Opto-electronic frequency divider circuit and method of operating same

[0002] The present invention refers to frequency divider circuits and has been developed by paying particular attention to their possible application to optical transmissions based on the so called OTDM (Optical Time Division Multiplexing) technique.

[0003] OTDM is particularly interesting in all situations where the need arises to increase the transmission capacity of an optical link, and it is an alternative to other solutions based on sharing the same physical carrier among multiple channels—typically by employing WDM (Wavelength Division Multiplexing) techniques—or on increasing the number of optical fibers available for the link.

[0004] This latter solution generally requires interventions on the installation (such as laying of new cables and performing excavations) and it does not completely exploit the extremely wide band made available by optical fibers.

[0005] The simultaneous transmission of many different channels on the same optical fiber, for example according to wavelength division multiplexing techniques, allows use of low speed opto-electronic components both in the transmitter and in the receiver, while obtaining a high aggregate capacity on the link. Wavelength division multiplexing further allows implementing at optical level some network functions like channel removal and insertion, dynamic routing, link protection, with a high reduction of the processing load for the electronic part in network nodes. The major inconveniences of such method are linked to the need of selecting and stabilizing the wavelengths for transmitters and optical filters used for channel selection, to the possible inter-channel interference due to non-linear phenomena in fiber propagation (for example the phenomenon known as Four Wave Mixing) or to the spectral nonuniformity of optical amplifier gain.

[0006] In OTDM systems, many optical signals, intensity modulated according to an RZ (return to zero) code, are interleaved into a single flow by acting on the relative delay of the pulse sequences. This solution retains the majority of advantages of WDM techniques related to the possible use of low speed opto-electronic components both in the transmitter and in the receiver, further avoiding the onset of some of the above-mentioned negative phenomena. A basic condition for the proper operation of an OTDM system is however that the different optical tributary flows must be well synchronized and composed of sufficiently narrow pulses in order to avoid interference among channels. Moreover, it is essential that a driving signal at tributary frequency and synchronous with the multiplexed flow be made available at the demultiplexing device.

[0007] The object of the present invention is to provide a solution to this latter need and, more generally, to provide a particularly simple opto-electronic frequency divider circuit, adapted to operate even at very high frequencies (typically with input frequencies of the order of several tens of Gbit/s) with good performance as regards the stability of frequency and phase locking between the signal resulting from the division and the input signal.

[0008] According to the present invention, such object is attained by means of an opto-electronic frequency divider circuit.

[0009] Opto-electronic frequency divider circuit which comprises:

[0010] electro-optical mixer means with a nonlinear behavior, adapted to receive as input a first optical signal (P_in) at a frequency to be divided (f₀) in addition to an electric signal (e₃) at a given frequency (f₁) and to generate as output a second optical signal (P_out) whose modulation spectrum includes, due to the mixing action, the frequency corresponding to the difference between the frequency to be divided (f₀) and at least one harmonic of the given frequency (f₁) generated due to the nonlinear behavior,

[0011] a feedback path comprising opto-electronic converter means to convert the second optical signal (P_out) into an electrical conversion signal (e₁) adapted to be sent back to the electro-optical mixer means,

[0012] filtering means associated with the feedback path to extract from the spectrum the component at the difference frequency, and

[0013] extracting means to derive from the feedback path as signal (e₂) resulting from the frequency division action, a signal at the difference frequency.

[0014] The electro-optical mixer means can include a Mach-Zehnder electro-optical modulator.

[0015] The electro-optical mixer means can be provided with a control input (V_bias) to select the order of the at least one harmonic. The electro-optical mixer means can exhibit a substantially sinusoidal transmittivity/input voltage characteristics, and in that the electro-optical mixer means are made to operate next to one of the intermediate points in the characteristics, so that the at least one harmonic is an odd-order harmonic. This harmonic can be the third harmonic of the given frequency.

[0016] The feedback path can include at least one delay element which may be an optical waveguide interposed between the electro-optical mixer means and the opto-electronic converter means. Alternatively the delay element can be a delay line operating on electrical signals and located, along the feedback path, downstream of the opto-electronic converter means. Within this feedback path, the filtering means are located downstream of the opto-electronic converter means. The feedback path can include gain control means to keep the electrical signal (e₃) at such a level as to ensure the nonlinear behavior of the electro-optical mixer means.

[0017] The filtering means can be connected in the feedback path, so that the component at the difference frequency is used as the electric signal (e₃) fed to the electro-optical mixer means.

[0018] The extracting means can be located downstream of the filtering means so that the component at the difference frequency is used as signal (e₂) resulting from the frequency division action.

[0019] The first optical signal (P_in) can be a signal belonging to an aggregate flow obtained by optically time division multiplexing a plurality of tributary flows, each one having a bit rate equal to the given frequency (f₁), and the signal (e₂) resulting from the frequency division action can be a signal synchronous with the tributary flows that forms a synchronism signal for demultiplexing the aggregate flow.

[0020] The invention also comprises a method of extracting from an optical signal (P_in) conveying an aggregate flow of a given number (N) of optical tributary signals interleaved according to an optical time division multiplexing scheme, a synchronism signal at the frequency (f₁) of the optical tributary signal. The method includes the following operations:

[0021] subjecting the optical signal (P_in) to a nonlinear electro-optical mixing operation with an electric signal (e₃) at the frequency (f₁) of the optical tributary signals in order to generate a further optical signal (P_out) having a modulation spectrum including, due to the mixing operation, a frequency corresponding to the difference between the frequency of the aggregate flow (f₀) and at least one harmonic of the frequency (f₁) of the optical tributary signals, generated due to the nonlinear behavior of the mixing operation,

[0022] converting the further optical signal (P_out) into an electrical conversion signal (e₁) that can be used to generate the electrical signal (e₃) for the electro-optical mixing operation, according to a general feedback path to which a filtering operation is associated to extract from the spectrum the component at the difference frequency, the signal thereby extracted being the synchronism signal.

[0023] The method, when applied to an aggregate flow of N optical interleaved tributary signals is characterized in that the electro-optical conversion operation is performed with such a nonlinearity degree that the at least one harmonic is an (N−1)th-order harmonic. The method can include the operation of delaying propagation of at least one of the steps of:

[0024] (1) delaying propagation of the further optical signal (P_out) within the feedback path,

[0025] (2) delaying propagation of the electrical conversion signal (e₁) within the feedback path, and

[0026] (3) delaying propagation of both the further optical signal (P_out) and the electrical conversion signal (e₁) within the feedback path.

[0027] In the preferred but not exclusive application to an OTDM transmission system, this circuit allows performing a division of the aggregate bit rate by employing components with a bandwidth not exceeding the tributary bit rate. The practical embodiment of the frequency divider therefore does not require a technological development degree higher than the one necessary to make the transmitter and receiver of the single channel in the transmission system.

[0028] The above and other objects, features, and advantages will become more readily apparent from the following description, reference being made to the accompanying drawing in which:

[0029] FIG. 1 shows, in block diagram form, the general arrangement of an OTDM optical transmission system to which a frequency divider circuit according to the invention can be applied;

[0030] FIG. 2 shows—also in block diagram form—the general structure of the divider circuit according to the invention; and

[0031] FIG. 3 is a diagram showing an operating characteristic of one of the components of the circuit in FIG. 2.

[0032] In FIG. 1 reference T globally designates an optical link realised according to the OTDM technique.

[0033] System T, operating according to criteria known per se, allows conveying on an optical fiber line W (with associated respective amplification/equalisation units schematically represented by blocks AE1 and AE2) an RZ optical signal intensity modulated according to RZ (return to zero) code and having a bit rate f₀equal, for example, to 40 Gbit/s.

[0034] Such optical signal is obtained by aggregating in a multiplexing unit (optical coupler) OC a number of optical signals (four in the example), defined as “tributary”, each one with bit rate f₁equal to 10 Gbit/s.

[0035] It is clearly apparent that reference to an aggregate flow with bit rate equal to 40 Gbit/s, obtained through multiplexing four 10 Gbit/s tributary flows, must be deemed purely as an example. Both the bit rate for tributary signals, and the number of such multiplexed signals, and—consequently—the bit rate of the resulting aggregate flow are design parameters adapted to be widely modified as a function of the specific application needs and of the components being used, without departing from the scope of the present invention.

[0036] In particular, the tributary flows (hereinbelow in the present specification four tributary flows will always be referred to, as an example) are generated starting from the RZ pulse train issued by a pulse source S, typically composed of a laser source with sufficiently narrow pulses that are time-domain limited. The signal from source S is split by a separator SP1 into a plurality of replicas each fed to a respective modulator M1 to M4.

[0037] The transmitter input data, in the form of electric signals ideally coming from a data source SD synchronously driven with source S due to the common slaving to a synchronism oscillator SYN (also operating, in the disclosed embodiment, at a frequency of 10 GHz, equal to the frequency of source S), are organised in a corresponding number of channels C1-C4, each with a bit rate of 10 Gbit/s. The signals present on channels C1-C4 drive the respective optical modulators M1-M4. The latter ones “write” the information on the related pulse trains and the signals obtained are sent to fiber W through optical coupler OC after having been mutually time-offset by a time interval Δt, for example in respective adjustable optical delay lines L1-L4.

[0038] The aggregate flow injected into fiber W thereby shows a typical time-domain multiplexed structure. In practice, the aggregate flow conveyed by fiber W is cyclically composed of a symbol corresponding to a datum coming from channel C1, a symbol corresponding to a datum coming from channel C2, a symbol corresponding to a datum coming from channel C3, a symbol corresponding to a datum coming from channel C4, etc.

[0039] At the receiving side, the aggregate flow is sent to an optical demultiplexer DMPX that orderly extracts from the received aggregate flow the signals corresponding to the tributary channels and routes them towards corresponding receivers RX1-RX4 to recover at the output the flows corresponding to channels C1-C4.

[0040] To be able to correctly extract the different tributary signals, the demultiplexer device DMPX requires a driving signal e₂(which is supplied by a synchronism recovery circuit globally designated as 1 and which is to be also sent to receivers RX1-RX4) at the tributary frequency and phase locked with the aggregate (or multiplexed) flow frequency.

[0041] When operating in RZ format, a line at the aggregate flow bit rate is always present in the multiplexed flow spectrum, while the spectral component at tributary frequency is absent (it disappears due to the multiplexing operation). In order to correctly operate, the synchronism recovery circuit must perform a division of the aggregate bit rate by a factor equal to the multiplexing factor, while keeping the phase locking between the original frequency and the divided one.

[0042] In the specific case, the signal corresponding to aggregate bit rate is extracted from the signal coming from fiber W through a component like separator SP2.

[0043] Synchronism recovery circuit 1, that is the subject matter of the present invention, is therefore a circuit that, starting from the optical signal at the aggregate frequency, is able to generate an electric signal with a frequency corresponding to the tributary signal frequency.

[0044] The circuit according to the invention ideally refers to a circuit known as Miller frequency divider. Such divider circuit, also known as regenerative divider since it is composed of a feedback system, is characterised by very low added noise levels. It found thereby use in generating sources with high spectral purity, obtained by dividing high frequency references. For a general description of the features of such known circuit, reference can be made to the paper by R. C. Miller “Fractional-frequency Generators Utilizing Regenerative Modulation” in Proceedings. IRE, Vol. 27, pages 446-457, July 1939.

[0045] Similarly to the Miller divider, circuit 1 according to the invention uses a component adapted to perform a mixer function. Here a multiplication is performed between the component at the frequency corresponding to the pulse repetition rate of the signal to be divided, and the harmonics of the electrical division signal coming from the feedback loop. Such harmonics originate inside such component, designated with reference 2 and characterised by a non-linear behaviour.

[0046] Any component showing this type of features can then be used in the invention. In the currently preferred embodiment, component 2 is an electro-optical Mach-Zehnder modulator having transmittivity characteristics as shown in the diagram in FIG. 3.

[0047] In such a diagram, the ordinate axis shows transmittivity T (normalized to unity as maximum value) versus a parameter that can be expressed as (V_bias+V_RF)/V_π, and that is characteristic of the driving conditions for modulator 2.

[0048] In particular, V_biasrepresents a bias voltage applied to a first driving input of modulator 2, while V_RFis a radiofrequency driving signal applied to a corresponding input. In the specific embodiment shown in FIG. 2, the concerned radiofrequency signal is designated as e₃. Parameter V_π is a normalization parameter: in practice, it is the voltage to be applied to move from maximum to minimum transmittivity. Obviously the transmittivity refers to the input/output behavior of the optical signal passing through the modulator. In the example shown, the optical input signal is represented by signal P_inwhile the output signal, designated as P_out, is equal to the input signal multiplied by transmittivity. With modulator 2 it is therefore possible to perform a multiplication between the spectral components in optical input power modulation and the driving signal harmonics.

[0049] If the radiofrequency signal V_RFis a sufficiently strong sinusoidal signal, transmittivity includes components with frequencies corresponding to the frequencies of the harmonics of signal e₃. The relative amplitudes of such harmonics depend on the working point position on the modulator characteristics (see FIG. 3), and therefore on bias voltage V_bias. In particular, it is possible to have even harmonics only, by placing the working point next to maximum or minimum transmittivity, or only the odd ones, by operating at the characteristics centre.

[0050] In the specific embodiment shown here, it is desired to perform a frequency division by four (N=4). Therefore, the choice has been operating at the characteristics center, that is next to one of the inflexion points shown by arrows in FIG. 3, in order to generate and use the third harmonic (N−1=3) of the radiofrequency signal.

[0051] In other words, the following results are obtained:

[0052] generation (due to non-linear behavior), starting from signal e₃at frequency f₁, of the (N−1)th order harmonic at frequency (N−1)f₁, and

[0053] generation (due to the typical multiplying behavior of modulator 2) of an output signal whose frequency is equal to the difference between the input signal frequencies (in addition to a signal with sum frequency).

[0054] In general, since frequency f₀conveyed by input signal P_inis N times frequency f₁(f₀=Nf₁), the output difference signal will have the following frequency

[0055] Nf₁−(N−1)f₁=f₁

[0056] this latter one being the desired frequency.

[0057] Obviously, by choosing a different working point and/or a component with different non-linear characteristics, it is possible to generate a different-order harmonic and therefore, to perform a division by a different factor. In any case it is important to note that, since harmonic generation occurs inside modulator 2, it is not necessary that the passband of radiofrequency input of modulator 2 itself extends beyond the frequency of signal with frequency f₁.

[0058] In summary, in the diagram in FIG. 2, the optical input signal P_in, intensity modulated at the frequency f₀to be divided (40 Gbit/s in the mentioned example) originates, through electro-optical modulator 2, optical signal P_out. This latter one is received, possibly through a section of optical fiber 3 operating as delay line, by a photodiode 4. In photodiode 4 the optical signal P_outis converted into an electric signal e₁. After passing through a variable electric delay line 5 located downstream of photodiode 4, an amplifying stage 6 and a variable attenuator 7 (whose function will be better described below), signal e₁is filtered in a pass-band filter 8 tuned around frequency f₁. The filtered signal thus obtained, designated as e₃, is sent back to input V_RFof modulator 2, thus providing a feedback. Usually, when passing from filter 8 (also located downstream of photodiode 4 within the feedback path) to modulator 2, the signal is also made to pass through an amplitude-adjusting amplifying stage 9 and an extracting device 10, such as for example a directional coupler. The latter one extracts a fraction of the feedback signal intended to build the output signal e₂for the device.

[0059] The circuit performs a phase locking between input signal P_inand oscillating signal e₃and this makes the circuit itself useful for applications like the tributary synchronism recovery device in the diagram in FIG. 1.

[0060] Experiments carried out by the Applicant have demonstrated the advantages of employing, within the general diagram in FIG. 2, some particular application embodiments. These embodiments are to be considered generally preferred, even if not mandatory per se.

[0061] In particular, the input signal P_inis preferably injected into modulator 2 through a polarisation controlling member, of a known type and not specifically shown.

[0062] The use of optical fiber 3, responsible for the majority of delays incurred by signals in the feedback path, is generally deemed preferred in order to stabilise the circuit operation, when necessary. The variable delay line 5 provided in the electrical section of the feedback loop is necessary for the fine control of the global phase shift. Preamplifier 6 and variable attenuator 7 allow accurately adjusting the circulating power.

[0063] The pass-band filter 8 does not require particular selectivity properties. Amplifier 9 is to raise signal e₃supplied to input V_RFof modulator 2 to a sufficiently high level (for example about 25 dBm) so that such signal, inside modulator 2, is subjected to the necessary distortion to produce the desired harmonic.

[0064] Obviously, without changing the principle of the invention, component parts and embodiments can be largely modified with respect to what is described and shown, without departing from the scope of the present invention.