JP3519374B2 - Optical packet compression circuit, optical packet decompression circuit, and ultra-high-speed optical packet transfer ring network - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an optical packet compression circuit and an optical packet expansion circuit for temporally compressing or expanding an optical packet. In particular, the present invention relates to an optical packet compression circuit and an optical packet decompression circuit suitable for ultra-high-speed optical packet communication with a link capacity of several hundred gigabits per second or more.

[0002]

2. Description of the Related Art Today, data is converted into packets called IP datagrams (hereinafter referred to as IP packets) in communication between devices such as computers (hereinafter referred to as communication terminals) that use packets for communication. And communicate. Further, in order to perform communication between arbitrary two of a large number of communication terminals connected to the communication network, a plurality of packet transfer devices called routers are installed in the network.

At the router, the I of the input IP packet
The packet output route is selected by recognizing the IP address indicating the logical number of the communication terminal that transmits and receives the data written in the portion called the P header. In a large-scale network such as the Internet in recent years, a packet transmitted from a communication terminal cannot be transferred to a destination communication terminal unless it is relayed by many routers.

Therefore, in the usual method in which the transfer processing in the router is performed on a software basis, the entire packet is first stored in the memory and then processed, and the delay due to the software processing time cannot be avoided. Therefore, there is a problem that the delays in all the networks passing through many nodes are accumulated and the total delay time becomes long. Further, the following problems are expected in a network using a router in which IP processing is accelerated by a dedicated IC.

In the near future, a large-scale network such as the Internet is required to have a network throughput of several Tbps (Tera bits per second) to several hundred Tbps. In order to realize this, a large capacity link and a high throughput router that determines the route between the links are required. In recent years, the link capacity between routers has been increased to several hundred Gbp by using wavelength multiplexing or optical time division multiplexing technology.
s (Giga bits per second, gigabit per second) and larger capacity has become possible. On the other hand, a router corresponding to this requires a transfer throughput of several hundred Gbps × the number of input / output routes. Even with a router in which IP processing is accelerated by a dedicated IC, the above throughput cannot be obtained with the conventional configuration.

As a solution to this problem, a method has been proposed in which a large number of ICs are arranged in parallel so that a high throughput can be obtained by performing parallel processing although the processing speed of each IC is slow. Interconnection between ICs is a problem.

Further, there has been proposed an optical path network in which different optical wavelengths are assigned to paths connecting a certain node and a certain node, and packets of the same destination node are transmitted as optical signals of the same wavelength. In this case, WDM (Wavelength Division Multi) is used between adjacent nodes.
plex) transmission, and at each node, a wavelength-dispersive optical element such as an AWG (Arrayed Waveguide Grating, Arrayed Wave Guide) is used to decompose only predetermined wavelengths into packets and the router processes them in packet units. Since other wavelengths are cut through to the adjacent nodes in the optical state, the load on the router section of each node is significantly reduced. At present, the ring topology (WDM / optical ADM ring network) is mainly studied.

However, in the above-mentioned method using the optical path network, since the optical wavelength is assigned to the path and the quasi-stationary path setting is performed for each wavelength, the traffic of the actual packet in the path is smaller than the path capacity. In that case, the throughput of the entire network does not increase. Further, in order to increase the throughput of the entire network, it is possible to dynamically change the path capacity by assigning a large capacity optical path to a path with a large traffic and a small capacity optical path to a small path. For this purpose, an optical wavelength selection device that can constantly monitor the traffic of all paths and change the wavelength allocation, its operation network, and its software are required, which complicates network operation.

Therefore, the group of the inventors of the present application can already operate with low delay, low delay jitter, large capacity and high scalability (flexibility) when performing communication using packets, and With the object of providing an ultra-high-speed optical packet transfer ring network that can be easily operated, an optical add / drop type demultiplexing node device used in this network, and an operating method of this optical add / drop type demultiplexing node device. Have invented the following ultra-high-speed optical packet transfer ring network and applied for a patent. That is, the above-mentioned invention refers to an optical add / drop branch in an ultrahigh-speed optical packet transfer ring network in which optical add / drop type demultiplexing node devices that drop, add, and pass optical packets are connected in a ring shape by an optical fiber transmission line. Type demultiplexing node device, if the optical packet arriving through the optical fiber transmission path is not a packet addressed to its own node device, let it pass through the optical node device as an optical packet signal,
If the optical packet is a packet addressed to the own node device, a packet transfer control unit that branches and takes in the optical packet is provided.

[0010]

An object of the present invention is to compress or decompress optical packets with high time accuracy in order to construct an ultrahigh-speed optical packet network of, for example, several hundred Gbps as described above. It is to provide an optical packet compression circuit and an optical packet decompression circuit that can do so.

[0011]

In order to solve the above problems, the present invention provides an optical pulse generator (3001, 3002) for outputting an optical pulse at intervals of time ΔT, and an output from the optical pulse generator. N optical pulses (N is a natural number)
The optical branching device (3003) for branching to the signal line of No. 1 and the data input in series are temporarily stored, and N
A buffering circuit (3010) that outputs in parallel, and the optical pulse that is provided on the N signal lines and that is output from the optical branching device is driven by each of the N data that is output from the buffering. And N optical modulators (3004) for modulating and optical delay lines (3008) which are respectively provided at the latter stage of the modulators on the signal line and delay the output of the modulators by an integral multiple of time Δt. And an optical multiplexer (300 that multiplexes the modulated optical pulses shifted by Δt by the optical delay line and outputs the multiplexed optical pulses onto an output line.
5) The optical packet compression circuit is characterized by comprising:

The optical packet compression circuit according to the present invention further comprises a plurality of input signal lines, and a plurality of buffering circuits (3010) respectively corresponding to the input signal lines,
A read control circuit that outputs a read signal for sequentially reading data from the plurality of buffering circuits (3
019) and between the plurality of buffering circuits,
And N selection circuits (3017) for selecting data for driving the corresponding N modulators.

Further, in the optical packet compression circuit according to the present invention, the buffering circuit (3010) converts the serially input data into N-parallel serial / parallel data.
It has a parallel conversion circuit (3011) and a memory (3012) which temporarily stores this data and outputs it based on a signal from the read control circuit, and the selection circuit includes an OR circuit (3017). Is characterized in that.

In the optical packet compression circuit according to the present invention, the number of the buffering circuits is N, and N
It is characterized in that input data of a series of books is compressed into one and outputted.

Further, according to the present invention, an optical serial / parallel conversion circuit (3101) for outputting continuously inputted optical pulse signals in N parallel (N is a natural number), and the optical serial / parallel conversion circuit.
N photodetectors (3103) that perform optical / electrical conversion on each optical pulse signal branched by the parallel conversion circuit
And at least one start bit detection circuit (3104) for detecting a start bit from the electric signals output from the N photodetectors, and N for distributing the electric signals output from the N photodetectors. Switches (310
5), and N for temporarily storing the data of each of the electrical signals distributed to the N by the N switches.
^{When a} start bit is detected by two memories (3106) and one of the start bit detection circuits, the memory corresponding to the start bit detection circuit is set as the first memory and the N ² memories are sequentially arranged in a predetermined order. And a read circuit (3107) for issuing a read signal to the memory, and N multiplexed data that is multiplexed on each of the N data of the N ² memories based on the read signal and output on the output signal line. Circuit (310
The gist is an optical packet decompression circuit characterized by including 8).

In the optical packet decompression circuit according to the present invention, the optical serial / parallel conversion circuit (310
1) is an OTDM / W that converts pulses at different time positions of continuously input optical pulse signals into different wavelengths.
It is characterized by comprising a DM conversion circuit (3111) and a wavelength branching device (3112) for branching the wavelength division multiplexed signal output from the OTDM / WDM conversion circuit into N wavelengths.

Further, according to the present invention, an OTDM / WDM conversion circuit (3121) for converting pulses at different time positions of continuously input optical pulse signals into different wavelengths, and output from the OTDM / WDM conversion circuit. The optical packet decompression circuit is characterized in that the optical packet decompression circuit is provided with a dispersion medium (3122) which receives different signals and causes the signals to have different delays depending on the wavelength.

In the optical packet decompression circuit according to the present invention, the dispersion medium includes a chirped fiber grating (3132) that reflects an incident signal with a delay time different depending on the wavelength, and the OTDM / WD.
And an optical circuit (3131) for directing the optical signal output from the M conversion circuit to the chirped fiber grating and directing the optical signal reflected by the chirped fiber grating to the output side. .

An optical packet decompression circuit according to the present invention is an optical packet decompression circuit configured by connecting a plurality of stages of the above optical packet decompression circuits in series, and the optical packet decompression circuits of each stage are input. The optical pulse train is decomposed into a plurality of partial pulse trains for delayed output.The optical pulse expansion circuit is connected in multiple stages in series to divide the input optical pulse signal into partial trains of smaller units. It is characterized by going.

The present invention also provides an optical add / drop type demultiplexing node device (1-
1, 1-2, 1-3, 1-4) are connected by an optical fiber transmission line (2) in a ring shape, and an optical packet arriving through the optical fiber transmission line is a packet addressed to the own node device. If there is no optical packet signal, the optical packet signal is allowed to pass through the optical node device. If the optical packet is a packet addressed to the own node device, the packet transfer control unit that branches and takes in the optical add / drop multiplexer / demultiplex node device. In the ultra-high-speed optical packet transfer ring network, the optical add / drop type demultiplexing node device includes the optical packet compression circuit according to the present invention, and the optical packet compressed by the optical packet compression circuit is transferred to the optical fiber. It is characterized by being inserted into a transmission line.

The present invention also provides an optical add / drop type demultiplexing node device (1-
1, 1-2, 1-3, 1-4) are connected in a ring shape by an optical fiber transmission line (2), and an optical packet arriving through the optical fiber transmission line is a packet addressed to its own node device. If there is no optical packet signal, it passes through the optical node device as it is, and if the optical packet is a packet addressed to the own node device, the packet transfer control unit that branches and takes in the optical add / drop multiplexer / demultiplex node device. In the ultra-high-speed optical packet transfer ring network, the optical add / drop type demultiplexing node device includes the optical packet decompression circuit according to the present invention, and the optical packet branched from the optical fiber transmission path is It is characterized in that it is expanded by an expansion circuit.

The reference numerals included in the description of the claims do not limit the interpretation of each claim.

[0023]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described in detail below with reference to the drawings. FIG. 1 is a diagram showing a configuration according to an embodiment of an ultrahigh-speed optical packet transfer ring network of the present invention. The ultrahigh-speed optical packet transfer network includes optical add / drop type demultiplexing node devices 1-1,
Each of the optical add / drop type demultiplexing node devices 1- 1, 2-3, 1-4 is connected in a ring shape by an optical fiber transmission line 2.
Reference numerals 1 and 1-2 connect communication terminals 3-1 and 3-2.

Next, the outline of the operation of this network will be described. For example, the communication terminal 3-1 that uses this network for the purpose of transferring packets is an optical add / drop type demultiplexer (optical ADM, optical ad) that forms a ring network.
d / drop multiplex) node device (hereinafter sometimes simply referred to as an optical node device) 1-1 is connected. The optical add / drop multiplexer / demultiplexer node apparatus 1-1 reads the header information of the packet passed from the communication terminal 3-1 to connect the destination communication terminal 3-2 to the optical add / drop multiplexer / demultiplex node. The address or route information of the device 1-2 is calculated.

In the optical node device 1-1, a label including the information of the destination node device 1-2 of this packet in a digital PCM (pulse code modulation) signal format is created. Both the packet and the label are converted into an optical signal by electric / optical conversion, and sent out to the same optical fiber transmission line 2 by polarization multiplexing or optical wavelength multiplexing.

In all of the optical add / drop type demultiplexing node devices 1-1 to 1-4 through which the optical packet and the label signal pass, the label signal is obtained by polarization separation or wavelength separation and converted into an electric signal, and an electronic circuit Determines whether the corresponding packet is addressed to its own node device, and if it is addressed to its own node device, it drives and drops the optical switch to which the optical packet is input. Drive and let the optical signal pass through. Even when passing through, the corresponding optical label is reproduced, multiplexed with the optical packet to be passed, and transmitted. That is, the route control is performed by the electric circuit, and the transfer control of the optical signal is performed by the optical switch, so that the transfer control of the ultra-high speed optical packet signal can be performed.

FIG. 2 is a block diagram showing an embodiment of an optical add / drop type demultiplexing node device of the present invention. The optical add / drop type demultiplexing node device of this example corresponds to the optical add / drop type demultiplexing node device (1-1 to 1-4) shown in FIG. 1, and includes a packet termination transmitting circuit unit 11 and a packet control transmitting / receiving unit. 12, packet reception termination circuit section 13 and optical circuit section 14
have.

FIG. 3 is a block diagram showing a detailed configuration example of the optical add / drop type demultiplexing node device shown in FIG. The packet termination transmission circuit unit 11 includes a plurality of termination circuit units 111,
Packet edit transmission circuit unit 112, E / O conversion unit 1132
And a plurality of transmission packages 113 including a packet compression circuit 1131 and a packet multiplexing unit 114. The packet control transmission / reception unit 12 includes a label O / E conversion unit 122,
E / O conversion unit (2) 123, packet control circuit unit 121
Is equipped with. The packet reception termination circuit unit 13 includes a packet separation unit 131, a packet expansion circuit 1321 and a packet O.
A plurality of reception packages 13 including the / E conversion unit 1322
2. Packet edit receiving circuit section 133, E / O conversion section 13
It is equipped with 4. The optical circuit unit 14 is a 2 × 2 optical switch 14
1, an optical label separating unit 142, and an optical label multiplexing unit 143.

Next, the operation of the network will be described in detail centering on the configuration of the optical add / drop type demultiplexing node device. The packet termination circuit unit 111 of the packet termination transmission circuit unit 11
It is assumed that packets from communication terminals connected to this network will be connected using various physical networks such as Ethernet (registered trademark) and SDH. Therefore, it is a circuit unit for terminating those formats and delivering the same format to the packet edit / delivery circuit 112 of the next stage, and there are the same number of communication terminals connected to the optical node device.

The packet edit transmission circuit 112 temporarily electrically buffers the data part of the input packet, its source / destination address, and service information (priority, allowable delay, etc.). In the packet edit transmission circuit unit 112, even if packets from other termination circuit units 111 have the same destination communication terminal and the same service level, these packets may be collectively edited as one new packet. it can. Further, the packet edit transmission circuit unit 112 has a table of which optical node device the destination communication terminal of the input packet is connected to. Address information of the node device is created as a label of each packet, and the correspondence between the packet and the label is managed. Then, this label is output to the packet control transmission / reception unit 12. In response to the optical packet transmission command signal from the packet control transmission / reception unit 12, the packet edit / transmission circuit unit 112 outputs the packet to the E / O conversion unit 1132 of the transmission package 113. The packet transmission command signal is transmitted only when there is no input packet from the optical fiber transmission line side or when the packet is addressed to the own node (see FIG. 6; described later).

The packet signal output from the packet edit transmission circuit section 112 is converted into an optical signal by the E / O conversion section 1132. The wavelength of the optical packet is this ring network 1
On the other hand, it is better to be close to only one predetermined wavelength (packet wavelength). This is because when the optical packet wavelengths sent by each optical add / drop type demultiplexing node device are different,
This is because chromatic dispersion in the optical fiber transmission line causes walk-off between packets (fluctuation of relative delay time between packets), and adjacent optical packets temporally overlap with each other, making packet separation impossible. In addition, the packet control transceiver 1
From 2, the label signal corresponding to the packet is output to the E / O converter (2) 123, and is converted into the optical label signal having the label wavelength by the E / O converter (2) 123.

The optical packet signal is inserted into the ring transmission path side optical fiber by the 2 × 2 optical switch 141 of the optical circuit section 14, and the label signal corresponding to this optical packet is transferred to the ring transmission path side optical fiber by the optical label multiplexing section 143. Polarization multiplexing or wavelength multiplexing is performed with a predetermined time difference from the corresponding optical packet. The transmission time difference between the optical packet and the optical label is transmitted such that the arrival time difference between the optical packet and the optical label transmitted to the next node device at the next node device is a predetermined value.

FIG. 4 shows a time chart for explaining the operating method of each of the above functional parts. As shown in this figure, the arrival time difference between the optical packet and the optical label (the optical label arrives first) is the processing time of the label discrimination processing 101b in the packet control transmission / reception unit 12 and the optical packet transmission in the packet end transmission circuit unit 11. It is determined to be longer than the time of the processing 101c. In particular, when the optical packet and the optical label are wavelength-multiplexed, the wavelengths of the two are different, so that the transmission time difference is determined in consideration of the group velocity dispersion of the optical fiber transmission line 2. Each processing step of the time chart of FIG. 4 will be described later.

In the all-optical add / drop type demultiplexing optical node devices 1-1 to 1-4 forming the network, the optical circuit section 14 is used.
The optical label demultiplexing unit 142 converts the optical label signal wavelength-separated or polarization-separated into an electric signal by the label O / E conversion unit 122 of the packet control transmission / reception unit 12, and the packet control circuit unit 121 converts the corresponding optical packet. It is determined whether or not the packet is for the own node device and 2 × 2 depending on the determination content.
A drive signal is sent to the optical switch 141, and an optical packet send command signal is sent to the packet end sending circuit unit 11.

It should be 2 × depending on the presence / absence of an input optical packet, the destination and the presence / absence of an insertion packet from the optical node device.
The state of the two-optical switch 141 is shown in FIGS.
Shown in.

In these figures, the 2 × 2 optical switch 141
2 shows the coupling state of the input / output port in the two states. If the optical packet input from the ring transmission line side is a packet for the own node device, the 2 × 2 optical switch 1
The drive signal is sent to 41 to transit to the cross state as shown in FIG. Lead. If there is a packet to be inserted at this time, the optical packet from the ring transmission line side is dropped and branched, and at the same time, the insertion packet is inserted into the ring transmission line side.

If the input optical packet is not destined for the own node device, a drive signal is sent to the 2 × 2 optical switch 141 to shift to the bar state as shown in FIG. The packet is directly passed to the next node device. Similarly to the above, the label signal terminated at this time is re-optically labeled, the time interval is adjusted so as to arrive at the next node device at a predetermined time interval with the corresponding optical packet, and the optical packet and the optical label multiplexing unit 143. In polarization multiplexing or wavelength multiplexing, the signal is transmitted. The table of FIG. 6 shows a list of state transitions of the 2 × 2 optical switch 141 described above.

Here, the operation timings of the packet end transmitting circuit unit 11, the packet control transmitting / receiving unit 12 and the optical circuit unit 14 will be described with reference to the timing chart of FIG. When the optical label signal arrives at the optical circuit unit 14 (step 101a), the packet control transmission / reception unit 12 performs a discrimination process of the destination of the optical label signal (step 10a).
1b), after the determination is completed (step 102b), if there is no input packet from the ring side or the corresponding packet is addressed to the own node device, an optical packet transmission command is issued to the packet termination transmission circuit unit 11 and the waiting time After that (step 103b), the drive signal of the 2 × 2 optical switch 141 is output to the optical circuit unit 14. On the other hand, if there is an input packet from the ring side or if the corresponding packet is not addressed to the own node device, the packet control transmission / reception unit 12 re-creates the arrival label and creates a new optical label (step 104b, 105b).

As a result, the packet end transmission circuit section 11
Performs optical packet transmission processing (step 101c),
The optical packet is transmitted (step 102c). In the optical circuit unit 14, when the optical packet on the ring transmission line side arrives at step 102a (step 102a), the 2 × 2 optical switch 141 is switched as described with reference to FIGS. 5A and 5B (step 103a). Then step 1
At 04a, the optical label reproduced by the packet end transmitting circuit unit 11 is transmitted, and the corresponding optical packet (transfer optical packet) whose time is adjusted by the propagation of the optical packet delay line is emitted to the ring transmission line (step 105a).

The packet edit reception circuit section 133 of the packet reception terminating circuit section 13 receives the packet signal from the packet O / E conversion section 1322 of the reception package 132, before being edited in the original packet edit transmission circuit section. Re-edit the packet format. When re-editing, the address information of the destination communication terminal such as the IP address of the original packet is read, and the re-edited packet is output to the corresponding output port based on this information.

The above-mentioned optical add / drop type demultiplexing node device 1
In a network using, when the traffic of the node device upstream of a certain optical node device is large and the packet occupation probability is close to 1, the insertion probability to the ring transmission line decreases. When the insertion is impossible, the memory of the packet end transmission circuit unit 11 is accumulated until an insertable slot arrives.
However, in this case, since the average value of the arrival intervals of empty slots is small and stochastic, transfer delay and transfer delay fluctuation increase, and in the worst case, the memory size is exceeded and packet loss occurs.

Therefore, in the present example, at least one of the optical add / drop type demultiplexing node devices included in the ring transmission line has the optical node only at the address of a specific optical add / drop type demultiplexing node device in the ring transmission line. Control is performed so that a label that allows packet insertion from the device is generated at a predetermined frequency. The slots of the packets corresponding to those labels are empty. Since each optical node device other than the designated optical add / drop type demultiplexing node device in the ring reads this label and determines that it cannot be inserted by the above-described routine, the slot of the packet corresponding to this label remains empty. Reach the optical node device.

That is, since packet slots that can be inserted arrive at the designated optical add / drop type demultiplexing optical node device, a packet transmission rate whose frequency is the minimum bandwidth is guaranteed. . The packet time length is set to 80 ns (nano second,
Nanosecond), the arrival cycle of the dedicated packet to the optical node device is 125 kHz (kilo hertz, kilohertz)
Becomes Assuming that the number of bits included in one packet is 1500 bytes (12000 bits), 1.5 Gbps (Gi
It is possible to guarantee a band of ga bits per second (gigabits per second) to the optical node device.

In this example, the network throughput can be increased by using the packet compression circuit 1131 and the packet expansion circuit 1321. Packet compression /
The conceptual diagram of extension is shown. Even with the same packet size (the number of bits included in a packet), increasing the bit rate reduces the packet duration area, enabling more packets to be transferred on the ring transmission line, thus improving network throughput. Will increase. By performing packet compression / expansion in the optical signal area, the packet edit transmission circuit unit 112 and the packet edit reception circuit unit 133
There is no need to change the configuration or performance of.

Packet compression / expansion circuits 1131 and 132
When a packet is inserted in the optical add / drop type demultiplexing node device using No. 1, a flow control method is used. Assuming that the packet compression ratio is N and the number of packet O / E conversion units 1322 is M, the decompression time of N / M packet time length is required in principle when the packet decompression circuit 1321 decompresses the packet. If the next optical packet destined for the optical node device arrives during this expansion processing, the packet reception termination circuit unit 13 cannot receive the packet due to interference between the packets. To avoid this,
When trying to insert a packet from each optical node device, confirm that the packet of the same destination is not within the N / M packet (including the planned insertion packet position) before and after the planned insertion packet time position, and insert only if not Specify the protocol to be used. A timing chart of an optical label and an optical packet for realizing this is shown in FIG.

The "granting delay due to other factors" shown in FIG. 8 is the time difference between the label and the packet required even when this flow control protocol is not used. In all the optical add / drop type demultiplexing node devices, in the packet control circuit unit 121, the label transmission timing is changed from the packet transmission timing to the packet compression / expansion circuit 11.
Compared to the case where there is no 31/3121, it is sent earlier by (N / M-1) packet slots. The packet insertion optical node device also monitors the packet slot N / M or more before the insertion target position in time.

If a packet destined to the same node device as the destination node device of the packet to be inserted by the optical node device exists in the N / M packets before and after the insertion target position, the insertion is stopped, If there is no packet destined to the same node device, insertion is executed. By this flow control protocol, it is possible to eliminate the existence of two or more packets destined to the same optical node device in a continuous N / M packet sequence, regardless of which packet sequence in the ring transmission path is extracted. Become.

In this example, the optical add / drop type demultiplexing node device establishes bit synchronization between labels and sends out the labels. In a packet transfer network in which bit strings between packets are asynchronous, it is very difficult to extract a clock having a frequency accuracy of 10 ⁻⁷ or more from a received packet. In the case of an application that distributes video data over a long time such as a movie in real time, the video encoding clock at the transmission terminal and the video reproduction clock at the reception terminal must match with high accuracy.

When the recovered clock at the receiving terminal is higher than the encoded clock at the transmitting terminal, data loss occurs at the receiving terminal, and when the recovered clock at the receiving terminal is lower than the encoded clock at the transmitting terminal, reception is performed. The buffer memory in the terminal overflows and the data at the rear of the video disappears. For example, a clock frequency of 600 MHz (Mega hertz) 2
High-definition video data of time has a clock frequency difference of 10 ^-5
In the case of transmission between the transmitting and receiving ends, the shift of 4.3 megabits occurs at the end of the video. By establishing bit synchronization between labels, all optical node devices in the network can share a common clock with a high accuracy of 10 ^-9 or more, and the above real-time application can also be transferred. . The specific method will be described below.

At least one optical node device of the optical add / drop type demultiplexing node devices included in the ring transmission line is a master node device for supplying a label clock. Each optical node device separates the optical packet and the polarization- or wavelength-multiplexed optical label and electrically terminates them. That is, each optical node device extracts the clock of the label and reproduces the bit of the label to discriminate the content. After that, an optical label is created by reproducing the bit based on the extracted clock and performing an electro-optical conversion, and the optical label is multiplexed with the optical packet by polarization multiplexing or wavelength multiplexing and is output to the next node device. That is, each optical node device can share the clock of the master node device by extracting the clock of the reception label. At this time, label recognition may be facilitated by performing frame synchronization using the label as a frame, and it is also possible to adjust the label issuing rate by inserting an appropriate empty bit between the labels. Of course, even in this case, synchronization at the bit level is established.

FIG. 9 is a table showing a design example of the ultrahigh-speed optical packet transfer ring network of the present invention. The ring length is 500 km. Ring transmission line fiber 1.55
The optical frequency deviation (wavelength deviation) of the optical packet light source of each optical add / drop type demultiplexing node device is set to 20 GHz (Giga hertz, gigahertz), the walk-off time after a packet transmitted from each optical add / drop type demultiplexing node device goes around the ring can be kept to 0.1 ns or less.

Since the bit synchronization between packets is asynchronous, the receiver of each optical node device needs to establish bit synchronization for each packet. When the bit phase is matched at the head bit of the packet, the bit phase is matched at the end of the packet.
z) needs to be Δf ≦ γ / Tb. Tb is the packet time length before compression, and γ is the allowable error of the bit phase at the end of the packet. 40G clock for uncompressed packet
Hz, packet bit length L = 12000 bits, γ =
When it is set to 2%, the required frequency accuracy is 66.7 kHz, and an independent clock having the accuracy shown on the left may be installed in each optical node device. At this time, only the bit phase is adjusted within the preamble time at the beginning of the packet. By setting the preamble time to 4 ns, it becomes possible to match the bit phase for each packet by using a phase comparison circuit such as a microwave mixer.

Here, an example is shown in which an optical packet of 40 Gbps is compressed to 1/4 and inserted into the ring transmission line side as an ultra high speed optical packet of 160 Gbps. The configuration of the optical add / drop type demultiplexing node device at this time is the same as that of FIG.
The bit length of the optical packet input to the packet compression circuit is fixed at 1500 bytes. A configuration example of the packet compression circuit is shown in FIG.

If the length of the pulse compression / expansion loop changes with time, the bit intervals after compression / expansion become non-uniform, so it is necessary to stabilize the loop effective length. Optical fiber thermal expansion coefficient for loop is β (/ ° C), temperature change is Δ
T (° C), the allowable non-uniformity ratio of the bit interval is γ (0 <γ <
In the case of 1), the relationship of βNLΔT <γ needs to be satisfied. Here, N is a packet compression ratio.

In this design example, assuming that γ = 2%, β˜
In the case of a fiber reduced by 10 ⁻⁷ , it may be stabilized within 5 ° C. near room temperature. The fiber loop length is about 15 m. In this example, the bit order is exchanged between the input bit string and the output bit string, but if the packet expanding circuit of the destination node device uses the expanding circuit of the same loop length, the same bit order as before the compression is reproduced after the packet expansion.

The optical switch used may be a Mach-Zehnder interferometer type optical switch in which an optical waveguide is formed in a lithium niobate (LiNbO ₃ ) crystal and a planar electrode is arranged. Since the modulation bandwidth of 10 GHz is sufficiently possible at the technical level of a commercially available optical switch, the switching transition time (the switching time between the cross state and the bar state) can be 0.1 ns or less.

The guard time between optical packets is determined in consideration of the transition time of the optical switch, the walk-off time between optical packets, and the jitter of the label sending circuit. Here, the total guard time is designed to be 1 ns including the margin of 0.7 ns. Together with the above-mentioned preamble time, the overhead time used for other than payload packet transfer within the transmission band is 5 ns, and its occupancy rate is about 6%.

Label information is transferred between adjacent nodes in a bit synchronous mode. One master node device for supplying the clock of the label is provided in the ring. The packet phase of the packet returned to the master node device after revolving the ring network varies due to linear expansion / contraction of the ring recirculation time due to environmental temperature fluctuations of the ring fiber transmission line. The packet transmission phase in 1 does not always match the phase. Therefore, it is necessary to adjust the packet transmission frequency (that is, the same as the label transmission frequency) to the fluctuation of the ring circulation time so that the packet transmission cycle is always synchronized with an integral fraction of the ring circulation time.

Since the duration of the 160 Gbps optical packet is 80 ns, if 80 bits are used as the label signal, the label bit rate is 1 Gbps and the label transmission frequency is 12.5 MHz. Therefore,
It becomes possible to perform label processing with the current electronic circuit technology such as Ethernet technology.

In the case of guaranteeing the band, the above-mentioned clock-supplied master node device may be a master node device for sending a dedicated label as shown in FIG. For example,
In order to guarantee a band of 2 Gbps for a certain optical node device k, a label with a destination of k may be issued at a rate of 156.25 kHz (that is, a rate that is 1/80 of the label sending rate). As a result, the other nodes pass this packet to the optical node device k, so that the optical node device k can be guaranteed to have the packet inserted at the minimum rate.

A calculation example of the optical add / drop type demultiplexing node device which can be accommodated in the case of the design shown in the above embodiment is shown in FIG.
2 shows. In this figure, the burst rate is the peak traffic with respect to the average traffic of the packets input to the network. η is the peak bit rate between nodes with respect to the ring side transmission capacity. 160 Gbps on the ring side,
If the peak bit rate between optical node devices is 1 Gbps and the burst rate of input packets is 5, then 36 optical add / drop type demultiplexing node devices can be accommodated between each optical node device.

In general, the network transfer delay time is the sum of packetized / non-packetized delay in the transmitting / receiving optical node device, queue delay and switch delay in each passing optical node device, and propagation delay of the transmission path. In this example, rather than recognizing the address information by converting an optical signal into an electric signal in the transmission node device for the transfer of an ultra-high-speed optical packet, by branching or passing it in the state of the optical signal, The transfer processing capacity is not limited by the bottleneck of the electric circuit. The packet processing by the electric circuit is performed only by the transmitting node device and the terminating node device, and the passing optical node device is cut through as an optical signal. Also, the transmitting node device suffers a queue delay only when a packet is inserted into the ring, but the delay amount is small.

FIG. 13 shows a queue delay amount, and FIG. 14 shows an example of calculation of queue delay variation. Even if the packet utilization rate is about 0.6, the delay is within 10 packets. In this example, since the ultra-high speed packet is used, the packet delay is extremely small.
In the embodiment, since one packet is 80 ns, the queue delay for 10 packets is 0.8 μs, which is extremely short in real time, and the total transfer delay is almost the propagation delay of the transmission path. Further, the delay fluctuation is only the queue delay fluctuation in the transmitting node device and is sufficiently small. In the case of an application that cannot tolerate delay fluctuation of several packets, the bandwidth guarantee algorithm described above is activated to enable packet transfer with the minimum fluctuation guaranteed without delay fluctuation.

Next, the protection method in the above label switch network will be described with reference to FIGS.
This will be described with reference to FIG.

Conventionally, a LAN (Local Area Net) constructed by a packet-based label switch network
work), a maintenance operation management packet is exchanged by a routing protocol to monitor a failure and notify operation information for maintenance. Further, the information terminal that receives the packet detects an error in the packet and requests retransmission of the packet if there is an error, but the quality of the transmission path between the two points is not monitored.

In a WAN (Wide Area Network) composed of a label switch network, the distance between nodes becomes large, so a path is set between two points,
SDH (Synchronous Digital Hierarchy) and SONE
It is transmitted after being multiplexed in a T (Synchronous Optical Network) frame. The quality of the overhead of the transmission frame is monitored by a monitoring byte, and when an abnormality is detected, the switching is controlled by an automatic switching byte.

In addition, The Patent Cooperation Treaty
International patent application "Redundant" published by (PCT)
Path Data Communication ", International Publicati
on Number WO 00/13376, International Publication D
The following technologies are disclosed in ate 9 March 2000. That is, when the same packet is transferred through different paths in the label switch network, if the packet is lost midway, the packet on the protection path is received. And counting the packet loss,
If the threshold is exceeded, an error will be notified or recorded. However, with such a technique, a failure cannot be detected unless a packet arrives, and a packet loss can be detected only at the receiving node, so it is not possible to know where the failure occurred. In addition, there is a problem that high-speed switching cannot be performed because the transmission quality is not constantly monitored. Further, since the above-mentioned international publication does not describe the algorithm for dealing with the failure, the configuration of the node device, and the operation thereof, technical details are unknown.

The packet-based label switch network has high link band utilization efficiency for burst traffic such as data communication between computers, but LA
In the conventional method of N, when the distance between the nodes becomes large, it is necessary to frequently exchange the maintenance operation management packet frequently in order to improve the reliability and speed of the protection, which reduces the actual data throughput. In WAN, not packet-based transfer,
The SDH (including SONET) system constantly monitors the transmission quality and is capable of high-speed protection, but sets a certain band path between two points to secure a certain band regardless of the increase or decrease in traffic. Therefore, the bandwidth utilization efficiency of the link is reduced for burst traffic.

In consideration of the above circumstances, the present invention realizes a packet-based packet transfer network with high throughput for burst traffic, and enables high-speed switching by constantly monitoring the transmission quality. Realizes protection with a level of reliability equal to or higher than SDH.

In the optical packet network,
It is difficult to realize protection with high reliability and high-speed switching when packets are transferred as optical signals, but the present invention solves this problem and is a quality monitoring method with high reliability and high-speed switching. I will provide a.

In the following, an embodiment of protection will be described with reference to FIGS. 15 to 33.

In a label switch network that transfers a packet based on the address information of the label, a communication path code of several bits for detecting or correcting an error in the transmission system is added to the frame of the label information, and the node monitors the label of the packet. To monitor the transmission quality. This monitoring is performed, for example, by using a parity check in which a 1-bit parity bit is added to an n-bit label signal and the number of "1" s in (n + 1) bits is always an even number or an odd number. . The parity bit enables detection of an error of 1 bit in (n + 1) bits. As a result, the bit error rate can be calculated,
The transmission quality between two points can be monitored.

<2-fiber 1: 1 unidirectional ring network> FIG. 15 is a flowchart for detecting an abnormality in a 2-fiber 1: 1 unidirectional ring network. 16 (a), 16 (b), and 16 (c) are schematic diagrams showing the label switch paths during normal times, fiber breaks, and ring breaks, respectively. Here, "fiber disconnection" means that the active system or the standby system is in a disconnected state. Further, “ring disconnection” means that the link between the nodes becomes disconnected at any place on the ring.

Hereinafter, according to the flow chart of FIG.
Description will be given with reference to FIGS. 16 (a), 16 (b) and 16 (c). In normal times, one of the two fibers is the active system and the other is on standby as a standby system. Then, each node sends a data packet to the active system. As shown in FIG. 16A, for example, the node 6 transmits a packet to each node via the label switch path. The receiving node performs reception in the active system, and monitors the quality of the active-system label signal by the monitor. As shown in FIGS. 16B and 16C, when a transmission path failure (fiber disconnection and ring disconnection, respectively) occurs between the nodes 1 and 2, the node 2 causes an abnormality in the active label signal. To detect. When the signal quality falls below a predetermined level, an abnormality is notified to the upstream node, node 1, via the standby system or an operation network (not shown), and the node 1 diverts the current system signal to the standby system, causing a failure. To avoid. After that, when the transmission line failure is restored, the original working system can be returned to.

FIG. 17 is a block diagram showing an example of the configuration of a node that operates according to the above-described fault avoidance and restoration algorithm. The optical label extraction circuit 2001 extracts the optical label from the optical packet arriving at this node through the ring network, and the extracted optical label is transmitted by the optical label reception circuit 2003 to the optical / electrical (O / E,
optical to electrical) is converted. And monitor 2
004 monitors this label signal converted into an electric signal.

Based on the label information, the 2 × 2 optical switch 20
In 06, the optical packet addressed to the own node is branched from the ring side, and is received by the optical packet receiving circuit 2007 to be optically / electrically converted. Then, the packet edit reception circuit 2008 restores the original packet and transfers it to the destination user.

The data signal from the user side is electrically terminated in the termination circuit 2010 according to the interface. The optical packet editing / transmitting circuit unit 201
In No. 1, the data part of the packet, its source / destination address, and service information (priority, allowable delay, etc.) are temporarily buffered electrically. Here, packets having the same destination or packets having the same service level may be collectively edited as one new packet. Also, a label signal indicating information including the address of the destination node device or the route to the destination node device is created and notified to the control circuit unit 2005. The control circuit unit 2005 sends a trigger to the optical packet transmission circuit 2012 at a timing for inserting a packet on the ring side to insert the optical packet on the ring side.

Further, the monitor 2 which monitors the label signal
When an abnormality is detected by 004, the control circuit unit 200
Reference numeral 5 sets the 2 × 2 optical switch 2014 in the cross state to perform switching control and divert the signal of the working system to the protection system.
By the same operation as described above, the standby optical packet is branched from the ring side for the one addressed to the own node, otherwise it is allowed to pass as optical.

In FIG. 17, reference numeral 2009 designates electric / optical (E / O, el) for transferring the packet received by the packet edit receiving circuit section 2008 to the user side.
electrical to optical) conversion circuit.

<2-fiber 1 + 1 Unidirectional Ring Network> Next, the case of the 2-fiber 1 + 1 unidirectional ring network will be described. FIG. 18 is a flowchart showing a process when an abnormality is detected in the network of the same form. FIG. 19 shows (a) normal and (b) when the transfer directions of the active system and the standby system are opposite.
It is the schematic which showed the label switch path at the time of fiber cutting | disconnection and (c) ring cutting | disconnection. 20A and 20B are schematic diagrams showing the label switch paths in the case where the transfer directions of the active system and the standby system are the same direction, respectively, in (a) normal operation and (b) fiber disconnection.

Hereinafter, according to the flow chart of FIG.
This will be described with reference to (a), (b) and (c) of FIG. 19 and (a) and (b) of FIG. In normal times,
Each node sends a packet of data to both the active path and the standby path (label switch path). FIG. 19
As shown in (a) and FIG. 20 (a), for example, the node 6 transmits a packet to each node via the active and standby label switch paths. The receiving node receives the data of the active system, electrically terminates the label signal in both the active and standby monitors, and monitors the quality of the signal. 19 (b), (c), and FIG.
When a failure (fiber break or ring break) occurs between the nodes 1 and 2 as shown in (b), an abnormality is detected only in the active label signal. When the signal quality falls below a predetermined level, the node switches the receiving end to the backup system. FIGS. 19B, 19C, and 20B show the label switch paths from the node 6 to other nodes at that time. In the 2-fiber 1 + 1 type, if a ring break occurs in a network in which the transfer direction of the active system and the transfer direction of the standby system are the same,
Failure avoidance cannot be performed by switching to the standby system.

FIG. 21 is a block diagram showing a node configuration example when the transfer direction of the active system and the transfer direction of the standby system are opposite to each other. In FIG. 21, reference numeral 2021 is a selector for selecting a packet from the active packet edit receiving circuit section 2008 or the standby packet edit receiving circuit section 2008. Further, 2022 is a bridge / selector that functions as a bridge that sends the same packet to both the working system and the protection system in normal times, and functions as a selector that sends the packet to either the working system or the protection system when the ring is disconnected. Is. Although FIG. 21 shows a case where the transfer directions of the active system and the standby system are opposite, when the transfer directions of the active system and the standby system are the same,
A similar configuration can be used so that the transfer directions of both systems are the same.

In the procedure shown in the flow chart of FIG.
A procedure for comparing and recording the packets received by both systems may be added. Such a procedure is shown in the flowchart of FIG. In the procedure shown in FIG. 22, a process of comparing and recording the received packets of the active system and the standby system branched from the ring is provided. As a result, it is possible to switch the receiving end for each packet between the active system and the standby system when a failure occurs, and it is possible to effectively utilize network resources such as optical fibers, nodes, and bands.

FIG. 23 is a block diagram showing an example of a packet switching type node configuration for realizing the procedure shown in FIG. The feature of the configuration shown in FIG. 23 is that the active packet edit receiving circuit 2008 and the standby packet edit receiving circuit 2
The point is that a packet comparison circuit 2023 for comparing the packets received by 008 and 008 is provided.

<4 Fiber 1: 1 Bidirectional Ring Network> Next, a case of a 4 fiber 1: 1 bidirectional ring network will be described. In the network of this form, the active system and the standby system each have a bidirectional optical transmission path. FIG. 24 is a flowchart showing a process when an abnormality is detected in the network of the same form. Further, FIG. 25 shows (a) normal and (b), respectively.
It is the schematic which showed the label switch path at the time of fiber cutting | disconnection and (c) ring cutting | disconnection.

Hereinafter, according to the flow chart of FIG.
This will be described with reference to FIGS. 25 (a), 25 (b), and 25 (c). During normal times, each node sends a packet in the direction of the shortest path of the active transmission path. For example, the node 6 transmits a packet to each node through the label switch path as shown in FIG. The receiving node performs reception using the active system and monitors the quality of the label signal on the monitor. As shown in FIGS. 25 (b) and 25 (c), for example, when a failure (fiber disconnection or ring disconnection) occurs between the nodes 1 and 2, the nodes 1 and 2 detect the abnormality of the active label signal. To detect. When the signal quality falls below a predetermined level, the nodes 1 and 2 bypass the working signal to the protection system.

FIG. 26 is a block diagram showing an example of a node configuration for performing the above-mentioned failure avoidance operation in a 4-fiber 1: 1 bidirectional ring network. In this node configuration, an optical label detection circuit 2001, an optical label reception circuit 2003, and a monitor 2004 are provided on each of a total of four fibers, and they are constantly monitored. Further, the optical packet addressed to the own node is a 2 × 2 optical switch 2006 or a 1 × 2 optical switch 2006-1.
At the optical packet receiving circuit 2007, is edited by the packet edit receiving circuit unit 2008, and is transferred to the user side. Further, the transmission packet is edited by the optical packet edit transmission circuit 2011, transmitted by the optical packet transmission circuit 2012, and
It is inserted in the × 2 optical switch 2006.

In normal times, the active system is used to send packets bidirectionally according to the destination, and when the monitor detects an abnormality in the label signal, the 2 × 2 optical switch 2006
By setting the cross state, the control for switching the active system to the standby system is performed.

<4 Fiber 1 + 1 Bidirectional Ring Network> Next, the case of the 4 fiber 1 + 1 bidirectional ring network will be described. FIG. 27 is a flowchart showing a process at the time of abnormality detection when the transfer directions of the active system and the standby system are the same in the network of the same form. FIG. 28 is a schematic diagram showing label switch paths when (a) normal, (b) fiber cut, and (c) ring cut, when the transfer directions of the active system and the standby system are the same, respectively. It is a figure. FIG. 48 is a flowchart showing a process at the time of abnormality detection when the transfer directions of the active system and the standby system are opposite in the network of the same form. Further, FIG. 29 shows (a) normal operation, (b) fiber disconnection, when the transfer directions of the active system and the standby system are opposite directions, respectively.
(C) A schematic view showing a label switch path when the ring is broken. FIG. 30 is a block diagram showing a configuration example of a 4-fiber 1 + 1 bidirectional node when the transfer directions of the active system and the standby system are opposite.

28 (a), (b), (c), and FIG. 29 along the flowcharts of FIGS. 27 and 48.
Description will be made with reference to (a), (b), (c), and FIG. In normal times, each node sends a packet of data to the shortest route path (label switch path) of both the active system and the standby system. For example, the node 6 is shown in FIG.
Then, the packet is transmitted to each node through the label switch path as shown in FIG. The node constantly monitors the label signals of all packets with a monitor.

Here, FIG. 28 (b) and FIG. 29 (b).
Consider a case where a transmission line failure (fiber disconnection) occurs between the nodes 1 and 2 as shown in FIG. The nodes 1 and 2 detect the abnormality of the active label signal from the side of the fiber cut portion. At this time, the standby system is not affected by the fiber break, and the label signal of the standby system is normal. When the quality of the signal of the working system falls below a predetermined level, the receiving end is switched from the working system to the protection system.

Next, consider a case where a transmission line failure (ring disconnection) occurs between the nodes 1 and 2. As shown in FIG. 28 (c), when a ring disconnection occurs between the nodes 1 and 2 when the transfer directions of the active system and the standby system are the same, the nodes 1 and 2 adjacent to the faulty site will have the active system and the standby system. Detects anomalies in both label signals in the system and recognizes that a ring break has occurred. Then, first, the ring-shaped packet in the direction toward the failure point in the active system and the standby system is branched and detoured to the active system and the standby system in the opposite directions, respectively. That is, the packet edit reception circuit 2008 shown in FIG. 30 transfers the packet to the optical packet edit transmission circuit 2011, and sends out the received packet so as to detour in the reverse direction. At that time, congestion may occur due to an increase in the number of packets to be processed. Therefore, the packets from the user side are distributed to the active system and the standby system to distribute the traffic. Then, the fact that a ring break has occurred at that location is notified using the active system, the standby system, or an operation network (not shown). All the nodes receiving this notification set the label switch path so as to avoid the faulty part and send the packet.

As shown in FIG. 29C, when the transfer directions of the active system and the standby system are opposite to each other, ring disconnection occurs between the nodes 1 and 2, and the quality of the label signal falls below a predetermined level. , The receiving end is switched from the active system to the standby system.

As shown in FIG. 30, this node normally uses both the active system and the standby system to transfer packets in the direction according to the destination. The data from the user side is terminated, and the bridge / selector 2022 sends the same packet to both the active system and the standby system during normal operation, and switches to only one of them when the ring is disconnected and sends the packet. Also, regarding reception from the ring side, the selector 2
In 021, the active packet edit receiving circuit 200
8 or the standby packet edit receiving circuit 2008 is switched.

Although FIG. 30 shows the configuration in which the transfer directions of the active system and the standby system are opposite, the same applies when the transfer directions of the active system and the standby system are the same. Can be used.

A procedure of comparing and recording packets received by both systems may be added to the procedure shown in the flowcharts of FIGS. 27 and 48, and the received packets of the active system and the standby system branched from the ring are compared. And a recording process is provided. As a result, it is possible to switch the receiving end for each packet between the active system and the standby system when a failure occurs, and it is possible to effectively utilize network resources such as optical fibers, nodes, and bands. Of such a procedure, the case where the transfer of the active system and the transfer of the standby system are in the same direction is shown in the flowchart of FIG.

FIG. 32 is a block diagram showing an example of a packet switching type node configuration for realizing the procedure shown in FIG. The feature of the configuration shown in FIG. 31 is that the active packet edit reception circuit 2008 and the standby packet edit reception circuit 2
The point is that a packet comparison circuit 2023 for comparing the packets received by 008 and 008 is provided.

Next, a method for identifying a node when an optical switch failure occurs in the node in the above-mentioned optical packet transfer ring network will be described. The ring network means that at least one of the optical add / drop multiplexer / demultiplexer node devices generates a label signal at a predetermined frequency that describes that only the optical node device of a specific address in the network can insert a packet. Then, in the optical node device that has transmitted the optical fiber transmission path and obtained the label, packet insertion is prohibited except for the optical node device designated as the source node device by the label. It is a packet transfer ring network.

FIG. 33 is a flow chart showing an algorithm for identifying a fault location of an optical switch in such a network. As shown in FIG. 33, a specific master node that supervises the ring sends pilot packets bidirectionally to each node. This pilot packet is a fixed data string. Each node receives the pilot packet, confirms whether the content is normal, and notifies the master node of the confirmation result by the active system, the standby system, or the operation network. This allows the master node to identify if the optical switch of the node in the ring has failed. When the master node identifies a node with a failed optical switch, it notifies it to replace it.

With the above technique, the signal quality of the transmission line can be constantly monitored in the packet-based label switch network, so that high-speed switching can be performed with high reliability even in a wide area communication network with a large node distance. Protection becomes possible. With such protection, the advantages of high throughput and flexibility due to the statistical multiplexing effect of the label switch network can be utilized. Further, in the optical packet transfer network, in the conventional technology, the signal quality of the transmission line could not be constantly monitored unless the optical packet was optically / electrically converted for each intermediate node, but only the optical label was electrically terminated and monitored. By doing so, constant monitoring becomes possible.

Next, the optical packet compression circuit and decompression circuit in the above label switch network will be described with reference to FIGS. 34 to 45.

The technique described below realizes an ultrahigh-speed optical packet transfer network and an interconnection in a node device which cannot be obtained by the direct electrical / optical conversion method. Such techniques are packet-by-packet (pa
cket by packet) is a means for increasing the throughput of optical packets from the electrical domain to the optical domain by increasing the throughput in order to reduce the probability of packet collision or reduce the queue delay in packet collision. It is a means of shortening the packet time length, a means of not reducing the throughput by packet compression even when the network traffic is congested, and a means of reducing the time occupying the band when transmitting a large amount of data and transmitting instantly. is there.

When an optical packet is generated by direct electrical / optical conversion of an ultrahigh-speed optical packet, there is a problem that the speed is limited by the response characteristics of the electric packet generation circuit and the optical modulator. The structure of a packet compression circuit according to the prior art for solving this problem is shown in FIGS.
6 shows.

In the optical packet compression circuit shown in FIG. 34, N optical pulse trains emitted from the laser oscillator 4002 are branched in parallel by an optical branching device 4003, and these pulse trains are relatively time-delayed by an optical fiber delay line 4005. Then, the optical coupler 4006 multiplexes them and outputs an N-multiplexed optical pulse bit string. Each external modulator 4004 is controlled by the shift register 4012 based on the input electric signal 4011 and modulates one pulse at a time. The delay amounts of the N optical fiber delay lines 4005 are 0, Δt · c / n, 2Δt · c / n, and
..... (N-1) .DELTA.t.c / n is set. Here, Δt is the light pulse interval, c is the speed of light, and n is the refractive index of the core.

The optical packet compression circuit shown in FIG.
2 optical switch 4022 and optical fiber delay lines 4023 and 2
There are k stages of a circuit including a × 1 optical multiplexer 4024 (k is a natural number). The odd-numbered pulses forming the optical packet are branched to the upper side by the optical switch 4022, the even-numbered pulses are branched to the lower side, and the upper pulse is branched to the fiber delay line 402.
3 causes a delay, and the optical multiplexer 4024 multiplexes the lower pulse with a slight time lag. After passing through such a circuit for k stages, the optical gate switch 4025 removes an extra pulse, so that a high-speed optical packet of 2 ^k bits can be generated.

The optical packet compression circuit shown in FIG. 36 includes an optical amplifier 4045 and an optical bandpass filter (OBPF) 40.
42, delay line 4046, and 2 × 2 optical switch 40
And a delay loop composed of 43 and an optical gate switch 404.
It is composed of four. In the 2 × 2 optical switch 4043, a part of the pulse of the optical packet circulates in the delay loop, so that the time interval with the next pulse becomes narrow. By repeating such a round, the time intervals of all the pulses are shortened, and the 2 × 2 optical switch 4043 is brought into the cross state to output all the power. Then, the optical gate switch 4044 outputs only the packet-compressed optical packet.

FIG. 37 is a block diagram showing an optical packet decompression circuit having a configuration similar to that of the optical packet compression circuit of FIG. This optical packet decompression circuit has k stages of circuits each including a 1 × 2 optical switch 4061, an optical fiber delay line 4062, and a 2 × 1 optical multiplexer 4063. The 1 × 2 optical switch 4061 distributes the odd-numbered pulses of the input high-speed optical packet downward and the even-numbered pulses upward. Since the pulse distributed to the upper side is delayed by the fiber delay line 4062, the pulse interval is doubled at the stage of being output from the 2 × 1 optical multiplexer 4063. After passing through such a circuit through k stages of circuits, the pulse interval is sufficiently widened and the input pulse order is maintained. The optical packet expanded in this way is
Photo / electric conversion is performed by a photo detector (not shown).

FIG. 38 is a block diagram showing an optical packet decompression circuit having a configuration similar to that of the optical packet compression circuit of FIG. This optical packet decompression circuit includes an optical amplifier 404.
5, an optical bandpass filter (OBPF) 4042, a delay line 4046, a delay loop including a 2 × 2 optical switch 4043, and an optical gate switch 4044. The pulse train goes around the delay loop, and the 2 × 2 optical switch 4043 sequentially outputs one pulse from the first pulse,
Extract while delaying with an optical delay loop. Then, the optical gate switch 4044 allows only necessary pulses to pass.

The compression circuit and decompression circuit according to the prior art described above have the following problems. Figure 3
In the conventional optical packet compression circuit shown in FIG. 4, since there is one optical packet compression circuit for one input data, its throughput is the same as that of the input electric signal. Therefore, a plurality of optical packet compression circuits are required to increase the throughput according to the amount of compression, which increases the scale and cost.

In the conventional optical packet compression circuit shown in FIG. 35, 1 × is used to compress an L-bit optical packet.
A circuit including two switches, an optical fiber delay line, and a 2 × 1 optical multiplexer is required for log2 (L) stages or more. As described above, when the circuits are connected in multiple stages, there is a problem that the loss of optical power, the circuit scale, the cost, and the like increase.
Further, since the optical fiber delay line is long, it is necessary to compensate the fiber expansion and contraction due to temperature. In addition, 1500 bytes (byte
s) That is, a packet of 12000 bits is transmitted at 10 Gbp
When compressing from s to a bit rate of 100 Gbps,
The required number of circuit stages k is 19, and c / n = 2 × 108 m / s (where c is the speed of light and n is the refractive index of the core), and the first pulse is 12000 in total in the optical fiber delay line. X (100 [ps] -10 [ps]) x (2
× 108 [m / s]) = 216 [m] is propagated. The coefficient of linear thermal expansion of the optical fiber is in the range of 10 ^-6 to 10 ^-5 , so if this is set to 10 ^-5 , 216 [m]
× 10 ⁻⁵ [/°C]=2.16 [mm / ° C]. Since the required bit interval Δt is 10 ps (pico seconds), the distance traveled in the optical fiber during that time is Δt.
t · c / n = 2 × 10 ⁻³ m (meter). Therefore, even if the jitter tolerance is set to 1/100, the required accuracy of the length of the optical fiber is 20 μm (micrometer) or less. That is, in order to maintain the accuracy of the length of the optical fiber of 20 μm or less with the jitter tolerance of 1/100, it is necessary to keep the temperature constant with the accuracy of about 0.01 ° C. or less. Also here, as the ultra-high speed is achieved, the accuracy required for the length and temperature of the optical fiber becomes high as described above, and it becomes difficult to realize the optical fiber.

In the conventional optical packet compression circuit shown in FIG. 36, since the number of rounds of the delay loop is repeated more for the first pulse, the number of times passing through the optical amplifier 4045 increases and the S / N ratio deteriorates. There is. Further, the expansion and contraction of the fiber due to the temperature also poses a problem. For example, 150
0 bytes, that is, a 12000-bit packet, 10G
When compressing from bps to 100 Gbps bit rate, the first pulse is a total of 12 optical fiber delay lines.
000 x (100 [ps] -10 [ps]) x (2 x 1
08 [m / s]) = 216 [m] will be propagated.
At this time, as described above, the optical fiber loop is set to about 0.0
It is necessary to keep the temperature constant with an accuracy of 1 ° C. or less and to control the delay line with a high accuracy of 20 μm or less in time. In addition, the number of pulses that can be packet-compressed is limited by the length of the delay loop.

The conventional optical packet decompression circuit shown in FIG. 37 has the following problems in addition to the above problems of the optical packet compression circuit shown in FIG. That is,
The 1 × 2 optical switch 4061 in the first stage needs to respond at a very high speed, which is a fraction of the pulse interval of the optical packet or less. Further, as the speed increases, bit phase synchronization becomes difficult.

The conventional optical packet decompression circuit shown in FIG. 38 has the following problems in addition to the above problems of the optical packet compression circuit shown in FIG. Ie 2
The × 2 optical switch 4043 is required to respond at a very high speed, which is a fraction of the pulse interval of the optical packet or less. Further, as the speed increases, bit phase synchronization becomes difficult.

As described above, it is difficult to compress or expand a packet of about 1500 bytes into an ultrahigh-speed optical packet of 100 Gbps or more only by using the conventional technique. It is an object of the present invention to provide a compression and decompression technique suitable for an optical packet having a bit rate of 100 Gbps or more for a packet of about 1500 bytes.

An optical packet compression circuit and an optical packet expansion circuit according to the embodiments of the present invention will be described below with reference to the drawings. FIG. 39 is a block diagram showing the configuration of the optical packet compression circuit according to the embodiment of the present invention. N series (# 1 to #N) data signals (clock frequency: 1 / ΔT) input as electric signals are input to the buffering circuit 3010. Each buffering circuit 3010 has a serial / parallel conversion circuit 3011. The input electric signal is output in N parallel by the serial / parallel conversion circuit 3011, and these N outputs are input to the memory 3012. Memory 3
012 has a capacity according to the packet length of the input electric pulse, and outputs the stored data in parallel for each N bits based on the read clock signal. The read clock signal has a time interval ΔT, and the clock signal control circuit 3
019, and the clock signal control circuit 3019
The outputs from the above are input to the respective memories 3012. The N OR circuits 3017 include N memories 30.
The logical sum of the data at the same time position output from 12 is obtained, and the output becomes a drive signal for the optical modulator 3004 described below. That is, the OR circuit 3017 has a function of selecting one of the N input electric signals.

Clock generator 3001 and laser oscillator 3
Optical pulse generated by 002 (repetition frequency:
1 / ΔT) is branched into N lines by the optical branching device 3003, and is modulated by the drive signal from the corresponding OR circuit 3017 in the branched optical modulator 3004 on each signal line. The optical multiplexer / demultiplexer 3005 is delayed in the delay line 3008 so as to be shifted by time Δt.
Is combined with.

FIG. 40 is a timing chart showing the timing of pulse trains at each port in the optical packet compression circuit shown in FIG. 40 (a) to (l)
Correspond to (a) to (l) on the circuit of FIG. 39, respectively. In the input electric pulse trains shown in FIGS. 40A, 40B, and 40C, the pulse interval is ΔT, that is, the bit rate is 1 / (ΔT) [bit / s]. Figure 40
The electric pulse train shown in (a) is input to the memory 3012 one pulse at a time as shown in FIGS. 40 (d), (e), (f) and (g) by the serial / parallel conversion circuit. Then, reading from the memory 3012 is performed based on the control signal output from the clock signal control circuit 3019. The short pulse output from the ultrashort pulse oscillator (laser oscillator 3002) is N-branched by the optical branching device 3003, and is modulated by each optical modulator 3004 by the electric signal read from the memory 3012. After the modulation, the optical pulse train of each port is time-shifted by Δt due to the optical delay line 3008, so the pulse trains in (h) to (k) of FIG.
(H) to (k). And the optical multiplexer 30
When multiplexed at 05, a plurality of optical packets are output as an optical pulse train with an interval of Δt, as shown in FIG. That is, the bit rate 1 / (ΔT) [bit /
[s] has a bit rate of 1 / (Δt)
It is compressed into an optical packet of [bit / s].

FIG. 41 is a block diagram showing the structure of an optical packet decompression circuit according to an embodiment of the present invention. Figure 41
As shown in FIG. 3, this optical packet decompression circuit includes an optical serial / parallel conversion circuit 310 for converting an optical signal to serial / parallel.
By 1, the pulse is branched into N lines for each pulse, and this is repeated in N pulse cycles. As an example of the optical serial / parallel conversion circuit, an OTDM / WDM conversion circuit 311 is used.
FIG. 42 shows a configuration that is composed of 1 and an AWG (Arrayed Wave Guide) 3112. This AWG acts as a wavelength splitter. For an embodiment of the OTDM / WDM conversion circuit, see K. Uchiyama et al., "Multiple-channel output a.
ll-optical OTDM demultiplexer using XPM-induced chi
rp compensation (MOXIC) ", Electronics Letters, Vo
See l.34 No.6 pp.575-576, 19 March 1998, etc.

The optical signal branched into N lines for each pulse is photoelectrically / electrically converted by the N photodetectors 3103, and at least one start bit detection circuit 3 is provided.
It is input to 104. Start bit detection circuit 3104
When the start bit is detected, the read circuit 3107
Fire a trigger for. Further, the electric pulse signal that has passed through the start bit detection circuit 3104 has N switches 3
At 105, the packets are allocated to the memory and temporarily stored in the electric memory 3106. Further, based on the trigger, the read circuit 3107 issues a read signal to each electric memory 3106 at intervals of ΔT, starting with the electric memory 3106 connected to the detected start bit, and thereby each electric memory 310.
The contents of 6 are sequentially read at intervals of ΔT. The memory (3106) # i-j receives the switch (3105) #i as an input and the OR circuit (3108) #j as an output (i, j = 1, 2, ..., N). The OR circuit 3108 outputs the logical sum of the electric signals output from the N memories.

43 and 44 are timing charts showing the timing of the pulse train at each port of the optical packet decompression circuit shown in FIG. 43 (a) to 44 (a) to (a) to (q) on the circuit of FIG. 41.
Each corresponds to (q). The input ultrahigh-speed optical pulse train shown in FIG. 43 (a) is distributed by the optical serial / parallel conversion circuit 3101 pulse by pulse, converted into an electric signal by the photodetector 3103, and shown in (b) to (e). It becomes a timing electric pulse, which is further distributed by the switch 3105 and is stored in the memory 31 for each packet as shown in (f) to (n).
It is temporarily stored in 06. Then, the read circuit 31
Since one pulse is sequentially read from the electric memory 3105 at time intervals of ΔT based on the signal from 07, the output from the OR circuit 3108 is shown in FIGS. 42 (o), (p), and (q).
The pulse interval as shown in (1) is output as a packet of an electric pulse train in which Δt is expanded to ΔT.

FIG. 45 is a block diagram showing the structure of a packet expansion circuit according to another embodiment. As shown in FIG. 45, this packet decompression circuit is configured by connecting the following partial circuits in m stages (m is the number of stages). That is, it is a partial circuit including an OTDM / WDM conversion circuit 3121 that converts an optical signal having a different wavelength for each predetermined number of pulses, and a dispersion medium 3122 that causes a different delay depending on the wavelength. The input optical signal is OTD
The M / WDM conversion circuit 3121 converts a predetermined number of pulses into different wavelengths, and the dispersion medium 3122 causes different delays depending on the wavelength. By repeating this for m steps, the pulse time interval of the entire packet is extended and output.

FIG. 46 shows an embodiment of the dispersion medium shown in FIG. This dispersion medium is a chirped fiber grating (CFG) that reflects an optical signal at a position different from that of the optical circuit 3131 depending on the wavelength.
3132, an optical signal converted into a different wavelength for every predetermined number of pulses and input is passed through the optical circuit 3131 and is incident on the chirped fiber grating 3132. Since the reflection position on the chirped fiber grating 3132 varies depending on the wavelength, the transmission path length varies depending on the wavelength. Therefore, at the time when the light is emitted from the chirped fiber grating 3132 and output from the optical circuit 3131 to the next stage, a time interval is left for each wavelength.

According to this method, in the OTDM / WDM conversion circuit 3121, the converted wavelength of the input optical pulse is
Even if drift occurs due to the deviation of the bit phase synchronization of the receiving circuit, it only causes the drift of the packet time position after the packet expansion, so that precise bit phase synchronization is not required.

FIG. 47 is a timing chart showing the timing of pulse trains at each port of the optical packet decompression circuit shown in FIG. 47 (a) to (e),
These correspond to (a) to (e) on the circuit of FIG. 45, respectively. As shown in FIG. 47A, the pulse train input at the time interval of Δt is the first-stage OTDM / WDM conversion circuit 3
By 121, several pulses are collected together and different wavelengths λ1, λ1 + Δλ, λ1 + 2Δλ, as shown in FIG.
λ1 + 3Δλ, ... At the stage after the signal shown in (b) enters and is emitted from the first-stage dispersion medium 3122, the timing is shifted according to the wavelength as shown in (c). That is, in this example, the wavelength (λ1 + Δλ) has a larger delay in the dispersion medium 3122 than the wavelength λ1, and in FIG. 46, the reflection position in the chirped fiber grating 3132 is far and the transmission path is long. It is output with a delay. Second OTDM /
In the WDM conversion circuit 3121, λ in (c)
Multiple pulses of one wavelength are combined into several pulses,
It is converted into different wavelengths λ2, λ2 + Δλ, .... Also,
The wavelength (λ2 + Δ
The signals of λ) and (λ2 + 2Δλ) are delayed, and the pulse timing becomes as shown in (d). Similarly, by sequentially shifting the pulse timings up to the m-th stage, one pulse is finally output at a time interval of ΔT as shown in (e). This time interval ΔT is determined by the delay amount of the wavelength interval of the dispersion medium 3122. Further, this ΔT is determined by the grid spacing of the chirped fiber grating 3132 in FIG.

The optical packet decompression circuit shown in FIG. 45 has a multi-stage structure in consideration of the relationship between the bandwidth that can be secured by OTDM / WDM conversion and the number of pulses to be expanded. With a wideband OTDM / WDM conversion circuit that can convert to all different wavelengths and a dispersion medium that covers the wideband, one-stage circuit
A one-stage configuration in which each pulse is extended may be used.

As described above in each of the embodiments, the optical packet compression circuit of the present invention has a simple control of the delay amount, a small bit fluctuation, and a small effect of temperature change, as compared with the prior art. (Currently 40 Gbps) to 100
It is possible to easily and highly accurately convert to a bit rate in the optical region of Gbps or higher. Further, since it is possible to generate continuous ultra-high-speed optical packets, it is possible to achieve N times the throughput when the compression ratio is N for one optical packet compression circuit as compared with the conventional technique.

Further, the conventional optical packet decompression circuit is
Bit phase synchronization and ultra high speed optical switch are essential,
Since current high-speed optical switches have switching speeds of several hundred ps at most, it is only possible to extend optical packets of 10 Gbps or less, and it is difficult to extend the pulse interval of ultra-high-speed optical packets of 100 Gbps or more to the electrical area. there were. However, the optical packet decompression circuit according to the present invention can easily perform 100 Gb without the need for an ultra-high speed optical switch or bit phase synchronization as described in the above embodiments.
Since it becomes possible to decompress an ultra-high-speed optical packet of ps or more and decompress continuous optical packets, the expansion ratio N can be increased for one optical packet expansion circuit as compared with the conventional technology.
Is N times the throughput.

In each of the above embodiments, the communication between the respective node devices for realizing the WAN has been described, but the same technique is applied to the interconnection for the communication between the constituent elements in a single device. It is also possible to apply. As a result, it becomes possible to perform signal transmission within the device at an extremely high speed, and even in the case of a signal line failure, the effect of the failure can be locally suppressed so as not to lead to the failure of the entire device. Although the embodiments of the present invention have been described in detail above with reference to the drawings, the specific configuration is not limited to these embodiments, and includes a design and the like within a range not departing from the gist of the present invention.

[0129]

As described above, by using the optical packet compression circuit of the present invention, the control of the delay amount is simple, the bit fluctuation is small, and the influence of the temperature change is small.
It is possible to easily and highly accurately convert the bit rate in the electric region (currently 40 Gbps) to the optical region bit rate of 100 Gbps or more.

Further, by using the optical packet decompression circuit according to the present invention, it becomes possible to easily decompress an ultra high speed optical packet of 100 Gbps or more without the need for an ultra high speed optical switch or bit phase synchronization.

Further, the optical add / drop type demultiplexing node device is provided with the above-mentioned optical packet compression circuit, and the optical packet compressed by this optical packet compression circuit is inserted into the optical fiber transmission line or the optical packet compression circuit is inserted into the device. By providing a packet decompression circuit and decompressing an optical packet branched from the optical fiber transmission line by this optical packet decompression circuit, it becomes possible to operate the ultra-high-speed optical packet transfer ring network in a stable manner.

[Brief description of drawings]

FIG. 1 is a schematic diagram showing a configuration according to an embodiment of an ultrahigh-speed optical packet transfer ring network according to the present invention.

FIG. 2 is a block diagram showing a configuration of an embodiment of an optical add / drop type demultiplexing node device according to the present invention.

FIG. 3 is a block diagram showing a detailed configuration example of the optical add / drop type demultiplexing node device shown in FIG.

FIG. 4 is a timing chart illustrating an operating method of each unit of the optical add / drop type demultiplexing node device shown in FIG.

5 is an explanatory diagram showing a state and case classification of the 2 × 2 optical switch shown in FIG.

6 is a table showing a list of state transitions of the 2 × 2 optical switch shown in FIG.

7 is a schematic diagram illustrating the concept of packet compression / decompression in the optical node device shown in FIG.

8 is a timing chart showing a flow control method at the time of packet insertion in the optical node device shown in FIG.

FIG. 9 is a table showing an example of design numerical values of an ultrahigh-speed optical packet transfer ring network of the present invention.

10 is a schematic diagram illustrating a configuration example of the packet compression circuit illustrated in FIG.

FIG. 11 is a timing chart showing a configuration example of a dedicated label transmitted when the master node device guarantees a band.

FIG. 12 is a graph showing the relationship between the burst rate and the number of nodes that can be accommodated in the design of an ultrahigh-speed optical packet transfer ring network.

FIG. 13 is a graph showing a queue delay amount indicated by a relationship between a utilization rate and an average waiting time [packet].

FIG. 14 is a graph showing a queue delay fluctuation represented by a relationship between a utilization rate and a wait time distribution [packet].

FIG. 15 is a flowchart showing a protection procedure in a two-fiber 1: 1 unidirectional network.

FIG. 16: In a two-fiber 1: 1 unidirectional network, (a) when normal, (b) when fiber is cut,
And (c) is a schematic diagram showing a label switch path from the node 6 to each of the other nodes when the ring is broken.

FIG. 17 is a block diagram showing a node configuration example in a two-fiber 1: 1 unidirectional network.

FIG. 18 is a flowchart showing a protection procedure in a two-fiber 1 + 1 unidirectional network.

FIG. 19 shows a case where the transfer directions of the active system and the standby system are opposite to each other in the two-fiber 1 + 1 unidirectional network,
FIG. 3 is a schematic diagram showing a label switch path from the node 6 to each of the other nodes when (a) normal, (b) fiber disconnected, and (c) ring disconnected.

FIG. 20 shows a node 6 in a two-fiber 1 + 1 unidirectional network in which the transfer directions of the active system and the standby system are the same, (a) when normal and (b) when the fiber is disconnected, respectively.
FIG. 6 is a schematic diagram showing label switch paths from a node to other nodes.

FIG. 21 is a block diagram showing a node configuration example in the case where the transfer directions of the active system and the standby system are opposite to each other in the two-fiber 1 + 1 unidirectional network.

FIG. 22 is a flowchart showing a procedure of protection in a packet switching two-fiber 1 + 1 unidirectional network.

FIG. 23 is a block diagram showing a node configuration example in the case where the transfer directions of the active system and the standby system are opposite to each other in the packet switching 2-fiber 1 + 1 unidirectional network.

FIG. 24 is a flowchart showing a protection procedure in a 4-fiber 1: 1 bidirectional network.

FIG. 25: In a four-fiber 1: 1 bidirectional network, (a) normal operation, (b) fiber disconnection,
And (c) is a schematic diagram showing a label switch path from the node 6 to each of the other nodes when the ring is broken.

FIG. 26 is a block diagram showing an example of a node configuration in a 4-fiber 1: 1 bidirectional network.

FIG. 27 is a flowchart showing a protection procedure in the case where the transfer directions of the active system and the standby system are the same in the 4-fiber 1 + 1 bidirectional network.

FIG. 28 shows a node 6 in the case where the transfer directions of the active system and the standby system are the same in the 4-fiber 1 + 1 bidirectional network, (a) normal, (b) fiber disconnected, and (c) ring disconnected, respectively. FIG. 6 is a schematic diagram showing label switch paths from a node to other nodes.

FIG. 29 shows a case where the transfer directions of the active system and the standby system are opposite to each other in the 4-fiber 1 + 1 bidirectional network,
FIG. 3 is a schematic diagram showing a label switch path from the node 6 to each of the other nodes when (a) normal, (b) fiber disconnected, and (c) ring disconnected.

FIG. 30 is a block diagram showing a node configuration example when the transfer directions of the active system and the standby system are opposite to each other in the 4-fiber 1 + 1 bidirectional network.

FIG. 31 is a flowchart showing a protection procedure in the case where the transfer directions of the active system and the standby system are the same in the packet switching 4-fiber 1 + 1 bidirectional network.

FIG. 32 is a block diagram showing a node configuration example in the case where the transfer directions of the active system and the standby system are the same in the packet switching 4-fiber 1 + 1 bidirectional network.

FIG. 33 is a flowchart showing a procedure for identifying a failure location of an optical switch.

FIG. 34 is a block diagram showing an example of a conventional optical packet compression circuit.

FIG. 35 is a block diagram showing another example of a conventional optical packet compression circuit.

FIG. 36 is a block diagram showing still another example of the optical packet compression circuit according to the conventional technique.

FIG. 37 is a block diagram showing an example of a conventional optical packet decompression circuit.

FIG. 38 is a block diagram showing another example of an optical packet decompression circuit according to the related art.

FIG. 39 is a block diagram showing a configuration of an optical packet compression circuit according to an embodiment of the present invention.

40 is a timing chart showing the timing of pulse signals in the optical packet compression circuit shown in FIG. 39.

FIG. 41 is a block diagram showing a configuration of an optical packet decompression circuit according to an embodiment of the present invention.

42 is a block diagram showing a detailed configuration of an optical serial / parallel conversion circuit in the optical packet decompression circuit shown in FIG. 41.

43 is a timing chart showing the timing of pulse signals in the optical packet decompression circuit shown in FIG. 41.

44 is a timing chart showing the timing of pulse signals in the optical packet decompression circuit shown in FIG. 41.

FIG. 45 is a block diagram showing a configuration of an optical packet decompression circuit according to another embodiment of the present invention.

46 is a block diagram showing a detailed configuration of a dispersion medium in the optical packet decompression circuit shown in FIG. 45.

47 is a timing chart showing the timing of pulse signals in the optical packet decompression circuit shown in FIG. 45.

FIG. 48 is a flowchart showing a protection procedure in the case where the transfer directions of the active system and the standby system are opposite to each other in the 4-fiber 1 + 1 bidirectional network.

[Explanation of symbols]

3001 clock generator 3002 Laser oscillator 3003 Optical splitter 3004 Optical modulator 3005 Optical multiplexer 3008 Optical delay line 3010 Buffering circuit 3011 Serial / parallel conversion circuit 3012 memory 3017 OR circuit 3019 Clock signal control circuit 3101 Optical serial / parallel conversion circuit 3103 Photodetector 3104 Start bit detection circuit 3105 switch 3106 Electric memory 3107 readout circuit 3108 OR circuit 3111 OTDM / WDM conversion circuit 3112 AWG 3121 OTDM / WDM conversion circuit 3122 Dispersion medium 3131 Optical circuit 3132 Chirped fiber grating

─────────────────────────────────────────────────── ─── Continuation of front page (56) References JP-A-7-38575 (JP, A) JP-A-6-261072 (JP, A) JP-A-7-321745 (JP, A) (58) Field (Int.Cl. ⁷ , DB name) H04L 12/42-12/46 H04L 12/54-12/56 H04J 3/00-3/26 H04J 14/00-14/08 H04B 10/00-10 / 28

Claims (7)

(57) [Claims] 1. An optical serial / parallel conversion circuit (3101) for outputting continuously input optical pulse signals in N parallel (N is a natural number), and each optical pulse branched by the optical serial / parallel conversion circuit. N photodetectors (3103) for converting each signal into light / electricity, and at least one start bit detection circuit (3104) for detecting start bits from the electric signals output from the N photodetectors. , N switches (3105) that distribute the electrical signals output from the N photodetectors, and N that temporarily stores the data of each of the electrical signals distributed to the N switches by the N switches. ^{When the} start bit is detected by the two memories (3106) and one of the start bit detection circuits, the start bit detection circuit is paired with the start bit detection circuit. The corresponding memory is set as the first and the N
^A read circuit (3107) for issuing read signals to the two memories, and N of the data read from the N ² memories based on the read signals, N pieces of which are multiplexed and output on output signal lines, respectively.
An optical packet decompression circuit comprising: a plurality of multiplexing circuits (3108). 2. The optical packet decompression circuit according to claim 1 , wherein the optical serial / parallel conversion circuit (3101) converts pulses at different time positions of continuously input optical pulse signals into different wavelengths. And an OTDM / WDM conversion circuit (3111) and a wavelength demultiplexer (3112) that branches the wavelength division multiplexed signal output from the OTDM / WDM conversion circuit into N wavelengths. Packet decompression circuit. 3. An OTD for converting pulses at different time positions of continuously input optical pulse signals into different wavelengths.
An M / WDM conversion circuit (3121) and a dispersion medium (3122) that receives the signal output from the OTDM / WDM conversion circuit and causes the signal to undergo different delays depending on the wavelength. Optical packet decompression circuit. 4. The optical packet decompression circuit according to claim 3 , wherein the dispersion medium reflects an incident signal with a different delay time depending on a wavelength.
2) and an optical circuit (313) which directs the optical signal output from the OTDM / WDM conversion circuit to the chirped fiber grating and directs the optical signal reflected by the chirped fiber grating to the output side.
1) An optical packet decompression circuit comprising: 5. An optical packet decompression circuit configured by connecting a plurality of optical packet decompression circuits according to claim 3 or 4 in series, wherein each stage of the optical packet decompression circuits outputs an optical pulse train. This is to divide it into multiple partial pulse trains and output it in a delayed manner. By connecting multiple stages of this optical packet decompression circuit in series, the input optical pulse signal is decomposed step by step into smaller partial trains. Optical packet decompression circuit. 6. An optical add / drop type demultiplexing node device (1-1, 1-2, 1-) that drops, inserts, and passes an optical packet.
If the optical packet arriving through the optical fiber transmission line is not a packet addressed to the own node device, the optical packet signal remains as it is. Ultra-high-speed optical packet transfer including a packet transfer control unit that branches and takes in if the optical packet is a packet addressed to the own node device in the ring network, the optical add-drop multiplex type node device according to claim 1 or
2. An ultra-high-speed optical packet transfer ring network, comprising the optical packet decompression circuit according to 2 , wherein the optical packet branched from the optical fiber transmission line is decompressed by the optical packet decompression circuit. 7. An optical add / drop type demultiplexing node device (1-1, 1-2, 1-) for dropping, inserting, and passing an optical packet.
If the optical packet arriving through the optical fiber transmission line is not a packet addressed to the own node device, the optical packet signal remains as it is. Ultra-high-speed optical packet transfer including a packet transfer control unit that branches and takes in if the optical packet is a packet addressed to the own node device In a ring network, the optical add / drop type demultiplexing node device is from claim 3.
The optical packet decompression circuit according to any one of 5 to 5 ,
An ultra-high-speed optical packet transfer ring network, wherein the optical packet expanded by the optical packet expansion circuit is inserted into the optical fiber transmission line.