CN113114354B - Method for simultaneously positioning optical switch structure switch and optical link fault in optical data center - Google Patents
Method for simultaneously positioning optical switch structure switch and optical link fault in optical data center Download PDFInfo
- Publication number
- CN113114354B CN113114354B CN202110414399.8A CN202110414399A CN113114354B CN 113114354 B CN113114354 B CN 113114354B CN 202110414399 A CN202110414399 A CN 202110414399A CN 113114354 B CN113114354 B CN 113114354B
- Authority
- CN
- China
- Prior art keywords
- optical
- switch
- configuration
- data center
- scheduling
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
- 230000003287 optical effect Effects 0.000 title claims abstract description 124
- 238000000034 method Methods 0.000 title claims abstract description 21
- 238000001514 detection method Methods 0.000 claims abstract description 54
- 239000011159 matrix material Substances 0.000 claims abstract description 25
- 230000001133 acceleration Effects 0.000 claims abstract description 12
- 239000004744 fabric Substances 0.000 claims description 10
- 230000005540 biological transmission Effects 0.000 claims description 5
- 239000000835 fiber Substances 0.000 claims description 3
- 230000000903 blocking effect Effects 0.000 claims description 2
- 238000013461 design Methods 0.000 abstract description 6
- 238000013507 mapping Methods 0.000 abstract description 3
- 238000012544 monitoring process Methods 0.000 abstract description 3
- 238000010586 diagram Methods 0.000 description 4
- 238000011160 research Methods 0.000 description 4
- 239000013598 vector Substances 0.000 description 4
- 239000012792 core layer Substances 0.000 description 2
- 238000000354 decomposition reaction Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 239000013307 optical fiber Substances 0.000 description 2
- 210000003516 pericardium Anatomy 0.000 description 2
- IAZDPXIOMUYVGZ-UHFFFAOYSA-N Dimethylsulphoxide Chemical compound CS(C)=O IAZDPXIOMUYVGZ-UHFFFAOYSA-N 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000002457 bidirectional effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000011109 contamination Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000004134 energy conservation Methods 0.000 description 1
- 239000010410 layer Substances 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000008054 signal transmission Effects 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B10/00—Transmission systems employing electromagnetic waves other than radio-waves, e.g. infrared, visible or ultraviolet light, or employing corpuscular radiation, e.g. quantum communication
- H04B10/07—Arrangements for monitoring or testing transmission systems; Arrangements for fault measurement of transmission systems
- H04B10/075—Arrangements for monitoring or testing transmission systems; Arrangements for fault measurement of transmission systems using an in-service signal
- H04B10/079—Arrangements for monitoring or testing transmission systems; Arrangements for fault measurement of transmission systems using an in-service signal using measurements of the data signal
- H04B10/0791—Fault location on the transmission path
Landscapes
- Physics & Mathematics (AREA)
- Electromagnetism (AREA)
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention discloses a method for simultaneously positioning the faults of an optical switch structure switch and an optical link in an optical data center, which comprises the following steps: the method comprises the steps of designing an optical data center structure by utilizing an electric cache and optical exchange principle, designing a packet scheduling model with QoS guarantee by adopting a space acceleration ratio, designing a packet scheduling algorithm based on multi-hop scheduling in a single configuration matrix, mapping optical switch configurations formed by decomposing an optical link and a flow matrix into edges of virtual network topology, forming an alarm code based on a topology design detection track, and placing an alarm. The invention has the following advantages: the idea of combining a pericardial scheduling algorithm and a fault positioning method in optical data based on a space acceleration ratio is adopted, an optical link and an optical switching structure switch required by current configuration are mapped to be the edge of virtual network topology, and an optimal detection track is designed based on the topology, so that the scheme of forming switch configuration to perform fault positioning based on a packet scheduling algorithm can greatly reduce the cost and time required by fault positioning while realizing accurate positioning of faults, only alarms with the least number are placed on the optical link to perform accurate and rapid fault positioning on the optical switching structure switch and the optical link simultaneously, and optimal configuration of network monitoring resources is realized.
Description
Technical Field
The invention relates to a method for simultaneously positioning faults of an optical switch structure switch and an optical link in an optical data center.
Background
With the rapid development of services such as internet, cloud computing, big data and virtualization, the traffic and bandwidth of a data center increase exponentially. Current electrical switching and connection technologies do not perform packet transmission well in large-scale networks. In order to realize high performance, large capacity and green energy conservation of a data center, an optical data center is a development direction of a future network. In optical data centers, optical fibers are used to replace conventional cables and optical switches are used to replace conventional electrical switches. Due to the maturity of Wavelength Division Multiplexing (WDM), the transmission capacity of a single optical fiber is already as high as Tbps, which causes a great amount of loss of service data due to any form of failure in the network (such as fiber link breakage). According to the university of minnesota research in the united states, an interruption of communication for only one hour results in losses of $ 2 million, $ 250 million, and $ 600 for insurance companies, airlines, and investment banks, respectively. If this condition persists for more than 2 days, the resulting loss is sufficient to close the bank. The emerging automatic driving network requires that the network is never interrupted in the future, but hardware always fails, so that the network is required to have sensing capability and be capable of detecting the occurrence of faults in real time, and the research on fault positioning of the optical data center is of great significance.
At present, a great deal of results are obtained in the study on fault location of optical links and electrical data centers in a wide area network, but the study on fault location of the optical data centers is little. Most of the fault location research on optical data centers is in the range of optical link breakage, and the fault location on optical switch fabric switches inside an optical switch is ignored. And the optical switching structure of the core layer of the optical data center continuously executes a great deal of packet scheduling work, and the probability of failure occurrence and the severity of the resulting effect are much higher than those of the optical link. In addition, since the optical switch fabric switch is packaged inside the optical switch, a failure caused by contamination or damage of an optical port thereof is difficult to be found. Therefore, this poses a serious challenge to the fault location of the optical switch fabric switch of the high-speed and large-capacity optical switch located in the core layer, and the fault of the hardware outside the optical switch (such as the fault of the optical switch port, etc.) can be located by observing the alarm.
In recent years, methods for optical link fault location based on hardware detection have been studied in large numbers. In fault location of single links of optical networks, researchers have proposed using a design method of detection rings (m-cycles) to reduce the number of alarms required. The design of the detection ring is realized in a physical layer, and the detection ring is divided into a simple detection ring and a complex detection ring. The design of the simple detection ring is simple, the simple detection ring is applied to a simple loop which is only used for detecting each node once, but the number of required alarms of the method is large and is close to O (| E |). The complex detection ring converts the problem of optical link fault location into the problem of binary coding of the optical link under certain constraint conditions, and the number of alarms is greatly reduced to O (log 2| E |). Both of these approaches are limited by the ring detection configuration. The concept of detection trajectory (m-trail) was subsequently proposed by Wu Bin, which jumps out of the limits of the ring detection architecture, allowing the detection channel to be freely routed in the network. This allows for greater flexibility because of the minimal restriction that is not imposed by the annular configuration. By considering effective compromise among alarm cost, detection light wave cost and system fault management overhead, the detection track can quickly locate the fault optical link on the basis of minimizing the detection cost. However, this method is only suitable for fault location of the optical link, and cannot locate a fault of the optical switch fabric switch of the optical data center. Therefore, the invention provides a fault positioning method based on a packet scheduling algorithm, so that the faults of the optical switch structure switch and the optical link can be positioned at the same time, and the defects of the prior art are overcome.
Compared with the traditional scheme, the method starts from a model and an algorithm of packet scheduling, generates the configuration of the switch based on the decomposition of a service flow matrix, forms a detection track coverage strategy based on the configuration to generate an alarm code and places a corresponding alarm so as to simultaneously perform fault location on an optical switching structure switch and an optical link of an optical data center. Therefore, the fault link can be quickly and accurately positioned on the basis of minimizing the monitoring resources.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provides a method for realizing the simultaneous fault location of an optical switch structure switch and an optical link in an optical data center. The method adopts the idea of combining a pericardium scheduling algorithm and fault location in optical data of space acceleration ratio. The method comprises the steps of firstly mapping optical switch configurations formed by decomposing an optical link and a traffic matrix to the sides of a virtual network topology, then forming an alarm code based on a topology design detection track and placing an alarm so as to realize fault location of an optical switching structure switch and the optical link in an optical data center when the number of the alarms is minimum. The purpose of the invention is realized by the following technical scheme:
a method for simultaneously positioning optical switch structure switch and optical link fault in optical data center includes following steps:
s1: an optical data Center structure based on electric cache and optical switching is designed by combining a packet scheduling algorithm, and comprises three modules, namely an Input Module (IM), an Output Module (OM) and a middle-level Module (CM), wherein VOQs are used for Input ports of each IM, and each port is provided with k VOQs, so that the total number of the VOQs in each IM is nk, namely nk VOQs are adopted on each IM to buffer data packets of T time slots, and n Output Queues (OQs) are placed on each OM;
s2: and designing an optical data center multilevel packet scheduling model based on the S1 optical data center structure. The packet scheduling model can ensure that the space acceleration ratio provided by the parallel work of the middle-stage optical switches replaces the hardware acceleration ratio in a single switch while realizing the QoS guarantee of each stage:
s21: in the convergence stage, equation (1) constraint is proposed to meet a traffic matrix, so that the maximum sum of rows or columns of a traffic matrix B formed after data packets are accumulated in T time slots does not exceed T;
s22: in the scheduling stage, the proposed configuration algorithm needs to calculate N in H time slotssA permutation matrix
Wherein
Is a 0-1 permutation matrix for describing the configuration state of the middle-stage exchanger;
s23: in the switching stage, in order to ensure that all data packets are transmitted, equation (2) is proposed to implement QoS guarantee (i.e. to implement no packet loss in advance of maximum packet delay), where ΦχFinger configuration matrix PχThe traffic matrix B must be N corresponding to the weight coefficientssThe weighted configuration matrix covers. In a multi-stage switching structure of m intermediate stage optical switches, the data packet transmission is completed
Secondary reconfiguration, the spatial acceleration ratio provided by the parallel operation of the m intermediate-stage optical switches can be used to replace the hardware acceleration ratio in the switches, and equation (3) is proposed to realize QoS guarantee, wherein δ NsIs a scheduling period NsReconfiguration overhead of individual configurations, 1/WNsφ is the total duration of packet delivery in the scheduling period, W represents the number of wavelengths in each ray, and m represents the spatial acceleration ratio. While the total time overhead of scheduling and switching is 1/m (deltaN)s+1/WNsPhi) can not exceed the maximum time T accumulated by the data packet at the input module, otherwise the scheduling of the middle-stage exchanger can generate a blocking problem;
s3: a packet scheduling algorithm based on multi-hop scheduling in a single configuration matrix is designed, so that the transmission efficiency of data packets is improved, and meanwhile, the fault positioning period is reduced;
s4: in order to simultaneously realize the fault location of an optical switching structure switch and an optical link of an optical data center, firstly, the configuration of an exchanger which changes in real time is mapped into an abstract virtual network topology one by one, namely, the optical link and the optical switching structure switch required by the current configuration are mapped into the edge of the virtual network topology, and the fault is located based on a detection track coverage strategy;
s5: in order to avoid the problem of repeated placement of the alarm devices caused by configuration switching, each group of configuration uses a fixed number of alarm devices, the emitter and the terminal alarm device of each alarm device are fixed on the optical link, the sides of the virtual network topology are correspondingly generated based on the optical link and each group of exchanger configuration, and the detection track is designed by considering the detection cost constraint to realize accurate and quick fault positioning.
The invention has the beneficial effects that:
(1) At present, the fault location research of the optical data center only stays in the range of optical link breakage, and the invention not only considers the fault of the optical link of the optical data center, but also considers the fault of the optical switch structure switch in the optical switch.
(2) In view of the large scale and the difficulty in detecting the fault of the optical switch fabric switch in the optical data center, the invention provides the fault positioning method based on the packet scheduling algorithm, which only detects the optical switch fabric switch corresponding to the decomposed configuration of the current traffic matrix, so as to detect the optical switch fabric switch in real time and reduce the detection scale at the same time.
(3) The invention adopts the idea of combining a pericardium scheduling algorithm and a fault positioning method in optical data based on a space acceleration ratio, maps an optical link and an optical switching structure switch required by current configuration as the edge of a virtual network topology, and designs an optimal detection track based on the topology, so that the optical switching structure switch and the optical link are accurately and quickly positioned at the same time by only placing alarm devices with the least quantity on the optical link.
Drawings
FIG. 1 is a flow chart of a method of the present invention;
FIG. 2 is a diagram illustrating multi-hop scheduling in a single configuration matrix;
fig. 3 is a diagram of an optical data center structure and a mapped virtual network topology in a certain configuration state.
FIG. 4 is a schematic diagram of voltage constraints using a monitoring trajectory.
FIG. 5 is a schematic diagram of locating a failed edge using an application detection trajectory.
Detailed Description
The technical solutions of the present invention are further described in detail below with reference to the accompanying drawings, but the scope of the present invention is not limited to the following.
As shown in fig. 2, to achieve an efficient packet scheduling algorithm while reducing the fault location period. The data packet firstly reaches OM (1) from IM (0) through a thick dotted line path, and then reaches OM (k-1) from IM (1) through a thick solid line path, and the two-hop scheduling passes through the same middle-stage module, namely CM (0). However, the signal loop from the output port of OM (1) to the input port of IM (1) is a key factor for implementing the multi-hop scheduling inside the optical switch. Current optical technologies can process headers in the electrical domain through O/E conversion, while data payload can remain in the optical domain. Thus the electronic header can be used for packet scheduling and the data payload can implement a loop in the optical domain under system control functions. Specifically, on the basis of an ADAPT matrix decomposition scheduling algorithm, the empty time slots are utilized by adopting the multi-hop scheduling mechanism so as to reduce the configuration matrix and improve the utilization rate of the bandwidth.
As shown in fig. 3, in order to implement the virtual network topology based on packet scheduling configuration dynamic mapping, in the optical data center model proposed in fig. 3 (a), the network topology represented by G (V, E) is shown in fig. 3 (b). Where V is the set of nodes in the network and E is the set of edges in the network (a directed edge may represent the direction of traffic in the network). There are two subsets of nodes V in G (V, E) with input/output modules as sets IO of network nodes. Since the input and output modules are identical in the network structure, they can be regarded as one node. And the input/output port of each intermediate stage switch serves as a network node, the set of which is Z. Thus, there is a set relationship of V = IO ≦ Z. In G (V, E), the input module switches are fibre-connected to the intermediate module switches, and the flow of fibre data is bidirectional, so that the flow into and out of the intermediate stage switches is represented by two reciprocal edges, the set of such edges being represented by IOZ. When the intermediate stage switch configuration is formed, the input/output ports form a fixed connection and the flow direction of the network is unidirectional, the set of such edges being DE. Thus, there is a set relationship E = IOZ @ DE. Packets flowing from the i input port to the i output port (i.e., flowing from the switch port itself to itself) do not appear in real network traffic and therefore are not present in the configuration matrix.
As shown in fig. 4-5, to implement the fault location detection trajectory based on network topology optimization. Firstly, all edges in the mapped virtual topology need to be found according to voltage constraint, specifically for each edge t in the virtual topologyjDefining a start transmitter T and a finish alarm R and being TjEach of the above directional vectors u → v defines a positive voltage
Representing the orientation vector u → v at the edge tjUpper and are
Thus, in addition to the terminal R, the opposite side tjEach point passing through the network needs to satisfy the voltage sum of the out-degree vector of the point and the voltage sum of the in-degree vector of the point, that is, as shown in fig. 4, the voltage constraint, the detection track can pass through one node one or more times, and all the detection tracks found in the network are guaranteed to be a single edge according to the voltage constraint. And then, each group of configuration is based on the fixed number of the alarms, and a transmitter T and a terminal alarm R of each alarm are fixed. The number of the alarms directly influences the number of the detection edges and the detection cost. One detection wavelength is used at each edge where the detection track passes, and thus the bandwidth of the detection edge is a part of the system cost. For Crossbar optical switches, only one packet can be transmitted in each time slot, each detection signal transmission occupies one time slot, and the delay cost is expressed by the number of the sum of configuration edges (intermediate stage optical switch fabric switches) passed by the detection signal. Therefore, the detection cost mainly comprises three aspects of the number of the alarms, the bandwidth cost of the detection signal and the delay cost, and the cost is ensured to meet the minimum detection requirement by compromising the relationship among the three aspects, as shown in a formula (4). Where γ and ω are weighting factors between the siren overhead, the detection bandwidth overhead and the delay overhead. And further constraining all detection tracks found in the virtual topology according to the compromise relation of the detection cost. If the track t is detectedjAnd recording as 1 when passing through each edge, and recording as 0 when not passing through each edge. Thus by marking whether a detection trace passes an edge, a binary code can be obtained (the number of bits of the binary code is equal to the number of all detection traces in the virtual topology).
Detection cost = number of alarms + γ × total number of detection wavelengths + ω × length of configuration side (4)
Then the obtained binary code is codedAnd converting the decimal alarm code into a corresponding decimal alarm code, and ensuring the uniqueness of the alarm code. In order to ensure the accuracy of fault edge positioning, each edge is corresponding to a unique alarm code. In order to solve the problem of complicated binary code bit-by-bit comparison, the binary code is converted into a corresponding decimal alarm code which is easy to compare. For two different edges (u, v) and (x, y), their corresponding decimal numbers are alphauvAnd alphaxyAnd using binary variables
Represents alphauv＞αxy，
Represents alphauv＜αxyAnd a small positive number β is introduced, and E represents the set of all links, the uniqueness constraint of the alarm code can be expressed by equation (5):
and finally, when a certain edge breaks down, accurate and quick fault positioning can be realized according to the alarm signal with the unique alarm code received by the alarm. Fig. 5 (a) shows a detection trace which cancels the restriction of the loop structure. FIG. 5 (b) shows a solution for detecting traces for network topology, t0、t1、 t2Are all detection traces in the network topology. As shown in fig. 5 (c), for each edge in the detection track, the binary code formed by marking whether the detection track passes through the edge is converted into a uniquely corresponding decimal alarm code, so that the fault edge (i.e. the optical switch fabric switch and the optical link) can be accurately and quickly positioned with the minimum detection cost.
Claims (1)
1. A method for simultaneously positioning faults of an optical switch structure switch and an optical link in an optical data center is characterized in that: it comprises the following steps:
s1: an optical data Center structure based on electric cache and optical switching is designed by combining a packet scheduling algorithm, and comprises three modules, namely an Input Module (IM), an Output Module (OM) and a Center Module (CM); VOQs are used for the input port of each IM and each port has k VOQs, so the total number of VOQs in each IM is nk, i.e., nk VOQs can be used on each IM to buffer packets of T slots, and n Output queues (OQ: output Queue) are placed on each OM;
s2: designing an optical data center multi-level packet scheduling model based on an optical data center structure of S1, wherein the packet scheduling model realizes QoS guarantee of each stage and simultaneously enables a space acceleration ratio provided by parallel work of intermediate-level optical switches to replace a hardware acceleration ratio in a single switch:
s21: in a convergence stage, the constraint of equation (1) is proposed to meet a traffic matrix, so that the maximum sum of rows or columns of a traffic matrix B formed after data packets are accumulated in T time slots does not exceed T;
s22: in the scheduling stage, the proposed configuration algorithm needs to calculate N in H time slotssA permutation matrix
Wherein
Is a 0-1 permutation matrix for describing the configuration state of the middle-stage exchanger;
s23: in the exchange stage, in order to ensure that the data packet is transmitted completely, the formula (2) is proposed to realize QoS guarantee, wherein phiχFinger configuration matrix PχThe traffic matrix B must be N corresponding to the weight coefficientssWeighted configuration matrix overlay is performed in a multi-stage switching fabric of m intermediate stage optical switches to complete packet transmission
Minor reconfiguration, requiring the use of the space provided by the parallel operation of m intermediate-stage optical switchesInstead of hardware acceleration ratios in the switch, the inter-acceleration ratio ρ is proposed by equation (3) to achieve QoS guarantee, where δ is the reconfiguration overhead of a single configuration, δ NsIs a scheduling period NsThe reconfiguration overhead of the individual configurations is,
is the total duration of packet delivery in the scheduling period, W represents the number of wavelengths in each fiber, and the total time overhead of scheduling and switching
The maximum time T accumulated by the data packet in the input module cannot be exceeded, otherwise, the scheduling of the intermediate-stage exchanger has the problem of blocking;
s3: a packet scheduling algorithm based on multi-hop scheduling in a single configuration matrix is designed, so that the transmission efficiency of data packets is improved, and meanwhile, the fault positioning period is reduced;
s4: in order to simultaneously realize the fault location of an optical switching structure switch and an optical link of an optical data center, firstly, the configuration of an exchanger which changes in real time is mapped into an abstract virtual network topology one by one, namely, the optical link and the optical switching structure switch required by the current configuration are mapped into the edge of the virtual network topology, and the fault is located based on a detection track coverage strategy;
s5: in order to avoid the problem of repeated placement of the alarms caused by configuration switching, each group of configuration uses a fixed number of alarms, the transmitter and the terminal alarm of each alarm are fixed on an optical link, the sides of virtual network topology are correspondingly generated based on the optical link and each group of exchanger configuration, and the detection track is designed by considering detection cost constraint so as to realize accurate and quick fault location.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110414399.8A CN113114354B (en) | 2021-04-16 | 2021-04-16 | Method for simultaneously positioning optical switch structure switch and optical link fault in optical data center |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110414399.8A CN113114354B (en) | 2021-04-16 | 2021-04-16 | Method for simultaneously positioning optical switch structure switch and optical link fault in optical data center |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113114354A CN113114354A (en) | 2021-07-13 |
| CN113114354B true CN113114354B (en) | 2022-11-01 |
Family
ID=76718201
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110414399.8A Expired - Fee Related CN113114354B (en) | 2021-04-16 | 2021-04-16 | Method for simultaneously positioning optical switch structure switch and optical link fault in optical data center |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113114354B (en) |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103297356B (en) * | 2013-06-20 | 2016-01-20 | 电子科技大学 | The method of a kind of optimization data center switching network cost |
| CN104753751B (en) * | 2013-12-27 | 2019-10-29 | 南京中兴新软件有限责任公司 | A kind of method and system being dynamically determined virtual network |
| US10003528B2 (en) * | 2016-10-14 | 2018-06-19 | Oracle International Corporation | Scalable framework for managing switch link failovers in a switch fabric network |
| CN110740054B (en) * | 2019-07-17 | 2022-04-01 | 东南大学 | Data center virtualization network fault diagnosis method based on reinforcement learning |
| CN112564964B (en) * | 2020-12-04 | 2022-06-24 | 中国石油大学(华东) | Fault link detection and recovery method based on software defined network |
-
2021
- 2021-04-16 CN CN202110414399.8A patent/CN113114354B/en not_active Expired - Fee Related
Also Published As
| Publication number | Publication date |
|---|---|
| CN113114354A (en) | 2021-07-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Xi et al. | A petabit bufferless optical switch for data center networks | |
| Harai et al. | Performance analysis of wavelength assignment policies in all-optical networks with limited-range wavelength conversion | |
| US9600440B2 (en) | Network topology of hierarchical ring with recursive shortcuts | |
| Kitsuwan et al. | Independent transient plane design for protection in OpenFlow-based networks | |
| Xia et al. | Petabit optical switch for data center networks | |
| Wang et al. | Dynamic wavelength assignment for multicast in all-optical WDM networks to maximize the network capacity | |
| CN111193971A (en) | Machine learning-oriented distributed computing interconnection network system and communication method | |
| CN113114354B (en) | Method for simultaneously positioning optical switch structure switch and optical link fault in optical data center | |
| CN114116596A (en) | Dynamic relay-based infinite routing method and architecture for neural network on chip | |
| EP1187402B1 (en) | Switching apparatus, communication apparatus, and communication system | |
| US20050044264A1 (en) | Device and method for transmitting a plurality of signals by means of multi-stage protocol processing | |
| Liu et al. | WRH-ONoC: A wavelength-reused hierarchical architecture for optical network on chips | |
| Yan et al. | Performance assessment of a novel HPC network architecture based on fast optical switches with HPC application traffics | |
| Yan et al. | On the performance of scalable fast optical switch based recursive network architecture for HPC applications | |
| Zhang et al. | The construction of high-performance optoelectronic interconnection network and the realization of cloud computing for traffic data recognition | |
| Kaur et al. | Effect of crosstalk on permutation in optical multistage interconnection networks | |
| Cheng et al. | Design and performance evaluation of hexagonal topology networks with novel routing algorithms | |
| Xingyun et al. | A fault tolerant bufferless optical interconnection network | |
| CN103001869A (en) | Method and system of cloud route data processing based on port deflection | |
| Aguas et al. | Failure modeling in a WDM optical burst switched network | |
| Soni et al. | The Data Vortex Switch Architectures—A Review | |
| CN110417681B (en) | Flow control method for high-performance computer photoelectric hybrid network | |
| Zhang et al. | Optical Interconnection Network Based on Distributed Switching for Data Center Application | |
| Wang et al. | Bidirectional demarcated vector-based bidirectional link failure localization method in optical network | |
| Comellas et al. | Optical Interconnection for Datacenters: To Switch or Not to Switch |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20221101 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |