JP2004254345A - Optical/optical and electric/optical signal converter and signal converting method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a small-sized and low power consumption optical/optical type serial/parallel converter and electric/optical type parallel/serial converter which is appropriate for a high-speed optical signal processing apparatus. <P>SOLUTION: The optical/optical type serial/parallel converter includes a branching filter for branching an input optical packet signal into (k) parallel optical signals, an optical delay unit for successively delaying the (k) optical signals bit by bit, a polarized beam splitter 208 where the (k) parallel optical signals retarded bit by bit and a circularly polarized control optical pulse are passed, a λ/4 wavelength plate 209 disposed all over the surface of one or two output sides of the polarized beam splitter or only in a middle portion where the control optical pulse is passed, a lens 210 for converging light transmitted through the polarized beam splitter and the λ4 wavelength plate to one point, and a reflection type surface type optical switch 211 for receiving light converged by the lens. The reflection type planar optical switch 211 is used, thereby extremely small-sized converter and reduction in power consumption as a whole can be realized. <P>COPYRIGHT: (C)2004,JPO&NCIPI

Description

  The present invention relates to an optical-optical type and an electrical-optical type signal conversion device and a signal conversion method for inputting an optical pulse train such as an optical packet or the like and converting the parallel electric signal into one optical signal.

  2. Description of the Related Art In recent years, with the explosion of data communication represented by the Internet, a demand for faster optical signals has been increasing. However, after converting an optical signal into an electric signal by a light receiving element, there is a problem that an electric signal of 10 Gbps or more is directly processed by a conventional electronic circuit. For example, in optical packet communication, a router or the like decodes address information contained in a label of an optical packet to determine a label output port and a packet signal for avoiding collision between optical packets. Although a buffer memory function that delays by an arbitrary time is required, conventionally, since the functions of the label recognition processing and the memory processing are configured by a silicon-based LSI, the operation speed is 1 Gbps or less. .

  Therefore, it is becoming difficult to realize the label recognition processing and the memory processing for a high-speed optical packet signal using a conventional silicon-based electronic circuit.

  In recent years, as shown in FIG. 14, a high-speed optical packet signal is converted into an electric signal by an O / E (optical / electric) receiving circuit 1 using a light-receiving element, and the electric signal is converted to a high-speed InP or GaAs-based electronic device. A clock signal is extracted by an electric clock signal generator 2 using circuit technology, and a high-speed electric signal is parallel-converted into a plurality of low-speed electric signals in an electric serial-parallel converter 3 by the clock signal. On the other hand, in the memory processing, the parallel-converted electric signals are stored in the memory cell array 4 of the SRAM, and a plurality of low-speed output electric signals read from the memory cell array 4 are also read when reading the electric signals. The electrical parallel-to-serial converter 5 using electronic circuit technology re-converts to a high-speed serial electrical signal. Forms, how to convert the optical packet signal is considered by the end of this high-speed serial electrical signal E / O (electrical / optical) transmission circuit 6.

  However, in such a method, the clock generation, the serial-parallel conversion, and the inverse conversion are all dependent on the electronic circuits 2, 3, and 5, so that a speed of about 40 Gbps is considered to be the limit. Furthermore, when serial-parallel conversion is performed by InP or GaAs-based high-speed electronic circuit technology to convert a plurality of low-speed signals, a high-speed electric signal is sequentially frequency-divided by half (40 GHz → 20 GHz →... →). (Several 100 MHz), which requires a considerable number of stages, and also causes problems such as clock extraction and phase control in each stage. Also, when these electronic circuits are used, the overall power consumption is expected to be considerably large. Moreover, in the conventional clock extraction using an electronic circuit, it is necessary to apply feedback by a PLL (Phase Locked Loop) to lock the oscillation frequency of a VCO (Voltage-Controlled Oscilltor). It is impossible to instantaneously extract a clock from a packet signal input in a burst.

  On the other hand, independently of the above system, some researches on parallel conversion of high-speed serial optical signals (also referred to as spatial conversion time) have been performed. As a conventional optical serial-to-parallel conversion method, a method in which a high-speed optical signal is split into a plurality of signals and each optical signal is converted into a low-speed optical signal by using a high-speed optical-optical switch can be considered. For example, in the case where a high-speed optical signal of 100 Gbps is parallel-converted into ten low-speed optical signals of 10 Gbps, ten optical-optical switches are used.

As other optical serial-parallel conversion methods, a method using a plurality of surface-emitted second-harmonic generations (Non-Patent Document 1), a method using an exciton-like giant nonlinear effect (Non-Patent Document 2), and a hologram are used. Method (Non-Patent Document 3).
Shih-Chen Wang et a1., J. Lihgtwave Techno1. Vo1.14, No.12, P.2736 (l996) K. Ema et a1., App1. Rhys. Lett. Vol. 59, No. 25, p. 2799 (1991) PCSuneta1., 0pt.Lett.Vo1.20, No.16, p.1728 (1995)

  However, the conventional method using a plurality of light-to-light switches for the light serial-to-light parallel conversion has a problem that the device is considerably large and the power consumption is large. Further, the conventional method using the surface-emitted second harmonic generation has a problem that the efficiency is extremely low and the loss is extremely large because the non-resonant optical nonlinear effect is used. Further, the conventional method using the exciton-like giant nonlinear effect has a problem that it is necessary to cool the nonlinear medium to a liquid helium temperature in order to obtain a large nonlinear effect. Further, the conventional method using a hologram has a problem that the loss is extremely large because the diffraction effect is used. Therefore, such conventional methods all have extremely high running costs, are inefficient, and have a problem that it is very difficult to maintain stable performance over a long period of time.

  Therefore, an object of the present invention is to solve the above-mentioned problems of the prior art, in which an input high-speed optical packet signal performs its own conversion to a low-speed parallel optical signal by various silicon-based electronic circuits. An object of the present invention is to provide an optical-optical type and an electrical-optical type signal converter and a signal conversion method for realizing optical signal processing with low power consumption and a relatively simple configuration.

  According to one embodiment of the present invention, an input high-speed optical packet converts itself into a low-speed parallel signal, thereby enabling signal processing in a silicon-based electronic circuit having a low operation speed but high functionality. Things. In the above-described conventional electronic circuit, a clock is extracted and the frequency division of the optical packet signal is repeated (10 GHz → 5 GHz, 2 lines → 2.5 GHz, 4 lines →...) To convert the signal into a low-speed parallel electric signal. Therefore, clock division and timing adjustment are required as needed. On the other hand, in the present invention, a single optical pulse is generated based on the first bit of the optical packet signal, and thereby, a part or all of the optical packet is parallel-converted collectively in the light state. With a very simple configuration, high-speed optical signal processing can be realized.

  Further, the present invention can operate independently of the polarization state of an incoming optical packet, and can always keep the polarization state of an output optical pulse constant (this characteristic is based on the optical-optical serial-parallel conversion). Essential to the vessel).

  A method of batch-converting a part or the whole of an optical packet at a time is performed by converting a single optical pulse generated by the above method and a parallel optical packet signal branched into k signals and sequentially shifted in phase by 1 bit. This is realized by irradiating one point of the type optical switch and increasing the transmittance or the reflectance at that point. In this method, since there is no external power supply and only one point of the surface type optical switch is used, it is possible to operate at an extremely small light intensity. A converter can be configured. Furthermore, by arranging a bundle of a plurality of optical fibers at each one input port, it is possible to operate the surface type optical switch at multiple points, and use the same optical-optical serial-parallel converter. In this case, the number of parallel conversions can be greatly increased. Further, in the method based on the conventional electronic circuit, a plurality of parallel converters are required to process a plurality of optical packets. However, according to the present method, the parallel conversion of a plurality of optical packets is performed by one device. Can be performed independently and simultaneously.

  According to another aspect of the present invention, the single optical pulse is converted into an optical pulse train by a loop-shaped optical waveguide, and the optical-optical serial-parallel converter converts the optical packet into k bits sequentially and collectively. Is repeated, the entire optical packet is converted into a low-speed k optical signal train, converted into a k electric signal train by a low-speed photoelectric converter, and processed by a silicon-based electronic circuit. This method is effective when the optical packet length is long and it is desired to process the whole.For example, by using a silicon-based electronic memory as the electronic circuit, it is possible to freely write an extremely high-speed optical packet signal to the electronic memory. It becomes. Further, the k parallel electric signals read from the electronic memory are reconstructed again into a high-speed optical packet signal by the electro-optical type parallel-serial converter of the present invention and output.

  As described above, according to the present invention, it is possible to configure an optical-optical and electro-optical signal conversion device and method suitable for a simple, compact, high-speed optical signal processing device with low power consumption. It becomes.

  Further, according to the present invention, it is possible to cope with a burst signal, and it is also possible to freely perform a partial processing or a whole processing of a high-speed optical packet.

  Further, according to the present invention, by combining the above-described techniques such as serial-parallel conversion, optical clock pulse generation, and parallel-serial conversion, and using a label recognition circuit or an electronic memory as an electronic circuit, the future extremely high speed is realized. High-speed optical label recognition (a function to read address information of optical packets and determine output ports), which is indispensable as a router function, and a high-speed optical buffer memory (to prevent optical packets from colliding at output ports, (A function of temporarily saving), optical bit rate conversion (conversion of high-speed optical packets to low-speed optical packets, or vice versa), and the like.

  Hereinafter, the best mode for carrying out the present invention will be described in detail with reference to the drawings.

[Optical-optical serial-parallel converter]
FIG. 1 is a diagram showing a configuration of an optical-optical serial-parallel converter 200 according to an embodiment of the present invention. In FIG. 1, reference numeral 201 denotes an optical demultiplexer that demultiplexes an input optical serial signal into k signals, 202 denotes an optical delay device that sequentially delays each demultiplexed parallel optical signal by one bit, and 203 denotes a condensing lens. Reference numeral 204 denotes a transmission-type surface switch (light-light switch), and reference numeral 205 denotes a condenser lens.

  The input optical packet is divided into k packets by the demultiplexer 201, and each packet is delayed so as to be shifted by one bit. At this time, focusing on a certain time timing of the k parallel optical signals, the optical pulse of k consecutive bits of the original optical packet is included one by one. Although the k parallel optical signals are represented two-dimensionally in FIG. 1, there is no problem if they are arranged two-dimensionally in the direction perpendicular to the plane of the drawing. Assuming that the array is arranged in a matrix of 5 × 5 in the direction perpendicular to the paper, only the central one is a control light pulse, and the remaining 24 around it are split by a splitter 201. This is a parallel optical signal. These twenty-four parallel optical signals and one control light pulse are condensed by the front condenser lens 203 at one point on the surface type optical switch 204.

  When a control light pulse is applied to the surface type optical switch 204, the surface-type optical switch 204 transmits light when the absorption coefficient decreases in the active layer and the transmittance increases because the refractive index in the active layer changes. An etalon type or the like whose rate changes is used. In any case, when the control light pulse (optical clock) is “0”, the transmittance of the surface-type optical switch 204 is extremely low, and these parallel optical signals cannot pass, but the control light pulse is “1”. In this case, the transmittance increases so that those parallel optical signals can pass.

  Since the 24 parallel optical signals spatially arranged are delayed by one bit for each port, the original optical signal continues at the timing when the control optical pulse reaches the optical switch 204. The 24-bit data arrives at the optical switch 204 in parallel at the same time. Therefore, when the transmittance of the optical switch 204 increases due to the control light pulse, the 24-bit data of all ports simultaneously passes through the optical switch 204 and is spatially developed again by the rear lens 205 in a spatially parallel manner. This means that the information of the high-speed optical signal (optical packet) is parallel-converted into 24 low-speed parallel optical signals synchronized with the control optical pulse. The period of the control light pulse at this time is 24 times the bit period of the original optical packet.

  FIG. 2 is a diagram illustrating an embodiment of an optical serial-to-parallel converter using a reflection type surface optical switch 211 as the surface type optical switch. A change in the absorption coefficient or refractive index in the active layer is enhanced by depositing a 100% mirror on one surface (the left side surface in FIG. 2) of the surface type optical switch 204 shown in FIG. Thus, it is possible to realize the reflection type surface optical switch 211 having a large extinction ratio. Here, one control light A and a plurality of signal lights B are incident from many optical fibers 207 attached to individual microlens arrays 206. Reference numeral 208 denotes a polarizing beam splitter (PBS) having the microlens array 206 attached to the light incident surface, and reference numeral 209 denotes a λ / 4 wavelength plate arranged on the PBS 208 on the side of the condenser lens 210.

  By setting the control light A and each signal light B to linearly polarized light passing through the PBS 208, the signal light B is reflected on the reflective surface type optical switch 211 and then reciprocates through the λ / 4 wavelength plate 209. , The signal light B is reflected by the PBS 208 and extracted as a parallel optical pulse train.

  However, in the case of this method, even when the control light A is “0”, a part of the signal light that is not completely absorbed by the surface light switch 211 may be reflected by the surface light switch 211 and output. In order to increase the On / Off ratio of the signal light B (the ratio of the intensity of the output signal light when the control light is “1” and the intensity of the output signal light when the control light is “0”), the absorption of the surface-type optical switch 211 is required. It is necessary to increase the change in the coefficient, and for that purpose, it is necessary to increase the intensity of the control light A. Therefore, in order to increase the On / Off ratio of the signal light B without increasing the control light A too much, it is essential to suppress the output of the signal light B when the control light A is "0" as much as possible. .

  Therefore, here, as shown in FIG. 3, the parallel optical signal and the control light obtained by demultiplexing the optical signal are preset to linearly polarized light passing through the PBS 208, and only the central control light is λ / The size and position of the λ / 4 wavelength plate 209 are set so as to pass through the four wavelength plate 209. For this reason, only the control light is converted into circularly polarized light by the λ / 4 wavelength plate 209, and the other parallel optical signals are irradiated to one point of the reflection type surface optical switch 211 by the lens 210 in a state of being linearly polarized. You. The operation principle at this time will be described below using an absorption-saturated surface optical switch as an example.

  Electrons and holes exist in which two states of up spin and down spin degenerate with respect to one energy state. Now, considering the exciton transition between an electron and a heavy hole, when a circularly polarized control light is incident on one point of a multiple quantum wells layer (MQW layer) of the optical switch 211, one of the spins is changed. Only if a parallel optical signal of linear polarization is incident, both spins will be excited. As a result, when only one of the spins is excited by the circularly polarized control light “1”, only the parallel optical signal in the polarization state interacting with the spin feels a change in absorption and refractive index. That is, in this case, when the parallel optical signal of linear polarization is irradiated, only the same circularly polarized light component as the control light is modulated by the control light, and the parallel optical signal reflected by the surface type optical switch 211 becomes an elliptically polarized light signal. Therefore, the light is reflected by the PBS 208. On the other hand, when the control light is “0”, the parallel optical signal cannot change its polarization state, and therefore returns to the original port without being reflected by the PBS 208, and the output light is almost “0”. ".

The switching speed of the planar optical switch 211 according to this method is determined by the shorter of the spin relaxation time and the carrier lifetime. In the case of a multi-quantum well layer grown at 500 degrees Celsius, which is generally used as an active layer, since the carrier lifetime is on the order of nanoseconds, the switching speed can be improved to a spin relaxation time of several tens ps. On the other hand, when a quantum well layer with a carrier lifetime of 10 ps or less is used by growing at a low temperature of about 200 degrees Celsius and adding a p-type element or Be as a dopant at 10 ³⁷ cm ⁻³ or more, an even faster surface light can be obtained. It becomes possible to make a switch.

  By the way, in the above method, the polarization state of the parallel optical signal is limited to linearly polarized light. However, when considering applications in actual optical communication and the like, it is necessary to operate on an optical signal having an arbitrary polarization state. Therefore, in FIG. 4, the reflection type surface optical switches 211 and 211A, the λ / 4 wave plates 209 and 209A, and the lenses 210 and 210A are respectively arranged on two adjacent sides of the PBS 208, and the control light is circularly polarized. , The polarization independence of the parallel optical signal is enabled.

  The central circularly polarized control light is split into two orthogonal linearly polarized lights of the same intensity by the PBS 208, and becomes circularly polarized again by passing through the λ / 4 wavelength plates 209 and 209A, respectively. To reflect the parallel optical signal. In this case, since the control light intensities applied to the surface type optical switches 211 and 211A are completely equal, the reflectances of both are equal.

  Therefore, the parallel optical signal incident with an arbitrary polarization is split by the PBS 208 according to the polarization state and reflected by the optical switches 211 and 211A. The sum of the parallel optical signal intensities respectively reflected by 211 and 211A is always equal, and is multiplexed again by PBS 208 and output. That is, this makes it possible to realize polarization independence for the optical signal.

  In the method of FIGS. 1 to 4, it is necessary to increase the number of microlens arrays 206 in order to increase the number of parallel conversions, and accordingly, the ratio of the beam diameter of each input light to the diameter of the condenser lens 210 is reduced. Since the spot size becomes smaller, the size of the spot when condensed by the condensing lens 210 increases, and it may be necessary to increase the energy of the control light pulse in order to obtain the same power density. In addition, since all optical signals are concentrated on the same spot, nonlinear effects such as a saturable absorption effect occur only with the optical signals, and the effect of irradiating the control light may be weakened. Therefore, there is a possibility that the effect of the generated heat may be large.

  Therefore, as shown in FIG. 5, a method is considered in which L optical fibers are bundled from one lens of the microlens array 206 and an optical signal is incident. Now consider the case of L = 2. When the optical fibers of A1 and A2 are arranged close to the input port of the control light pulse, the control light pulses emitted from each of them propagate with a slight difference in angle. The light is collected at two points. Similarly, if the optical fibers for B1 and B2 are arranged close to each other at the port for inputting the optical packet signal, the optical packet signals input from B1 and B2 are focused on the same spot as A1 and A2, respectively. As shown in FIG. 6, the branched optical packet signal is delayed one bit at a time, and the two control light pulses are simultaneously transmitted at the same time when all the bits are aligned. Is irradiated. Light reflected at two different points on the surface-type optical switch 211 propagates again at different angles, and is collected at two different points by the microlens array 206 arranged on the output side. The light receiving element array 106 is arranged so as to be able to independently receive these parallel-converted light pulses. According to this method, the number of parallel conversions can be increased L times without increasing the size of the microlens array.

  On the other hand, in the conventional electrical method, the same number of electrical serial-to-parallel converters were required to process a plurality of optical packets, but using this method, as shown in FIG. , And B2, the other optical packet signals are input, so that the parallel conversion processing of a plurality of optical packet signals can be performed independently and simultaneously using one optical-optical serial-parallel converter. .

  An optical signal processing apparatus for performing optical serial-optical parallel conversion, for example, as shown in FIG. 8 (corresponding to FIG. 1), by reversing the optical signal and control light input ports shown in FIGS. It becomes possible. That is, in FIG. 8, two types of optical signals are used, a high-speed signal light A and a low-speed repetitive probe light B. In FIG. 1, a serial optical signal A is input from a port for inputting control light, and an optical signal is input in FIG. A parallel probe light B (hereinafter, referred to as a probe light instead of a control light due to a difference in action) is input from a port to be connected, so that one serial signal light A is converted into a serial-parallel signal by a plurality of probe lights B. Convert. At this time, the light pulse train of the signal light A and the probe light B is condensed by the condensing lens 203 to one point on a transmission type surface optical switch 204 having a semiconductor multiple quantum well layer, for example, as in FIG. Is done.

  The plurality of probe lights B spatially arranged are delayed by one bit of the signal light A for each port, and have different phases. If the number of the probe lights B is k, the cycle of each probe light B is k times of 1 bit of the signal light A. Therefore, a certain bit of the signal light A is the first probe light, the second bit is the second probe light, the third bit is the third probe light,..., The k-th bit is the k-th probe light. Is synchronized with For this reason, when the bit of the signal light A synchronized with the probe light B is “1”, the signal light increases the transmittance of the surface-type optical switch 204, so that the probe light becomes “1” and the probe light becomes “1”. When the bit of the signal light A synchronized with B is “0”, the probe light becomes “0” and appears on the lens 205 side.

  In this way, the probe light B of each port is sequentially modulated in transmittance by the multiple quantum well layer of the surface type optical switch 204 by the signal light A, and is developed spatially in parallel again by the rear condenser lens 205. . Thus, the information of the high-speed signal light A is parallel-converted into a plurality of lower-speed optical signals.

  Also in the configurations of FIGS. 2 to 7, it is possible to perform optical serial-optical parallel conversion in the same manner as described above. Furthermore, by integrally configuring the duplexer 201, the optical delay unit 202, and the optical fiber array 207 with a glass waveguide such as a PLC, it is possible to further reduce the size as a whole.

  In the above-described embodiment, since only one point of the surface type optical switch 204 or 211 is used, only one or two surface type optical switches 204 or 211 are required. An element can be manufactured.

  The optical-optical serial-parallel conversion uses a transmission-type or reflection-type surface-type optical switch having a semiconductor multiple quantum well layer, thereby realizing extremely small size and low power consumption as a whole. Even in the electric-optical parallel-serial conversion, an extremely simple configuration is used as described later. As described above, the configuration in which light is actively used for the input / output portion and the Si-based electronic memory circuit is used for the memory portion enables a very high-speed burst optical signal to be handled, and a large capacity, small size, An optical memory with low power consumption can be realized.

[Optical memory]
Next, an optical memory using the optical-optical serial-parallel converter and the electro-optical parallel-serial converter of the present invention will be described. In this example, as in the above-described embodiment, the first bit of the input optical packet is always set to “1”, and each time an optical packet is input, a single optical pulse is always generated at the same timing. This is put into a loop-shaped optical waveguide, and an optical pulse train having a period k times the bit period of the optical packet is generated. Further, using the optical pulse train, an input optical packet is converted into k parallel optical signals by an optical-optical serial-parallel conversion, and the k parallel optical signals are converted into k parallel optical signals by a light receiving element. After being converted into an electric signal, it is simultaneously written into k memory circuits constituted by electronic circuits. For reading from the k memory circuits, k parallel electric signals are simultaneously read, converted into one optical pulse train by electro-optical parallel-serial conversion, and output as an optical packet.

  The optical-optical serial-parallel conversion uses a transmission-type or reflection-type surface-type optical switch having a semiconductor multiple quantum well layer, thereby realizing extremely small size and low power consumption as a whole. Even in the electric-optical parallel-serial conversion, an extremely simple configuration is used as described later. As described above, the configuration in which light is actively used for the input / output portion and the Si-based electronic memory circuit is used for the memory portion enables a very high-speed burst optical signal to be handled, and a large capacity, small size, An optical memory with low power consumption can be realized. The details will be described below.

  FIG. 9 is a diagram showing the overall configuration of an optical memory device including the light-light type and electric-light type signal conversion devices of the present invention as constituent elements. The input optical packet P is split into two by a demultiplexer (not shown), and one of the two is input to an optical pulse train generator (not shown), where it has a low speed of k times the bit period of the original optical packet. An optical pulse train is generated. The optical pulse train is input to the above-mentioned optical-optical serial-parallel converter 200, where the other optical packet is converted into k low-speed parallel optical pulse trains (having a period k times the bit period of the original optical packet). Convert to a signal. This parallel optical pulse train signal is output simultaneously for each k bits. Each of the k-bit parallel optical pulse train signals is converted into an electric signal by the low-speed light-receiving element array 300, and the electric signal is converted into one electric signal in k arrays of the Si-based memory cell array 401 in which the column and row addresses are shared. Are held at the same time.

  As shown in FIG. 10, an Si-based memory cell array 400 (hereinafter referred to as a memory array) has k memory cell arrays 401 corresponding to k light receiving elements 301 included in the light receiving element array 300. Then, reading and writing are performed by the column address circuit 403 and the row address circuit 404 controlled by the control circuit 402. That is, when one address is designated by both address circuits 403 and 404, k data are simultaneously written into each memory cell array 401. As described above, the optical packet is converted into parallel data by k bits one by one while the optical packet is continued by the optical pulse train generated by the optical pulse train generator 100 and written into the memory array 400. It will be written to the memory circuit. At this time, the bit rate of the optical packet signal can be increased up to k times the writing speed of the memory cell array 401.

  When reading data written in each memory cell array 401, k data is simultaneously output every time an address is designated by both address circuits 403 and 404. The electric data read out in this way is output again as one high-speed optical packet signal by the electric-optical parallel-serial converter 500.

  As a method of sending k output electric signals from the memory array 400 to the electro-optical type parallel-serial converter 500, as shown on the left side of FIG. 11, the surface-emitting laser 405 converts the electric signals into optical signals. Thereafter, the contact point 406 and the contact point 502 are adhered by an optical I / O method in which the light signal is converted again into an electric signal by the light receiving element 501 or a solder bonding method as shown on the right side of FIG. A sending method can be used.

[Electrical-optical parallel-serial converter]
FIG. 12 is a diagram showing a configuration example of the above-mentioned electro-optical type parallel-serial converter 500 according to the present invention. An optical pulse light source 503 oscillates a pulse having a period k times the bit period of the original optical packet. Reference numeral 504 denotes a demultiplexer, 505 denotes k optical modulators, 506 denotes an optical delay device, and 507 denotes an optical multiplexer. Of these, the waveguide portion is constituted by PLC. Note that the optical delay unit 506 may be provided between the demultiplexer 504 and the optical modulator 505.

  In the memory array 400, when an address of written data is designated and read, k low-speed electric signals are simultaneously output from k memory cell arrays 401 whose addresses are shared. These k parallel electric signals are supplied to k optical modulators 505, respectively. The light pulse from the light pulse light source 503 is divided into k pulses by the demultiplexer 504, and when passing through the k light modulators 505, the light pulse is modulated by the k electric signals and modulated. The optical pulses are delayed by one bit by the optical delay unit 506 and then combined by the multiplexer 507. As described above, the k low-speed parallel electric signals are reconstructed into one high-speed optical packet whose bit period is 1 / k times again.

  FIG. 13 is a diagram illustrating another configuration example of the electro-optical type parallel-serial converter 500. In the electro-optical type parallel-serial converter 500 shown in FIG. 12, if k is large, a large amount of optical modulator is required.

  Therefore, when k is large as described above, as shown in FIG. 13, k parallel electric signals are bundled in units of n to create m parallel electric signal trains (k = n × m), Similarly to the above, the original high-speed optical packet signal is output again by the m optical modulators 505A. As a result, the number of required optical modulators can be significantly reduced (1 / n).

  To bundle the n parallel electric signals, n of the k parallel electric signals output simultaneously from the RAM array 400 are supplied to the drains of the n transistors 508 as a bias power supply. Although FIG. 13 shows only one set of n transistors 508 and related parts, m sets of such circuits are used. Then, an electric pulse is input to the gate of each transistor 508 to sample a part of the electric signal, and the sampling signal charges a capacitor 509 connected to the source of the transistor. The above-described transistor 508 and capacitor 509 constitute a sample-and-hold unit, where n electrical signals are sampled and held at the same time.

  Further, the optical clock (the cycle is k times the bit cycle of the original optical packet) is delayed one bit at a time by an optical delay unit 511 composed of a PLC or the like, and is sequentially irradiated on the n photoconductive switches 510. The electric charges accumulated in each capacitor 509 are sequentially discharged, and are converted by the load resistor 512 into an electric pulse train signal in which n pulses are continuous. The photoconductive switch 510, the optical delay unit 511, and the load resistor 512 constitute an electric-electric type parallel-serial conversion unit.

  The electric pulse train signal obtained as described above is input to one of the optical modulators 505A. In this case, k parallel signals are modulated by the optical modulator 505A in n divisions every m lines. The pulse cycle output from the optical pulse light source 503A used here is set to be m times the bit cycle of the original optical packet. In this way, k data are sequentially read, and the original optical packet is output as a whole.

  As described above, according to the above-described embodiment of the present invention, a high-speed burst optical packet signal is converted to a Si-based memory circuit (shown in the figure) by using the optical-optical serial-parallel converter 200 according to the present invention. No.), and using the electro-optical parallel-serial converter 500 according to the present invention, a large-capacity optical memory device that can be freely read again as a high-speed optical packet can be realized with small size and low power consumption.

  As can be understood from the above description, the present invention realizes functions such as label processing, optical memory, and optical bit rate conversion by combining the above-described serial-parallel conversion and parallel-serial conversion techniques according to the present invention. It is possible to realize high-performance optical information processing devices or systems, such as high-performance routers and optical computers, by combining them, and is expected to greatly contribute to the development of industry. it can.

FIG. 3 is an explanatory diagram of one embodiment of the present invention that performs serial-parallel conversion of an optical signal sequence. FIG. 3 is an explanatory diagram of one embodiment of the present invention that performs serial-parallel conversion of an optical signal sequence. FIG. 3 is an explanatory diagram of one embodiment of the present invention that performs serial-parallel conversion of an optical signal sequence. FIG. 3 is an explanatory diagram of one embodiment of the present invention that performs serial-parallel conversion of an optical signal sequence. FIG. 3 is an explanatory diagram of one embodiment of the present invention that performs serial-parallel conversion of an optical signal sequence. FIG. 3 is an explanatory diagram of one embodiment of the present invention that performs serial-parallel conversion of an optical signal sequence. FIG. 3 is an explanatory diagram of one embodiment of the present invention that performs serial-parallel conversion of an optical signal sequence. FIG. 3 is an explanatory diagram of one embodiment of the present invention that performs serial-parallel conversion of an optical signal sequence. 1 is a block diagram illustrating a configuration example of an optical random access memory device using an optical-optical serial-parallel converter and an electrical-optical parallel-serial converter according to the present invention. FIG. 3 is a detailed explanatory diagram of a Si-based RAM array part. FIG. 4 is an explanatory diagram of a method of transferring electric data output from a Si-based RAM array to an electric-optical parallel-serial converter. FIG. 1 is an explanatory diagram of an electro-optical type parallel-serial converter according to an embodiment of the present invention. It is explanatory drawing of the electro-optical type parallel-serial converter of another example in one Embodiment of this invention. FIG. 11 is a block diagram illustrating a configuration example of a conventional optical packet communication system.

Explanation of reference numerals

DESCRIPTION OF SYMBOLS 1 O / E receiving circuit 2 Electric clock signal generator 3 Electric serial-parallel converter 4 Si memory cell array 5 Electric parallel-serial converter 6 E / O transmission circuit 104 Optical-optical serial-parallel converter 200 Optical-optical Type serial-parallel converter 201 duplexer 202 optical delay device 203,205 condensing lens 204 transmission type surface type optical switch 206 micro lens array 207 optical fiber (optical fiber array)
208 Polarizing Beam Splitter (PBS)
209 λ / 4 wavelength plate 210 Condensing lens 211 Reflective surface type optical switch 300 Light receiving element array 301 Light receiving element 400 Si-based RAM array 401 Memory cell array 402 Control circuit 403 Column address circuit 404 Row address circuit 405 Surface emitting laser 406 502 contacts 500 electro-optical parallel-serial converter 501 light receiving element 503, 503A optical pulse light source 504, 504A demultiplexer 505, 505A optical modulator 506, 506A optical delay unit 507, 507A multiplexer

Claims (8)

A demultiplexer for demultiplexing an input burst optical packet signal into k parallel optical signals;
An optical delay unit for sequentially delaying each of the k optical signals demultiplexed by the demultiplexer by one bit;
A polarizing beam splitter through which k parallel optical signals delayed by one bit by the optical delay unit and control light pulses of circularly or linearly polarized light pass;
A λ / 4 wavelength plate disposed on the entire surface on one or two output sides of the polarizing beam splitter, or only on a portion through which a central control light pulse passes;
A lens for condensing the transmitted light of the polarization beam splitter and the λ / 4 wavelength plate at one point,
And a reflective surface type optical switch for receiving the light condensed by the lens. A demultiplexer for demultiplexing the input burst optical packet signal into k × L parallel optical signals;
An optical delay unit for sequentially delaying k × L optical signals demultiplexed by the demultiplexer by one bit,
A demultiplexer that demultiplexes the input control light pulse, which is the single light pulse, into L light beams;
A total of (k + 1) × L optical waveguides for propagating the k × L optical signals and the L control optical pulses that are demultiplexed by the demultiplexer and delayed by the optical delay device;
A lens array including k + 1 lenses for converting the light output from the optical waveguide into parallel light;
A polarization beam splitter through which the k parallel optical signals delayed by one bit and the control light pulse of circular polarization or linear polarization pass;
A λ / 4 wavelength plate disposed on the entire surface on one or two output sides of the polarizing beam splitter, or only on a portion through which a central control light pulse passes;
A lens for condensing light transmitted through the polarizing beam splitter and the λ / 4 wavelength plate;
A reflective surface-type optical switch that receives light collected by the lens;
And a lens array configured by k + 1 lenses for separating and condensing the light reflected by the surface-type optical switch and output from the polarization beam splitter. A parallel converter. L demultiplexers for demultiplexing each of the L burst optical packet signals input in parallel into k parallel optical signals;
L optical delay devices for sequentially delaying each of the k optical signals demultiplexed by the demultiplexer by one bit;
A demultiplexer that demultiplexes the input control light pulse, which is the single light pulse, into L light beams;
A total of (k + 1) × L optical waveguides for propagating the k × L optical signals and the L control optical pulses that are demultiplexed by the demultiplexer and delayed by the optical delay device;
A lens array including k + 1 lenses for converting the light output from the optical waveguide into parallel light;
A polarization beam splitter through which the parallel optical signal and the circularly or linearly polarized control light pulse pass,
A λ / 4 wavelength plate disposed on the entire surface on one or two output sides of the polarizing beam splitter, or only on a portion through which a central control light pulse passes;
A lens for condensing light transmitted through the polarizing beam splitter and the λ / 4 wavelength plate;
A reflective surface-type optical switch that receives light collected by the lens;
And a lens array configured by k + 1 lenses for separating and condensing the light reflected by the surface-type optical switch and output from the polarization beam splitter. A parallel converter. m (m = k / n) sets of sample-and-hold units for sampling k parallel electric signals in units of n;
K photoconductive switches for taking out the electric charges sampled and stored in units of n units in the sample and hold unit as one electric pulse train signal,
An optical pulse light source,
A demultiplexer for demultiplexing the optical signal output from the optical pulse light source into m optical signals;
M optical modulators for modulating the m parallel optical signals split by the splitter with the m parallel electrical signals output from the photoconductive switch;
An optical delay unit that delays the m parallel optical signals by one bit at an input side or an output side of the m optical modulators;
A multiplexer for combining the m parallel optical signals delayed by the optical delay unit into one optical pulse train to form an optical packet signal. . The control light pulse of the circularly polarized light or the linearly polarized light is passed through a polarizing beam splitter to be split into two linearly polarized lights transmitting and reflecting 90 degrees, each of which is converted into a circularly polarized light again by a λ / 4 wavelength plate. Then, the light is irradiated to one point of the surface type optical switch by the condenser lens, and the reflectance at that point is modulated.
On the other hand, the input optical packet signal is branched into k pieces, and the phases are sequentially shifted by one bit, and the k optical packets are spatially parallel passed through the polarizing beam splitter. Irradiate the same point on the surface type optical switch,
Only the optical pulse in the optical packet that has entered the surface-type optical switch at the same timing as the control optical pulse is reflected by the surface-type optical switch, thereby converting a part or all of the optical packet in the optical packet into parallel. A light-light type serial-parallel conversion method. The control light pulse of the circularly polarized light or the linearly polarized light is branched into L light beams, each of which is spatially parallel and propagates with a slight difference in angle. Modulate the transmittance or reflectance at those points,
On the other hand, an input optical packet signal is branched into k × L signals, the phases of which are sequentially shifted by 1 bit, and the k × L optical packets are propagated spatially in parallel at different angles (for each k packets). The same angle), irradiating the same L points of the surface type optical switch with the condenser lens,
Only the optical pulse in the optical packet that has entered the surface-type optical switch at the same timing as the control optical pulse is reflected by the surface-type optical switch, thereby converting a part or all of the optical packet in the optical packet into parallel. A light-light type serial-parallel conversion method. L control light pulses of different circularly polarized light or linearly polarized light are respectively propagated in a spatially parallel manner with a slight difference in angle, and irradiate different L points of the planar optical switch with a condenser lens, Modulate the transmittance or reflectance at those points,
On the other hand, L different optical packet signals that are input independently are respectively branched into k signals, and the phases are sequentially shifted by 1 bit, and these k × L optical packets are spatially different in parallel from packet to packet. Propagate at an angle, and irradiate the same L points of the surface type optical switch by the condenser lens,
Only the optical pulse in the optical packet that has entered the surface-type optical switch at the same timing as the control optical pulse reflects the surface-type optical switch, thereby parallelizing some or all of the optical pulses of the plurality of optical packets. A light-optical serial-parallel conversion method. A part of k parallel electric signals is sampled and stored as a charge in a capacitor,
By sequentially discharging the charges accumulated in the sample-and-hold unit by the n photoconductive switches or transistors, an n-bit electric pulse train signal is output (m = k / n),
The optical pulse output from the optical pulse light source is split into m light pulses, and is passed through an optical modulator to which the m electric pulse trains are applied, thereby converting the light pulses into m optical signal trains and delaying them by one bit. An electrical-optical parallel-serial conversion method, wherein the signal is converted into one high-speed optical signal sequence by combining the signals with a multiplexer again.

Images