Optical packet header identification system, integrated chip and identification method

Abstract

This application discloses an optical Baotou identification system, integrated chip, and identification method. The system comprises an input layer, an optical reservoir module, and an output layer, connected in sequence. The optical reservoir module comprises a plurality of reservoir nodes connected in a predetermined manner, each of which includes a microring resonator. This application employs microring resonators as reservoir nodes, which act as a time delay. This system features low power consumption, a small footprint, and stable operation, enabling on-chip integration of the optical Baotou identification system and increasing its flexibility.

Description

Technical Field

[0001] The present application relates to the field of optical fiber communication technology, and in particular to an optical packet header identification system, an integrated chip, and an identification method. Background Art

[0002] With the rapid development of modern society, the massive growth of data has greatly promoted the development of communication technology. Optical communication networks offer numerous advantages, such as high bandwidth, large capacity, and low loss, making them irreplaceable in wired communication systems. Compared with existing optical circuit switching (OCS) networks, optical packet switching (OPS) networks have the potential for higher bandwidth utilization and lower power consumption. Routers in OPS networks can extract the optical packet header of an optical signal and identify the header type. Based on the header type, the optical signal is sent to the corresponding port. In other words, the identification of the optical packet header determines whether the routing information is correctly obtained, becoming a key technology in OPS networks.

[0003] Initially, the recognition of optical packet headers in OPS networks used recurrent neural networks (RNNs), a type of computer neural network. However, RNNs require a large number of neurons and layers during training for effective recognition. Training such a large number of parameters requires a large number of training sets and takes a long time, so it cannot meet the needs of real-time optical fiber communication systems.

[0004] In recent years, the use of optical neural networks for optical packet processing has become a hot topic. Single-node reservoir computing, a novel artificial neural network consisting of a single nonlinear node and a delay line, significantly reduces hardware costs compared to traditional neural networks. As a single-node reservoir computing (RC), optoelectronic feedback reservoir computing can be easily implemented using multiple optoelectronic devices. Through a very simple training process on its output layer, the RC system can achieve various pattern recognition and time series prediction tasks.

[0005] However, the current optical neural network processing of optical packet headers is based on the traditional optical delay line method. The construction of the reservoir uses vertical cavity surface emitting lasers (VCSELs), fiber delay lines, semiconductor ring lasers, etc. This reservoir structure occupies a large area and is difficult to achieve on-chip integration.

[0006] Therefore existing technology still needs to be improved and improved. Summary of the Invention

[0007] The technical problem to be solved by this application is to provide an optical packet header recognition system, an integrated chip and a recognition method in response to the deficiencies of the existing technology.

[0008] In order to solve the above technical problems, the first aspect of an embodiment of the present application provides an optical packet header recognition system, wherein the system includes an input layer, an optical reservoir module, and an output layer connected in sequence, wherein the optical reservoir module includes a plurality of reservoir nodes connected in a preset connection manner, and each reservoir node includes a microring resonator.

[0009] In the optical packet header recognition system, the preset connection mode is one of a series connection mode, a parallel connection mode, a series-parallel connection mode, a hybrid connection mode and a sparse connection mode.

[0010] In the optical packet header recognition system, the microring resonator is a single ring resonant cavity structure or a multi-ring resonant cavity structure, wherein the multi-ring resonant cavity structure is formed by at least two single microring resonant cavity structures according to a preset topology.

[0011] In the optical packet header recognition system, the preset topology is one of a cascade type, a parallel type, a nested type, and a three-coupler type.

[0012] In the optical packet header recognition system, the optical reservoir module includes at least one activation optical device with a power nonlinear response, and the activation optical device is connected to a reservoir node among the plurality of reservoir nodes.

[0013] In the optical packet header recognition system, the activation optical device is a semiconductor optical amplifier.

[0014] The optical packet header identification system, wherein the optical reservoir module includes a plurality of couplers, the plurality of couplers correspond one-to-one to the last reservoir node located in a preset manner among the plurality of reservoir nodes, and each coupler is connected after its corresponding reservoir node.

[0015] The optical packet header recognition system further comprises an optical domain masking unit, and the optical domain masking unit is connected to the input layer.

[0016] A second aspect of an embodiment of the present application provides an integrated optical chip for optical packet header recognition, characterized in that the integrated optical chip integrates the optical packet header recognition system described above.

[0017] A third aspect of the embodiments of the present application provides an optical packet header recognition method, characterized in that the method applies the optical packet header recognition system described above, and the method includes:

[0018] Inputting the optical packet header signal to be identified into the optical packet header identification system;

[0019] The optical packet header type corresponding to the optical packet header signal is determined by the optical packet header identification system.

[0020] Beneficial Effects: Compared to the prior art, this application provides an optical packet header recognition system, integrated chip, and recognition method. The system comprises an input layer, an optical reservoir module, and an output layer connected in sequence. The optical reservoir module comprises a plurality of reservoir nodes connected in a preset manner, each of which includes a microring resonator. This application employs microring resonators as reservoir nodes, which act as a time delay. This results in an optical packet header recognition system with low power consumption, a small footprint, and stable operation. This allows for on-chip integration of the optical packet header recognition system, improving its flexibility. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without inventive work.

[0022] Figure 1 This is a schematic diagram of the structure of the optical packet header identification system provided in this application.

[0023] Figure 2 This is a structural principle diagram of the serial mode of the optical packet header identification system provided in this application.

[0024] Figure 3 This is a structural principle diagram of the sparse connection method of the optical packet header identification system provided in this application.

[0025] Figure 4 Schematic diagram of the cascaded dual ring resonator structure.

[0026] Figure 5 Schematic diagram of the parallel dual-ring resonator structure.

[0027] Figure 6 Schematic diagram of the nested double-ring resonator structure.

[0028] Figure 7 Schematic diagram of the three-coupler double micro-ring resonator structure.

[0029] Figure 8 Schematic diagram of the Mask processing process. DETAILED DESCRIPTION

[0030] This application provides an optical packet header identification system, integrated chip, and identification method. To make the purpose, technical solution, and effects of this application more clear and explicit, the application is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only intended to explain this application and are not intended to limit this application.

[0031] It will be understood by those skilled in the art that, unless expressly stated otherwise, the singular forms "a", "an", "said" and "the" used herein may also include the plural forms. It should be further understood that the term "comprising" used in the specification of the present application refers to the presence of the features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof. It should be understood that when we refer to an element as being "connected" or "coupled" to another element, it may be directly connected or coupled to the other element, or there may be intermediate elements. In addition, "connected" or "coupled" as used herein may include wireless connections or wireless couplings. The term "and / or" used herein includes all or any units and all combinations of one or more associated listed items.

[0032] It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by those skilled in the art to which this application belongs. It should also be understood that terms such as those defined in common dictionaries should be understood to have meanings consistent with their meanings in the context of the prior art and will not be interpreted in an idealized or overly formal sense unless specifically defined as herein.

[0033] It should be understood that the sequence numbers and sizes of the steps in this embodiment do not imply the order of execution. The order of execution of each process is determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of this application.

[0034] After research, the inventors found that with the rapid development of modern society, the massive growth of data has greatly promoted the development of communication technology. Optical communication networks have many advantages such as high bandwidth, large capacity, and low loss, which make them play an irreplaceable role in wired communication systems. Compared with the existing optical circuit switching (OCS) network, the optical packet switching (OPS) network has the potential for higher bandwidth utilization and low power consumption. The routers in the OPS network can extract the optical packet header of the optical signal and identify the packet header type; based on the packet header type, the optical signal is sent to the corresponding port, that is, the identification of the optical packet header determines whether the routing obtains the information correctly, which becomes one of the key technologies of the OPS network.

[0035] Initially, the recognition of optical packet headers in OPS networks used recurrent neural networks (RNNs), a type of computer neural network. However, RNNs require a large number of neurons and layers during training for effective recognition. Training such a large number of parameters requires a large number of training sets and takes a long time, so it cannot meet the needs of real-time optical fiber communication systems.

[0036] In recent years, the use of optical neural networks for optical packet processing has become a hot topic. Single-node reservoir computing, a novel artificial neural network consisting of a single nonlinear node and a delay line, significantly reduces hardware costs compared to traditional neural networks. As a single-node reservoir computing (RC), optoelectronic feedback reservoir computing can be easily implemented using multiple optoelectronic devices. Through a very simple training process on its output layer, the RC system can achieve various pattern recognition and time series prediction tasks.

[0037] However, the current optical neural network processing of optical packet headers is based on the traditional optical delay line method. The construction of the reservoir uses vertical cavity surface emitting lasers (VCSELs), fiber delay lines, semiconductor ring lasers, etc. This reservoir structure occupies a large area and is difficult to achieve on-chip integration.

[0038] To address the aforementioned issues, in an embodiment of the present application, the system comprises an input layer, an optical reservoir module, and an output layer, all connected in sequence. The optical reservoir module comprises a plurality of reservoir nodes connected in a predetermined manner, each of which includes a microring resonator. This application employs microring resonators as reservoir nodes, which act as a time delay. This results in an optical packet header recognition system with low power consumption, a small footprint, and stable operation. This allows for on-chip integration of the optical packet header recognition system, improving its flexibility.

[0039] The application content will be further explained below through description of embodiments in conjunction with the accompanying drawings.

[0040] This embodiment provides an optical packet header recognition system, such as Figure 1As shown, the system includes an input layer, an optical reservoir module, and an output layer. The input layer is connected to the optical reservoir module, which is in turn connected to the output layer. The input layer, the optical reservoir module, and the output layer form an optical reservoir neural network structure, which performs optical packet header recognition. The optical reservoir module includes a plurality of reservoir nodes, which are connected according to a predetermined connection method. Each reservoir node includes a microring resonator. The microring resonator has a certain bandwidth and can also cause a delay in the optical packet header signal. Thus, the microring resonator can transmit optical packet header information and provide delay for the optical packet header signal. This embodiment uses microring resonators as reservoir nodes, making the optical reservoir module power-efficient and occupying a millimeter-scale footprint. Furthermore, the use of microring resonators as reservoir nodes makes the optical reservoir module easy to process and stable in operation. This allows the optical packet header recognition system provided by this embodiment to be integrated on a chip, increasing the flexibility of the optical packet header recognition system.

[0041] In one implementation, the preset connection mode can be one of a series connection mode, a parallel connection mode, a series-parallel connection mode, a hybrid connection mode, and a sparse connection mode. That is, a plurality of reservoir nodes can be a reservoir node series structure, a reservoir node parallel structure, a reservoir node series-parallel structure, a reservoir node hybrid connection structure, or a reservoir node sparse connection structure. For example, Figure 2 As shown, several reservoir nodes are connected in series, or as Figure 3 As shown, the plurality of water reservoir nodes are sparsely connected, etc. In addition, the number of the plurality of water reservoir nodes can be determined according to actual usage, for example, the number of the plurality of water reservoir nodes can be 20.

[0042] In one implementation, the microring resonator may adopt a single-ring resonant cavity structure or a multi-ring resonant cavity structure, wherein the multi-ring resonant cavity structure is formed by at least two single microring resonant cavity structures according to a preset topology. It is understandable that the multi-ring resonant cavity structure may be a dual-ring resonant cavity structure formed by two single-ring resonant cavity structures, or a triple-ring resonant cavity structure formed by three single-ring resonant cavity structures. In addition, when the microring resonator is a multi-ring resonant cavity structure, the topology and number of the multi-ring resonant cavity structures may be determined based on the transmission rate of the optical packet header to be processed, the length of the optical packet header, and the capacity of the single-shaped resonant cavity. This allows the structure of the optical reservoir module to be flexibly changed, so that the optical packet header recognition system can effectively recognize optical packet headers with non-fixed bits.

[0043] Furthermore, the preset topology is one of cascade, parallel, nested and three-coupler types, for example, Figure 4As shown, the preset topology is cascaded, and the microring resonator is a cascaded double-ring resonant cavity structure, wherein the cascaded double-ring resonant cavity structure includes two optical waveguides and a cascaded single ring resonant cavity located between the two optical waveguides; Figure 5 As shown, the preset topology is parallel, and the microring resonator is a parallel double-ring resonant cavity structure; Figure 6 As shown, the preset topology is nested, and the microring resonator is a nested double-ring resonant cavity structure; Figure 7 As shown, the preset topology is a three-coupler type, and the microring resonator is a three-coupler double microring resonator cavity structure.

[0044] In one implementation, the optical reservoir module includes at least one activating optical device and several couplers, wherein the activating optical device is connected to one of the several reservoir nodes, and the several couplers correspond one-to-one to the last reservoir node located in a preset manner among the several reservoir nodes, and each coupler is connected after its corresponding reservoir node. The activating optical device has a power nonlinear response and is used to provide nonlinear expression capabilities for the optical reservoir module. The coupler is used to separate a portion of the optical packet header signal transmitted by the last reservoir node, so that the separated portion of the optical packet header information can be received by the PD to perform linear regression or ridge regression. In a typical implementation, the activating optical device can be a semiconductor optical amplifier, using the nonlinear saturation region of the SOA to provide nonlinearity for the entire reservoir network architecture, and the coupler can be a splitter. Of course, in practical applications, the activating optical device can also be other devices with power nonlinear response, such as nonlinear gain media, ring resonator (power spectrum), photonic crystal structure, etc.

[0045] In one implementation, the optical packet header recognition system further includes an optical domain masking unit, which is connected to the input layer. The optical domain masking unit is configured to perform an optical domain masking (MASK, multi-order filter) operation on the optical packet header signal input by the input unit, and then input the optical packet header signal after the optical domain masking (MASK) operation into the optical reservoir module. For example, Figure 8 The 3-bit optical packet header signal on the upper left side of the middle is obtained after the Mask (multi-order filter) operation. Figure 8 Output signal on the right side of the diagram.

[0046] In one implementation, the optical reservoir neural network structure configured in the optical packet header recognition system, which is formed by the input layer, the optical reservoir module and the output layer, is trained before use. Figure 1As shown in the figure, the signal is input into the optical reservoir module through the input layer. In the optical reservoir module, the reservoir nodes do not have the concept of layers. The reservoir nodes transmit signals through sparse hybrid connections, and finally connect the signals to the output layer to obtain the output. Therefore, by training the connection weights between the input layer and the optical reservoir module, and the connection weights between the output layer and the optical reservoir module, the optical reservoir neural network structure can be trained.

[0047] In addition, in order to further illustrate the optical packet header recognition system provided by this embodiment, a series structure is adopted between several water tank nodes for illustration. Figure 2 As shown, several reservoir nodes are connected in series in sequence to form a series ring, wherein one node in the series ring is a semiconductor optical amplifier, and the last structure reservoir node of the series ring separates part of the optical signal through a splitter; the separated part of the signal is received by the PB for linear regression or ridge regression, and the optical reservoir neural network structure is trained by linear regression or ridge regression to obtain the weight coefficient corresponding to the optical reservoir neural network structure, and then the obtained optical reservoir neural network structure is configured in the optical reservoir neural network structure of the optical packet header recognition system to obtain the optical packet header recognition system.

[0048] The above-mentioned series structure is used to recognize 3-bit optical packet headers with a transmission rate of 10 Gb / s. The training set consists of 8000 items, the noise range is ±0.45, the ridge coefficient λ of ridge regression is 8e-3, the total delay τ per round is kept constant (500 ps), the time the signal spends in the reservoir is kept constant (3000 ps), the delay that each node can provide is 25 ps, the number of nodes in the reservoir network is 20, the maximum communication bandwidth of the nodes is 50 GHz, the number of test sets is 5000, and the final recognition error rate (WER) of the test set is 7.5e-4.

[0049] Table 1 Parameter list of the water reservoir structure for 3-bit it optical packet header signal recognition

[0050]

[0051] The above-mentioned series structure is used to recognize 6-bit optical packet headers with a transmission rate of 10Gb / s. The training set consists of 8000 items, the noise range is ±0.45, the ridge coefficient λ of ridge regression is 8e-3, the total delay τ per round is kept constant (500ps), the time the signal spends in the reservoir is kept constant (3000ps), the delay that each node can provide is 25ps, the number of nodes in the reservoir network is 20, the maximum communication bandwidth of the nodes is 50GHz, the number of test sets is 5000, and the final recognition error rate (WER) of the test set is 3.5e-3.

[0052] Table 2 Parameter list of the water reservoir structure for 6-bit it optical packet header signal recognition

[0053]

[0054]

[0055] In summary, this embodiment provides an optical packet header recognition system, characterized in that the system includes an input layer, an optical reservoir module, and an output layer connected in sequence. The optical reservoir module includes a plurality of reservoir nodes connected in a preset connection method, each reservoir node including a microring resonator. This application uses microring resonators as reservoir nodes, which act as a delay mechanism, resulting in low power consumption, a small footprint, and stable operation of the optical packet header recognition system. This allows for on-chip integration of the optical packet header recognition system, improving its flexibility.

[0056] Based on the above-mentioned optical packet header recognition system, this embodiment provides an integrated optical chip for optical packet header recognition, wherein the integrated optical chip integrates the above-mentioned optical packet header recognition system, wherein the connection weights between the input layer and the optical reservoir module in the optical packet header recognition system, and the connection weights between the optical reservoir module and the output layer are all trained, that is, before or after the optical packet header recognition system is integrated with the optical chip, the optical packet header recognition system will be trained to obtain the trained connection weights, so that the optical packet header signal can be directly recognized through the integrated optical chip.

[0057] Based on the above optical packet header recognition system, this embodiment provides an optical packet header recognition method, characterized in that the method applies the above optical packet header recognition system, and the method includes:

[0058] Inputting the optical packet header signal to be identified into the optical packet header identification system;

[0059] The optical packet header type corresponding to the optical packet header signal is determined by the optical packet header identification system.

[0060] Specifically, the optical signal (OOK modulation format) received by the relay station or receiving end of the optical communication system extracts the optical packet header signal through an optical switch and other equipment; after the optical packet header signal is subjected to a mask (MASK) operation in the optical domain, the optical packet header type corresponding to the optical packet header signal is identified by the optical packet header identification system. In addition, it is worth noting that the optical packet header identification method provided in this embodiment can be applied to the identification of packet header signals of optical fiber communication signals, and can also be applied to the identification of header signals with other frequencies as transmission media, such as wireless communication, terahertz communication, microwave communication, visible light communication, etc. In other words, the optical packet header identification system provided in the above embodiment can also be applied to the identification of header signals with other frequencies as transmission media, for example, wireless communication, terahertz communication, microwave communication, visible light communication, etc.

[0061] In addition, the working process and principle of the above-mentioned integrated optical chip and optical packet header recognition method have been explained in the optical packet header recognition system and will not be described here one by one.

[0062] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. An optical packet header recognition system, characterized in that: The system includes an input layer, an optical reservoir module, and an output layer connected in sequence. The optical reservoir module includes a plurality of reservoir nodes connected in a preset connection manner. Each reservoir node includes a microring resonator. The microring resonator has a bandwidth not exceeding 50 GHz and also causes a delay in the optical packet header signal. The microring resonator is a single ring resonant cavity structure or a multi-ring resonant cavity structure, wherein the multi-ring resonant cavity structure is formed by at least two single microring resonant cavity structures according to a preset topology; the preset topology is one of a cascade type, a parallel type, a nested type, and a three-coupler type; The optical reservoir module includes a plurality of couplers, each of which corresponds to a last reservoir node in a preset manner among the plurality of reservoir nodes, and each coupler is connected after its corresponding reservoir node; The optical packet header recognition system further includes an optical domain masking unit, which is connected to the input layer, wherein the optical domain masking unit is used to perform an optical domain masking operation on the optical packet header signal input by the input unit, and then input the optical packet header signal after the optical domain masking operation into the optical reservoir module; Several reservoir nodes are connected in series to form a series ring, where one node in the series ring is a semiconductor optical amplifier. The last structure of the series ring, the reservoir node, splits out part of the optical signal through the optical splitter. The separated part of the signal is received by the PD and subjected to ridge regression. The optical reservoir neural network structure is trained by ridge regression to obtain the weight coefficient corresponding to the optical reservoir neural network structure. Then, the obtained optical reservoir neural network structure is configured in the optical reservoir neural network structure of the optical Baotou recognition system to obtain the optical Baotou recognition system. The serial ring structure is used to identify the optical packet header, and the preset ridge coefficient of ridge regression is used to ensure that the total delay of each circle is fixed, and the time of the signal in the reservoir is fixed, so that each node can provide the target delay; The optical reservoir module includes at least one active optical device with a power nonlinear response, and the active optical device is connected to one of the reservoir nodes; the active optical device is a semiconductor optical amplifier.

2. An integrated optical chip for optical packet header recognition, characterized in that: The integrated optical chip integrates the optical packet header recognition system as claimed in claim 1.

3. A method for identifying an optical packet header, characterized in that: The method uses the optical packet header recognition system according to claim 1, and the method includes: Inputting the optical packet header signal to be identified into the optical packet header identification system; The optical packet header type corresponding to the optical packet header signal is determined by the optical packet header identification system.

Images