ALL-OPTICAL DATA SIGNAL PROCESSING DEVICE

The present invention relates to all-optical signal processing devices.

BACKGROUND OF THE INVENTION

The all-optical processing of digital signals is highly desirable in optical information systems in order to avoid converting the signal into electronic form. Various functions such as optical pulse re-amplification (1 R), reshaping (2R) and retiming (3R), wavelength conversion, pulse gating (or demultiplexing) have been demonstrated using several kinds of devices including optical fibre based nonlinear optical loop mirrors (NOLM) and semiconductor optical amplifier Mach-Zehnder interferometer (SOA-MZI) devices. However, fibre based devices are large in size and sensitive to environmental disturbances, and are difficult to integrate into more complicated systems involving multiple functions. SOA based devices need high current to operate, and therefore have high power consumption. It is also are difficult to obtain high performances (such as signal extinction ratio) due to limited fabrication accuracies and phase drift between the two arms of the interferometer. The SOA sizes are typically in the order of millimetres, making it difficult to integrate multiple devices monolithically on the same chip.

A prior art device used two semiconductor ring lasers (SRL) to achieve two digital states represented by the two directions of operation in these SRLs. This relied on the mutual locking between these SRLs to maintain the direction of operation not changing from one to another. This device has the drawback of having to use two identical SRLs. Any mismatch in the two involved SRLs (such as a lasing frequency difference) will cause the operation to be disrupted. It also occupies twice the area of the SRL device.

Another prior art device, as described in US 2005/0012982, assigned to Binoptics Corporation, uses a ring laser in the form of a cavity having at least two facets as a wavelength convertor apparatus. A first optical signal is supplied to the laser cavity input at a first facet, with this signal being in the form of a light beam at a wavelength A ₂ at a first angle to the first facet. This input signal results in laser propagation in a counter-clockwise mode within the ring laser cavity to produce an output R of laser light at the wavelength A ₂ at the second, or output facet. A second optical input signal A of laser light having a wavelength A ₁ , is directed into the laser cavity at a second angle to

the first facet. If the second optical input is stronger than the first, and the first and second angles are symmetric about the perpendicular to the first facet, injection locking and light propagation in the clockwise mode is produced, substantially eliminating the output R. In this manner, the output signal R at wavelength A ₂ is switched on and off by the absence or presence, respectively, of an input signal at wavelength λi, thereby converting the input signal at λi to an inverted output signal at A ₂ .

However, the configuration of the device is such that it may have relatively high levels of back-scattering (i.e. light being scattered back towards to the direction of incidence) at the facets. This is because it relies on partial reflection at these facets to obtain output power from the ring cavity. This output mechanism requires that the two waveguides form an acute angle between them, so that each form an angle with the facet normal that is less than the critical angle for total internal reflection, and such acute angles are more likely to produce high levels of back-scattering. Back-scattering adversely affects the performance characteristics of the device, such as the extinction ratio (the ratio between power levels at 'high' and 'low' states at any output) and also prevents the device from operating stably in one direction in the absence of an external input beam.

Yet another prior art device can be found in US 5,132,983, assigned to Cornell Research Foundation, Inc. This patent describes a similar form of device to the above document. As optical logic involves more than one input beam, this prior art device has multiple facets to allow access by multiple input beams. Each leg in the cavity has a facet between two acutely angled waveguides, and each leg is connected to the rest of the cavity by total internal reflection mirrors. It also uses external reflection mirrors to force the device to automatically return to one particular direction after having been set to the other direction.

However, as there are more facets in the cavity, there are more sources of back- scattering. This not only increases the general level of back-scattering, but also causes a greater disturbance to the operation of the device due to the uncertain and varying phase differences between these sources.

Thus these prior art devices achieve a unidirectional operation by forcing light into a certain direction using external reflectors. However, this results in devices that can

only perform one or a small number of functions, as both the strength of reflection and its direction are fixed once the device is fabricated.

Therefore, it is an object of the invention to provide an optical signal processing device that does not suffer from the disadvantages associated with the prior art devices.

SUMMARY OF THE PRESENT INVENTION

According to the present invention, there is provided an optical signal processing apparatus comprising a semiconductor ring laser device having first and second directions of laser action, first and second inputs for coupling light into the first and second directions of laser operation of the semiconductor laser device respectively, and first and second outputs for coupling light out of the first and second directions of laser operation of the semiconductor laser device respectively.

The device consists of a semiconductor ring laser (SRL). The semiconductor ring laser is designed to operate at one of the two possible stable states at any moment, one being the laser light propagates in the clockwise direction inside the ring, the other being the laser light propagates in the counter-clockwise direction inside the ring. These two states are used to represent binary data of '1 ' or O'. The state of the semiconductor ring laser may remain unchanged in the absence of any external input light or assisted by external light input. Mechanisms are included to couple light in and out of the semiconductor ring laser. This device can be operated to perform multiple all- optical digital data processing functions including, but not limited to, all-optical data pulse re-shaping, re-timing, wavelength conversion, gating, de-multiplexing, and data envelope detection.

BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1 is a schematic illustration of all-optical digital signal processor based on semiconductor ring laser;

Figure 2 illustrates the operation of the device in the presence of two varying input signals;

Figure 3 illustrates an all-optical pulse reshaping operation of the device of Figure 1 ;

Figure 4 is a schematic illustration of the pulse reshaping process;

Figure 5 illustrates an all-optical pulse reshaping and wavelength conversion operation;

Figure 6 illustrates an all-optical pulse reshaping and retiming operation;

Figure 7 shows the timing relationship between input data pulses, optical clock, and reshaped/retimed output data pulses;

Figure 8 illustrates an alternative all-optical pulse reshaping and retiming operation;

Figure 9 illustrates an all-optical pulse gating or de-multiplexing operation;

Figure 10 illustrates timing of the optical gating/data de-multiplexing operation;

Figure 11 illustrates timing of an alternative optical gating/data de-multiplexing operation;

Figure 12 illustrates an all-optical data envelope detection scheme;

Figure 13 is a timing schematic of the data envelope detection;

Figure 14 illustrates optical single pulse detection operation;

Figure 15 is a timing schematic of the optical single pulse detection operation;

Figure 16 illustrates an all-optical programmable logic device;

Figure 17 illustrates a timing schematic for the programmable logic device operating as an AND and NAND gate;

Figure 18 illustrates a timing schematic for the programmable logic device operating as an OR and NOR gate;

Figure 19 illustrates a timing schematic for the programmable logic device performing NZR to RZ conversion;

Figure 20 illustrates an all-optical XOR gate; and

Figure 21 illustrates a timing schematic for the all-optical XOR gate.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

Embodiments of the present invention aim to provide an all-optical digital data pulse processing device and methods of operating this device to provide multiple all-optical data processing functions. In order to illustrate the general principle of one such device, a schematic illustration is provided in Figure 1.

A device embodying the present invention comprises a single semiconductor ring laser (SRL) 10. The semiconductor ring laser 10 has a ring cavity, and a semiconductor optical gain medium that forms at least part of the ring cavity. The ring cavity is an enclosed path of any shape that allows light propagation in such a way that light wave starting from any point along the path will return to the same point after traversing any extended section of the path only once. Examples of the ring cavity shape include circular, annular, oval or ellipsoidal, rectangular, polygonal, a 'racetrack' shape that comprises two parallel straight sides connected by curved paths at both ends, etc.

As described above, it is desirable for the semiconductor ring laser 10 to be able to support stable laser operation in any one of the two directions of propagation - the clockwise and counter-clockwise directions - when no external light input is present, and when the laser's output light beam is not reflected back into the cavity by external reflection. As mentioned above, in prior art devices, such unidirectional operation is achievable only when a continuous external light input is present or external reflection is used. The key issue is to reduce any back-scattering in the device (i.e. light energy that is reflected or scattered back in the direction opposite to its propagation in the ring cavity). Such back-scattering or back-reflection may result from locations both inside and outside the ring cavity.

Thus, in accordance with the invention, self-sustained unidirectional operation of the semiconductor ring laser 10 is achieved by using evanescent wave couplers to input

the light to, and output the light from, the semiconductor ring laser 10. In some embodiments, the evanescent wave couplers can be directional couplers, which are evanescent wave couplers with two parallel waveguides. Evanescent couplers are known to have very low levels of back-scattering or back-reflection therefore are conducive to achieving self-sustained unidirectional operation.

In addition to the above, back-scattering can be further reduced and therefore self- sustained unidirectional operation improved by using any or all of the following: a substantially smooth light path in the waveguide (i.e. there are no discontinuities in the boundaries of the light path); using an evanescent wave coupler or a directional coupler to input the light to, and extract the light from, the ring cavity; terminating access waveguides by forming a non-vertical angle with the chip facets (as access waveguides always terminate at the edge of the chip, where light is coupled into or from optical fibres, the waveguide can be tilted away from the vertical such that reflection is significantly reduced); and applying an anti-reflective film to the facets (in order to further reduce the reflection from the facets).

In a preferred embodiment of the invention, the semiconductor ring laser 10 has a ring cavity that does not have any discontinuities in the light path, so any change in the direction of the light occurs gradually. Such a cavity can be in the form of a circular waveguide loop. In order for the output coupling mechanism used in the device to avoid introducing significant levels of back-scattering, an evanescent wave coupler (also known as directional couplers) is formed between each access waveguide and the ring cavity waveguide by placing both in close proximity.

In further embodiments, as a circular cavity does not couple a significant amount of power into or from a straight access waveguide, the ring cavity is provided with two straight sections that are each parallel to an access waveguide for a distance sufficiently long enough for significant coupling to take place. Thus, the cavity comprises two substantially parallel straight waveguides connected together at each end by a semi-circular waveguide, thereby forming a closed loop in the shape of a stadium or racetrack. In yet further embodiments, the racetrack shaped cavity can be further improved by treating the end facet of the access waveguide where it terminates at the edge of the chip, by means such as either forming a non-vertical angle with the facet, or by applying an anti-reflective film coating, or both, in order to minimise any

reflection back into the access waveguide that will be coupled into the opposite direction in the ring cavity.

In the descriptions hereafter a circular or annular shaped ring cavity is used to illustrate the principles of operation.

In the exemplary device shown in Figure 1, the semiconductor ring laser 10 is formed by a substantially circular closed loop optical waveguide. It includes an input coupler 18 where inputs 14, 15 can be coupled to each direction and output coupler 19 where outputs 16, 17 can be coupled from each direction. The clockwise direction 11 is indicated by dashed lines and the counter-clockwise 12 direction is indicated by solid lines. In this embodiment, the input coupler 18 for each direction comprises an evanescent wave coupler and the output coupler 19 for each direction comprises an evanescent wave coupler. Each of these couplers is formed by placing the access waveguides 8 or 9 in close proximity to semiconductor ring laser 10.

The semiconductor ring laser 10 is designed to operate at one of the two possible stable states at any moment, one being the laser light propagates in the clockwise direction 11 inside the ring, the other being the laser light propagates in the counter- clockwise direction 12 inside the ring.

When back-scattering and back-reflection are minimised as described above, the achievement of the operation in one or other direction of the SRL 10 can rely on an effect known as nonlinear gain competition between the two directions inside a single SRL 10, in which any direction that has a slightly higher optical power will have higher optical gain than the opposite direction. This will result in the growth of the stronger direction further, and hence creating even less gain for the weaker direction. This effect can be strong enough to result in the SRL 10 essentially operating in one direction only and essentially suppress the operation in the opposite direction, therefore making the state of the SRL 10 self-sustained.

Furthermore, when back-scattering and back-reflection are minimised as described above, the SRL 10 can achieve self-sustained laser operation at any one of multiple wavelengths in each of the two operation direction. The multiple wavelengths (or modes) are essentially defined by the resonance wavelengths of the ring cavity. Because the laser material usually has a gain spectrum that is characterised by a

maximum at a central wavelength A ₀ with reducing gain values at either side of λ ₀ , not all of the multiple wavelengths will coincide with this central wavelength. The fact that the laser can operate at a wavelength away from the gain maximum is again due to a nonlinear gain competition favouring a laser mode with higher intensity, therefore enabling its stable operation even when it is located away from the gain maximum A ₀ , without reverting to operating at a wavelength closer to λ ₀ .

When the SRL device is provided with a supply of energy such as an electric current injection that provides optical gain above the threshold for laser operation, it is usually observed that the device will take a direction of operation that is randomly chosen from the two possible directions. The device will also take a wavelength of operation that is close to the maximum wavelength A ₀ of the gain of the semiconductor materials that forms the active layer of the device.

The operation of the SRL 10 in a designated direction can also be achieved by intentionally inputting an external light wave into the desired direction. Thus, the SRL 10 can be operated in the clockwise direction by coupling light into the SRL 10 using input 14, and in the anticlockwise direction by coupling light into the SRL 10 using input 15. Furthermore, in each direction, the laser operation wavelength can be decided by the wavelength of the input light, as long as the input light wavelength is essentially aligned with one of the resonance (modes) of the ring cavity. Furthermore, the SRL 10 will continue to operate in the direction and at the wavelength set by the input light wave after the input light wave ceased to exist.

The SRL 10 can also be operated to receive external input optical waves into both directions, i.e. via both inputs 14, 15. In this case the direction of operation will depend on the strengths and the wavelengths of the two input waves.

The response of a device to the presence of two input light beams is shown in Figure 2. In Figure 2, a "control" beam with a first power value and a first wavelength is injected into the counter-clockwise direction (i.e. using input 15), and this holds the SRL 10 in that direction and the first wavelength so that the output from the clockwise direction (i.e. from output 16) takes a low value as indicated by 'off'.

A "signal" beam is subsequently injected into the clockwise direction via input 14. The signal beam may have a second wavelength that is essentially aligned with that of the

'control' beam. The signal beam may also have a second wavelength that is essentially different from that of the 'control' beam in that it is aligned with another resonance wavelength of the SRL 10. Until the "signal" beam power reaches a switch- on threshold value, the SRL 10 remains operating in the counter-clockwise direction. The SRL 10 is switched to operating in the clockwise direction as the "signal" beam power surpasses the switch-on threshold value, and the output power from the clockwise direction (via output 16) makes an abrupt transition to a high value indicated as 'on', and its wavelength makes an abrupt transition to the second wavelength. Simultaneously, the output power from the counter-clockwise direction (measurable via output 17) also makes an abrupt transition from a high value to a low value as the counter-clockwise direction is switched off.

When the "signal" beam power is reduced, the SRL 10 remains operating in the clockwise direction until the "signal" beam power reaches a switch-off threshold value. The SRL 10 only returns to operating in the counter-clockwise direction and its wavelength transits back to the first wavelength when the "signal" power reduces below the switch-off threshold value. It is usually the case that the switch-off threshold value is smaller than the switch-on threshold value. Such an output-input relationship is commonly known as a hysteresis loop.

It is the property of the devices used in this invention that the values of both the switch- on and switch-off thresholds are variable by varying the "control" beam power injected into the SRL 10. Furthermore, the difference between these two threshold values is also variable by the same means. In general, as the "control" beam power increases, the threshold values increase, but the difference between them diminishes.

The two directional states in which the SRL can operate are used to represent binary data of '1 ' or '0'. This forms a 'bistable' device. The multiple possible stable wavelengths at each direction form a 'multistable' device. A device embodying this invention also includes mechanisms which couple light into and out of both directions of the bistable SRL 10.

The SRL bistable device 10 can be used to perform all-optical data pulse re-shaping in the following manner, with reference to Figures 3 and 4. By inputting 14, 15 a continuous wave (CW) optical signal into one of the two directions 11 , 12, the SRL 10 is forced to operate in that direction. The optical data pulse to be regenerated is

coupled 15, 14 into the opposite direction 12, 11. When the incoming data pulse power is lower than a switch-on threshold value that is related to the CW optical power, the SRL 10 remains operating in its original direction. When the incoming signal pulse power rises above the switch-on threshold value, the direction of the SRL 10 changes to the direction receiving the incoming data pulse. When the signal pulse power decreases below the switch-off threshold value, the direction of the SRL 10 changes back to the direction receiving the CW light. In this way the shape of a distorted optical data pulse can be reshaped to be closer to an ideal square wave. The thresholds of the optical pulse regeneration are primarily decided by the CW optical power as described above with reference to Figure 2. Furthermore, the required difference between the maximum signal pulse power and the minimum power between pulses - a parameter known as the input extinction ratio (ER) - of the incoming data pulse in order to satisfy both the switch-on and switch-off thresholds is also decided by the CW optical power. With reference to Figure 2, in particular it is possible to obtain a high output signal extinction ratio with a low input signal extinction ratio when using a sufficiently high CW beam power.

As shown in Figure 3, two copies of the optical pulse are obtained from the outputs 16, 17. These two copies can be measured from the two directions of the SRL 10 (i.e. clockwise and anticlockwise), and are logically complementary to one another. The waveforms illustrating the operation of the device in Figure 3 are shown in Figure 4. Both copies of the reshaped data are at the same wavelength A ₁ as the input data and CW light.

Figure 5 illustrates the device being used to perform all-optical data pulse re-shaping, re-timing, and all-optical data wavelength conversion. By using a wavelength λ-i for the CW input 15 that is different from the input data pulse wavelength A ₂ , the SRL 10 is forced to operate at A ₁ in the counter-clockwise direction 12 and at A ₂ in the clockwise direction 11 each time its operating direction changes. Two copies of the data pulse are generated and have different wavelengths, one at Ai and the other at A ₂ . In this way, optical data is transferred from A ₂ to λ-i and in the meanwhile its pulse has been reshaped and re-timed.

Figure 6 illustrates the same device performing all-optical data pulse re-shaping and re- timing, as will be described below. Figure 7 shows the waveforms at the inputs and outputs of the device. A relatively high power optical clock pulse train is coupled into

one direction (in this example, the counter clockwise direction). Thus, the SRL 10 is forced to operate in that direction during the clock pulse. The optical data pulse to be re-shaped and re-timed is coupled into the opposite direction. When the data pulse power is low (e.g., representing a logic O') following an optical clock pulse, the SRL 10 remains operating in its original direction (this is shown in Figure 8 by the "missing" middle data pulse). When the data pulse power is high (e.g., representing a logic '1 ') following an optical clock pulse, the direction of the SRL 10 changes to the direction receiving the data pulse because during this time there is low power at the clock input. When the next clock pulse arrives, due to its higher power, the direction of the SRL changes back to the direction receiving the clock pulse, therefore setting the data output back to low power. In this way the shape of a distorted optical signal pulse can be reshaped to be close to an ideal square wave at output, and the output data pulse position is confined to between two consecutive optical clock pulses. The data pulse therefore has been re-shaped and re-timed. Again, two copies of the optical data pulse are obtained. These two copies can be measured from the two directions of the SRL 10, and are logically complementary to one another.

Figure 9 illustrates the device being used to perform all-optical data gating and demultiplexing. Figure 10 shows the associated waveforms. An optical data pulse train to be gated or de-multiplexed is coupled into one direction of the SRL 10 (in this example, the clockwise input 14). When no light is coupled into the other direction, the SRL 10 then operates in the direction receiving the data pulses by the first '1 ' data bit, and will remain operating at that direction regardless of the subsequent data stream. This effectively stops the data flow from passing through the SRL 10. When a suitable optical gating control light is coupled into the other direction, the SRL 10 will operate as described in the re-shaping operation, thus allowing the data signal to pass through the SRL 10. By changing the duration and timing of the gating control signal, a part of the data can be de-multiplexed from a longer stream of data pulses.

The same configuration as shown in Figure 9 allows the realisation of another method of all-optical data gating and de-multiplexing as shown in Figure 11. In this method, when the gating control is sufficiently high so that the switching threshold is higher than the input data pulse power, switching by the data pulse is prevented so the data will not be allowed to pass through. When the gating control is reduced so that the switching threshold is between the 'high' and 'low' levels of the input data pulses, the data pulses will switch the direction of the SRL and pass through whilst being reshaped.

Figure 12 illustrates the device being used to detect an envelope of an incoming signal, with Figure 13 illustrating the timing diagram. A continuous wave optical signal is input into the counter clockwise direction 12 via the second input 15. This optical input causes the SRL 10 to lase in the counter clockwise direction. When an input data signal in the form of a plurality of pulses is input into the clockwise direction 11 via the first input 14, the lasing direction of the SRL 10 will be changed to the clockwise direction. If, as illustrated, the input data pulses are closely spaced in time, then the

SRL 10 does not have time to switch lasing direction, and so the outputs represent the duration of the pulse series input to the device.

The same device can further be used to perform all-optical single pulse detection. The SRL 10 device is set to operate at a pre-set direction by a 'clearing' pulse. The SRL 10 will remain operating in this direction until the other input of the SRL 10 receives a single optical pulse to be detected. This single pulse will set the SRL to operate in the opposite direction therefore indicating the arrival of the single pulse.

Figures 14 and 15 illustrate the device being used to detect an incoming pulse supplied to the first input 14. Initially, a clear pulse is supplied to the counter clockwise direction 12 via the second input 15 in order to cause the SRL 10 to lase in the counter clockwise direction 12. An input pulse for detection is supplied to the first input 14 and when the pulse arrives, the SRL 10 is caused to lase in the clockwise direction 11 , until a further clear pulse is supplied to the second input 15 to return the SRL to the counter clockwise direction 12. In this manner, a pulse on the second input can be detected by the SRL 10, and the outputs 16 and 17 thereof can supply detection pulses.

In addition to the above operations, the device can also be used to perform all-optical logic operations.

An all-optical AND gate and OR gate can be achieved using an all-optical logic unit comprising the SRL device 1 as illustrated in Figure 16, with two optical logic inputs A and B being combined by an optical coupler 20 to generate a single input F that is input to the SRL device 1 via input 15. A control signal C is provided to the other input of the SRL device 1 (input 14). This configuration provides a programmable logic gate that can selectively operate as an AND and NAND gate, or an OR and NOR gate

depending on the applied optical control signal C. The outputs of the optical logic gate from output couplers 17 and 16 are denoted D and D respectively.

The logic relationship to be achieved by the device 1 depends on the strength and waveform of control input C. Logic inputs A and B each have two states: the '1' state with high power and '0' state with low (but usually non-zero) power.

The waveforms illustrating how the optical device 1 operates as an all-optical AND gate are shown in Figure 17. The two inputs A and B are combined using the optical coupler to produce input signal F. The power of the combined signal F therefore has three levels. F is 'high' when A and B are both '1', 'medium' when one of A or B is 1 and 'low' when both A and B are '0'. The control input C is a CW optical signal. The power of the CW signal is such that the switching threshold is between the 'high' and 'medium' levels of signal F.

The SRL 10 cannot be switched to operate in the counter-clockwise direction by inputs A=1 or B=1 alone as C holds the SRL 10 in the clockwise direction. Only the combined optical power of A=1 and B=1 can cross the threshold level to switch the SRL 10 to operate in the counter-clockwise direction. Thus the output D associated with the counter-clockwise direction corresponds to the logic AND function of inputs A and B

(D=AB), while its complementary output D js the NAND function of A and B (^ = -AB ).

The waveforms showing how the optical device 1 operates as an all-optical OR gate are shown in Figure 18. Here, in contrast to Figure 17, the control input C power is reduced such that the SRL 10 switching threshold is between the 'medium' and 'low' levels of the signal F. The SRL 10 can now be switched to operate in the counterclockwise direction by either input A= 1 or B=1 or both. However, the input power when A=O and B=O is less than the threshold. Thus the output D associated with the counter-clockwise direction corresponds to the logic OR function of inputs A and B, while its complementary output D j _s the logic NOR function of A and B.

Therefore by simply varying the input power of a control light beam C, the same optical device 1 and optical coupler 20 can be used to achieve different logic functions.

The same device 1 and optical coupler 20 can be operated to achieve an all-optical conversion from a none-return-to-zero (NRZ) data format to a return-to-zero (RZ) data format, as illustrated in Figure 19. The NRZ binary optical data is input via either of inputs A or B, with the other being unused. It is assumed that A is used for the input A and B=O in the description below. Alternatively, this conversion can be performed using the device as shown in Figure 1 (i.e. there is no optical coupler 20, and therefore no input for signal B).

Instead of using a CW input as the control signal C, an optical clock signal is used, so that C is now a series of periodic optical pulses having a repetition rate identical to the optical data and a desired pulse-space ratio known as the duty cycle. The C input is such that during the clock pulse, the SRL switching threshold is higher than the A=1 power, while during the clock space the SRL switching threshold is lower than the A=1 power, but higher than the A=O power. Therefore, the SRL 10 is held in the clockwise direction when A=O and when A=1 during a clock space. The SRL 10 only switches to the counter clockwise direction during the clock space and when A=1. Hence the NRZ input A is converted to an RZ output D with its pulse width essentially equal to the clock space width. Logically, the output D is related to inputs A and C by ^{D =} AC _wn j| _e the complementary output D ⁼ AC = A + C _m

Furthermore, the above all-optical logic unit can be connected into more complicated logic circuits. An example is given in Figure 20 which shows that an all-optical exclusive OR (XOR) logic gate can be constructed from the relationship of

A ® C = AC + AC j|η _e | _O gj _c component on the right hand side of this relationship is already described above in Figure 19.

Thus, in Figure 20, the XOR gate comprises two optical devices 1 , with a logic input A being split into two paths by a first optical coupler 21 , and the control input C being split into two paths by a second optical coupler 22. Each path is directed through a respective attenuator 23, 24, 25 or 26 which ensure that the power levels in each path are adequate for the specified switching thresholds. The two paths from the first optical coupler 21 are provided to the counter-clockwise input 15 of the first optical device 10 and the clockwise input 14 of the second optical device 1 respectively. The two paths from the second optical coupler 22 are provided to the clockwise input 14 of the first optical device 1 and the counter-clockwise input 15 of the second optical device 1

respectively. The outputs from the anticlockwise output 17 of each of the first and second optical devices 1 (denoted D ₁ and D ₂ respectively) are combined by optical coupler 28 to produce an output, denoted E.

Thus, it can be seen that the XOR gate configuration comprises two devices connected in parallel, with each part comprising an identical SRL 10 and performing the logic function of ' ⁼ and 2 ⁼ , respectively. As described above with reference to Figure 19, the output of each part not only depends on the position of inputs A and C in the part, it also depends on their relative power levels. The outputs of each part D ₁ and D ₂ are combined using the 3dB coupler 28 to complete the A ® C = AC + AC function. The waveforms illustrating the operation of the XOR gate are shown in Figure 21.

It is to be appreciated that the above-described functions of the SRL device are exemplary, and variations can be developed. In particular, a continuous wave optical signal can be input to the first input, and hence the clockwise direction, and the input optical signal can be supplied to the counter clockwise direction via the second input.

In all functions where the wavelengths of the inputs are not specified, it can be appreciated that they may be essentially the same as aligned to one ring cavity resonance, and they may also be essentially different in that each input to the device may be aligned to a different ring cavity resonance.

The functions described above serves as examples of the all-optical digital functions that may be realised using the semiconductor ring laser based all-optical regenerator device. These by no means form an exhaustive list of the possible functions provided by the device of this invention, and the skilled can infer further functions based on the same principles of operation.