 5650 which in turn is transformed by the two-dimensional discrete fast Fourier transform (FFT) processing sequence 5620, into the desired bispectrum matrix 5660. The triple product of a signal s(t) is defined as:

In this definition r, and r₂ are time delays appropriately selected in accordance with the fixed number of samples n generated by the ADC structure 5400 during the symbol period of the original optical signal 4030, depicted in Figure 4. For example where 4 sample defined by the set {di , d_2> d₃, d₄} are generated by the ADC structure 5400 per symbol period, there are n-1=3 delay intervals. The triple correlation processing sequence consists of first generating all possible triple product terms for all possible combinations of time delays r, and τ. in circular progression through the delay interval around the symbol period.

The next step towards generating the bispectrum is the derivation of the triple correlation terms from the triple product terms. Given the triple product of a signal as defined above, the triple correlation of the signal is defined as the correlation between the signal s(t) with two delayed versions of the signal, mathematically set out as:

S(T,,T₂) = JsiMt + Tjsit - rJdt In the digital domain and based on the circular generation of triple product terms, triple correlation terms are generated for all possible combinations of the time delays τ, and r₂ through the summations of the triple products along the rows of the triple product matrix. The resulting autocorrelation terms generated for this current example are shown in Table 1. Each of the summation terms then becomes an entry in a two dimensional triple correlation matrix with rows and columns designated by the time delays τ, and r₂ for a finite integer number of samples n per symbol period.

The final step of this processing sequence 5601 is the two-dimensional Fast Fourier transformation (FFT) of the values of the triple correlation matrix 5650 into corresponding frequency domain values of the bispectrum matrix 5660. Upon transformation into the frequency domain the rows and columns of the matrix transform to and h with integer values respectively corresponding to time T, andr₂.

The second processing sequence 5602 for transforming the time sequenced sample sets 4033 and 4043 into the bispectrum matrix 5660 transforms the sample data sequences through a one dimensional FFT sequence 5630 into a one dimensional spectrum data matrix 5655. The bispectrum generation sequence 5640 then derives the desired bispectrum matrix B(fi,f₂) 5660 based on the following formula: fi(/„/₂) - FifJFifJF'^ + f₂)

Here we note that fi and f₂ correspond to the time domain time delays r, and τ_¾

Regardless of the processing sequence used to generate the bispectrum matrix 5660 the first overriding objective of the present invention is to transform the digitized form of the optical signal 4030 into the data set contained in the bispectrum matrix, thereby enabling significantly higher level of accuracy than achieved to date in the recovery of data symbols carried in a highly noisy optical signal 4030. The second overriding objective of the present invention is to exploit the data set contained in the bispectrum matrix generated according to the present invention and the effects of optical transmission impairments on the transmitted data symbols. In so doing the receiver 9200 according to the present invention is able to provide for the first time comprehensive optical transmission performance diagnostic information from the receiver of the transmission system. The common ways of obtaining optical transmission with a similar level of comprehensive diagnostic information involve invasive monitoring using specialized monitoring systems external to the transmission system.

This way of transmission impairment monitoring is inherently limited in scope to only a few points in the transmission network and is inherently cost inefficient to deploy. Encapsulation of impairment monitoring capability into the receiver system as made possible by the present invention enable ubiquitous and cost efficient monitoring of impairments within optical transmission network.

Processing sequences applied on the contents of the bispectrum matrix 5660 will now be described. As indicated earlier the two objectives of the processing of the bispectrum data set is first to recover the continuous stream of data symbols and secondly to detect the effects of optical transmission impairments.

The data symbol recovery sequence 5700 may be constructed based on well known methods including but not limited to the least squares method, and linear programming.

It is the express objective of the present invention to attain comparatively superior bit error rate (BER) performance for any given signal to noise ratio (SNR) at the input of the receiver 9200 so as to enable construction of optical transmission systems employing higher order optical modulation schemes capable of 2 bits per symbol, 3 bits per symbol, 4 bits per symbol, and greater. For example as illustrated in Figure 8, the bispectrum receiver 9200 according to the present invention exhibits superior BER/SNR performance for QPSK modulation compared with the current standard method employed in the reception of such modulated signals. The comparative standard method of symbol recover employs the least squares method directly to the signal samples 4033 as depicted in Figure 5, while the bispectrum method according to the present invention shows equally superior performance results for both processing sequences 5601 and 5602.

As can be seen from Figure 8, the bispectrum receiver 9200 requires a lower SNR to achieve the same BER as the standard receiver. Also for increased SNR levels the comparative performance differential increases in favour of the bispectrum receiver 9200.

A further key distinctive capability of the bispectrum receiver 9200 according to the present invention relates to its superior responsiveness to increases in the number of samples per symbol period used in the symbol recovery processing sequences. While the bispectrum processing sequences are able to recover symbols with progressively improving BER at a given SNR in direct response to increased samples per symbol period, the standard receiver improves marginally with increased samples per symbol period. Therefore the bispectrum receiver 9200 according to the present invention possesses the inherent scalability for deployment in systems with greater than 2 bits per symbol because increased samples can be used to maintain BER performance at the desired SNR within the transmission system.

Significant extensions to the present invention based on the use of the optical pre-processor 5100 will now be described. According to the present invention the optical pre-processor 5100 may be constructed in two forms. The first form is illustrated in Figure 6(a) while the second is illustrated in Figure 6(b).

In the first form illustrated in Figure 6(a) the optical pre-processor 5100 may be constructed in a multiplicity of physical forms including but not limited to a Planar Lightwave Circuit (PLC) 5100 comprising a light splitting chamber 5105, the first waveguide 5410, the second waveguide 5420 through to the J^m waveguide 5499. The second waveguide 5420 through to the J"¹ waveguide 5499 are designed to apply to the light wave progressively larger transmission delays 5110 through to 5199, where the total delay is less than the symbol period of the original light wave 4030. The objective according to the present invention is to achieve the effect of generating J output signals Si 5410, S₂ 5420 and so on to Sj 5499 such that the first output signal Si 5410 is a corresponding (undelayed) version of the input signal 4030, the second output signal S₂ 5420 is a version of the input signal 4030 but delayed by one delay quantity 5110 and so on up to the J^th output signal Sj 5499 also being a version of the input signal 4030 but delayed by J- delay quantities. Each of the J output signals is then coherently detected to generate the characteristic output signal pairs lsi 5210 and Q_Si 5210, Is₂ 5230 and Qs2 5240, and so on up to l_Sj 5270 and Q_Sj 5810. These signal pairs are individually processed each through a dedicated coherent receiver within the coherent receiver structure 5200 to produce I and Q signal pairs 5210 and 5220, 5230 and 5240, through to 5270 and 5280. These I and Q signal pairs are then digitized by the ADC processing structure 5400 to provide the (I and Q) signal data samples into the bispectrum generating sequence 5600.

When these data sample sets are eventually compiled into the final sample sequence 4033 within the bispectrum generation sequence 5600 the total number of data samples registered for each symbol period is increase in proportion to the number J of the different versions Si 5410, S₂ 5420 and so on to Sj 5499 of the original optical signal 4030. For example as illustrated in Figure 7 the original optical signal plus three progressively delayed versions of it have been generated by the optical pre-processor 5100 to provide a corresponding number of I and Q signal pairs out of the coherent receiver structure 5200.

Shown in Figure 7 are the samples generated by the ADC structure for the I components of the four coherent receivers. As shown in Figure 7 where the sampling rate of the ADC structure is set to four sample over the symbol period, the net effect of sampling these progressively delayed versions of the same signal achieves sixteen distinct data samples of the signal much the same as if the signal had been samples at sixteen sample over the symbol period. This apparent increase in the sampling rate provides a significantly larger bispectrum matrix to characterize the symbol period of the received signal. According to the present invention the ability of the receiver 5200 to negate the effect of transmission impairment noise accumulated over the originally transmitted symbol signal is progressively improved by processing the received signal through progressively larger bispectrum matrices.

Since commercial optical transmission systems will continue to transmit at rates significantly higher than the prevailing speeds of electronic sampling technologies, the optical pre-processor according to the present invention provides means of improved matching of capacities between optical transmission technologies and the available electronic sampling technologies.

The second form of the optical pre-processor 5100 according to the present invention is illustrated in Figure 6(b). In the second form of the optical pre-processor 5100 the optical pre-processor 5100 may be constructed in a multiplicity of physical forms including but not limited to a first planar Lightwave circuit 5501 comprising a light splitting chamber 5106, the first waveguide 5411 , the second waveguide 5412 with a specified time delay τ, 511 1 and the third waveguide 5413 with another specified time delay τ₂ 5112.

At the exit point of the planar light wave circuit 5501 the three light waves 541 1 , 5412, 5413 are recombined constructively in the combining chamber 5510, together with added pump signal 5150. The resulting output wave is then propagated through a third order highly non-linear waveguides 5190 constructed using appropriate highly non-linear light wave circuit 5500. The specific objective of the present invention is to exploit the third order susceptibility commonly referred to as x(3) nonlinearity of the highly non-linear light wave circuit 5500 to generate a fourth wave related to the three co- propagating waves. According to the present invention the resulting output waves from the highly non-linear light wave circuit 5500 comprise a mixture of signal components related to the original three input signals and the fourth signal component also related to the original three signals is related specifically to the three co- propagating waves as the triple product of the three waves. Mathematically the fourth wave vector is proportional to the triple product of the co-propagating three wave vectors as follows:

where k is proportionality constant. To extract the fourth wave signal component an optical band pass filter 5520is employed to obtain the desired signal 4031.

The filtered signal 4031 is then passed through the PLC 5100 as previously described and illustrated in Figure 6(a) to generate the necessary input signals for the coherent receiver structure 5200 also as previously described where they are coherently detected to generate the characteristic output signal pairs lsi 5210 and Q_Si 5210, l_S2 5230 and Q_S2 5240, and so on up to l_Sj 5270 and QSJ 5810. These I and Q signal pairs are then digitized by the ADC processing structure 5400 to provide the (I and Q) signal data samples into the bispectrum generating sequence 5600. The important feature of the I and Q samples generated here is that they are samples of the triple product function of the original signal received, entirely processed in the optical domain prior to coherent detection. Referring to Figure 5 it can be seen that triple correlation step 5610 of the bispectrum processing sequence 5601 as described earlier is simplified by removing the triple product generation step thus avoiding a large number of multiplication operations. Only the summation step is required. This reduction in the number of operations carried out in the bispectrum processing engine 5600 directly leads to improved throughput and can therefore enable processing of higher symbol rates than otherwise possible. Various modifications may be made in details of design and construction and process steps, parameters of operation etc] without departing from the scope and ambit of the invention.

Claims

A method and apparatus for highly sensitive detection and recovery of optically transmitted signals, the method including the step of.

a) Receiving an optically transmitted signal in the form of a modulated optical carrier signal comprising N-bits per symbol, where N≥1 [that is 1 , 2, 3, 4 or greater] and where each of the two polarization modes of the carrier is used to carry its own set of symbols of the same bit capacity; b) Separating the received signal via a polarization de-multiplexer (4500) into at least a first separated signal (4030) and a second separated signal (4040);

c) Separating the first separated signal (4030) and a second separated signal (4040) into a set of parallel optical signals [via a optical preprocessor] S , S₂, ... Sj, where J = 2^N [comprising the first signal Si 5410, the second signal S₂ 5420 up to the J^,h signal Sj 5499];

d) Passing each parallel optical signal Si, S₂, ... Sj through a coherent receiver (4032), each parallel optical signal Si, S₂, ... Sj thus generating a pair of output electrical signals being a real signal (I) and an quadrature signal (Q) (S_1t/S_1Q, S₂|/S_2Q, ... SJI/SJQ) (5210, 5220, ... 5299);

e) Digitizing each of the pair of output electrical signals, being the in-phase signal (I) and an quadrature signal (Q) (Sn/Si_Q, S₂I/S₂Q, ... SJI/S_JQ), [analogue to digital converter] into a stream of digital time sequenced sample sets (4033)

f) Passing the time sequenced sample sets (4033) into a bi-spectrum

generator (7000) wherein the time sequenced sample sets (4033) are separated into:

i) Symbol region data (9220); and

ii) Optical health transmission data (9250);

g) Wherein the Optical health transmission data is data outside of the

Symbol region data.