WO2001091387A1

WO2001091387A1 - Statistical multiplexer

Info

Publication number: WO2001091387A1
Application number: PCT/JP2001/004346
Authority: WO
Inventors: Takeshi Ota
Original assignee: Photonixnet Kabushiki Kaisha
Priority date: 2000-05-26
Filing date: 2001-05-24
Publication date: 2001-11-29
Also published as: AU2001258824A1

Abstract

Statistical multiplexers (1a, 2a) comprise a group (5) of uncompressed transmission channels, and a group (4) of compressed transmission channels. The transmission capacity of the group (5) of uncompressed transmission channels is greater than the transmission capacity of the group (4) of the compressed channels. The statistical multiplexer comprises means for extracting packet part of signals sent from the group (5) of uncompressed transmission channels, and first means for rearranging the extracted packets dynamically in the group (4) of compressed transmission channels. Signals are compressed by reducing the proportion of idle signals in the signals sent to the compression transmission channels (4) by the extraction means and the first rearrangement means.

Description

Description Statistical multiplexer Technical field

The present invention relates to a statistical multiplexing device suitable for use in optical communication. In particular, the present invention relates to a statistical multiplexing apparatus suitable for being applied to a communication network in which a plurality of transmission paths are provided in parallel by a method such as wavelength multiplexing. Background art

Due to a sudden increase in Internet traffic, the line capacity of the Internet knock-bon has become insufficient. In the world-wide telecommunications network, optical fibers were laid based on the conventional demand forecast of voice line (telephone) traffic, and the number of laid optical fibers was insufficient. In order to solve such a problem, a wavelength multiplexing device is used. A wavelength multiplexing device modulates different signals with light of different wavelengths and transmits / receives the same, so that if one optical fiber is multiplexed, for example, if eight wavelengths are multiplexed, substantially eight light beams are transmitted. This technology can be treated as fiber.

However, for example, when wavelength multiplexing of eight wavelengths is performed, each wavelength channel is treated as a completely independent channel, and traffic such that one channel is full but the adjacent channel is empty. There was unevenness in the work. This is like a highway with no lane change. In addition, due to the speed of technological innovation, existing network equipment and facilities were not sufficiently amortized, but became outdated and existing facilities were destroyed. Disclosure of the invention

In order to solve the above-mentioned problems, a statistical multiplexing apparatus according to the present invention comprises: an uncompressed transmission channel group including a plurality of transmission channels; and a compressed transmission channel including a plurality of transmission channels. In the statistical multiplexing apparatus having a group, the sum of the transmission capacities of the uncompressed transmission channel groups is larger than the sum of the transmission capacities of the compressed channel groups, and the signal transmitted from the uncompressed transmission channel group is Extracting means for extracting a packet portion from the packet; and first rearranging means for dynamically rearranging the extracted packets to a group of compressed transmission channels, wherein the extracting means and the first rearrangement are provided. Means for compressing the signal by reducing the ratio of the idle signal in the signal transmitted to the compressed transmission channel.

The statistical multiplexing apparatus according to the present invention includes: a destination detecting unit configured to detect a destination of the packet from a dynamically rearranged signal transmitted from the compressed channel group; A second rearrangement unit for rearranging the packet, wherein the destination is expanded by the destination detection unit and the second rearrangement unit.

The statistical multiplexing apparatus according to the present invention is characterized in that the transmission rates of the uncompressed channel and the compressed channel are different.

The statistical multiplexing apparatus according to the present invention is characterized in that the speeds of transmission channels included in the uncompressed channel group are not uniform.

The statistical multiplexing device of the present invention is characterized in that the speeds of transmission channels included in the compressed channel group are not uniform.

A statistical multiplexing device according to the present invention is characterized in that it has a crossbar switch.

The statistical multiplexing device according to the present invention is further characterized in that at the time of traffic congestion, traffic is bypassed to an internal channel equipped with a large capacity memory.

According to the above configuration, the idle signal can be removed from the transmission signal of the uncompressed transmission channel group and sent to the compressed transmission channel group.

According to the above configuration, it is possible to detect a destination address for each packet from among the compressed signals sent via the compressed transmission channel group and restore (expand) the original signal.

According to the above configuration, any transmission standard can be adopted for the uncompressed transmission channel and the compressed transmission channel, and the existing network equipment and facilities can be effectively used. Can be

According to the above configuration, traffic control between the uncompressed transmission channel group and the compressed transmission channel group can be performed by cross-parsing. In addition, the cross-persistent method can realize statistical multiplexing for high-speed transmission lines.

According to the above configuration, traffic overflowing during traffic congestion can be temporarily saved in the large-capacity memory. BRIEF DESCRIPTION OF THE FIGURES

FIG. 1 is a diagram showing an optical communication device provided with the statistical multiplexing device of the present invention. FIG. 2 is a diagram illustrating the operation principle of the statistical multiplexing device.

Fig. 3 is a block diagram (compression) showing the internal configuration of the statistical multiplexing device.

FIG. 4 is a diagram showing the structure of NIC and CIC in the statistical multiplexing device. FIG. 5 is a diagram showing a state of basic control of the statistical multiplexing device.

Fig. 6 is a time chart showing the principle of statistical multiplexing (compression).

FIG. 7 is a diagram showing a temporal change of a connection pattern of the crossbar switch. Fig. 8 is a block diagram (expansion) showing the internal configuration of the statistical multiplexing device.

FIG. 9 is a time chart showing the principle of statistical multiplexing (expansion).

FIG. 10 is a diagram showing the principle of processing at the time of traffic congestion.

FIG. 11 is a diagram showing a statistical multiplexing device according to a second embodiment of the present invention.

FIG. 12 is a diagram showing the structure of 0CIC in the statistical multiplexing device according to the second embodiment of the present invention. BEST MODE FOR CARRYING OUT THE INVENTION

Hereinafter, examples of the present invention will be described.

[First embodiment]

FIG. 1 shows an optical communication device including the statistical multiplexing devices 1a and 1b of the present invention. 8 The signal added from the channel independent signal transmission channel array 5 is converted (compressed) into four signals by the statistical multiplexing device 1a and sent to the wavelength multiplexing device 2a. The statistical multiplexing device 1a and the wavelength multiplexing device 2a may be connected by a copper wire, or may be connected by an optical fiber.

The optical transmission channel array 5 is composed of, for example, an aggregate of eight Gigabit Ethernet transmission lines. The transmission medium of Gigabit Ethernet may be optical fiber or copper wire (coaxial cable or twisted pair). These eight gigabit Ethernet (transmission rate: 1.25 Gbps) signals are converted (compressed) into four transmission rate 1.25 Gbps signals by the statistical multiplexer 1a. To the wavelength multiplexing device 2a. These four signals are transmitted according to the physical layer of Gigabit Ethernet.

The wavelength-multiplexed optical signal 4 is transmitted on the optical fiber 3 and sent to the wavelength multiplexer 2b. The wavelength-multiplexed optical signal is composed of four wavelengths, for example, center wavelengths 1310 nm, 1345 nm, 1448 nm, and 1545 nm. The optical signal of each wavelength is modulated at a rate of 1.25 Gbps. The four wavelength optical signals received by the wavelength multiplexing device 2b are sent to the statistical multiplexer 1b in the form of four electrical signals or optical signals, and the eight optical signals are transmitted by the statistical multiplexer 1b. Converted (expanded) to signal transmission channel array 6.

The signal flow from the statistical multiplexing device 1a to the statistical multiplexing device 1b has been described above.The signal flow in the opposite direction, from the statistical multiplexing device 1b to the statistical multiplexing device 1a, has been described. The same multiplexing (compression-expansion) is performed for the flow of the signal to be transmitted. FIG. 2 is a diagram illustrating the operation principle of the statistical multiplexing devices 1a and 1b. FIG. 2 (a) shows the traffic of each of the eight independent signal transmission channel arrays 5 as 5a to 5h. Since the traffic always fluctuates, the total traffic sum of each channel traffic 5a to 5h of the signal transmission channel array 5 is as shown in Fig. 2 (b). Therefore, by rearranging the sum of the traffic of the signal transmission channel array 5, the signals of eight channels can be contained (compressed) in the signals of four channels of the same speed. Fig. 4 (c) shows that the signal compressed in this way is transmitted as a wavelength-multiplexed signal 4a or 4b. It shows the traffic when it is possible. Since compression is performed using the statistical effect, the above-mentioned compression / expansion is called statistical multiplexing, and the ratio between the original number of signals and the actual transmission path is called the compression ratio. . In the above case, the compression ratio is 2: 1 because eight signals are compressed into four signal transmission lines of the same speed.

In Fig. 1 and Fig. 2, for convenience of explanation, 8 gigabit Ethernet signals (transmission speed of 1.25 Gb ps) are statistically multiplexed into 4 signals of the same speed (1.25 Gb ps). (Compression-expansion) has been described. Generally, the average traffic on the signal transmission line is about 20% to 30% of the peak traffic, and it is difficult to obtain a compression ratio of about 4 times using the above-mentioned statistical multiplexing principle. Absent. That is, it is possible to statistically multiplex (compress and decompress) 16 giant Gigabit Ethernet signals into four signals of the same speed (1.25 Gbps).

When the compression ratio is set high, a situation 8 occurs in which the instantaneous traffic exceeds the maximum transmission capacity 7 (in this case, the total transmission capacity of the WDM signal path 4), as shown in Fig. 2 (b). Sometimes. The processing in this case will be described later.

FIG. 3 is a block diagram showing the internal configuration of the statistical multiplexer la or 1b. Network interface card (NIC) array 11, cross bar switch interface card (CIC) array 12, cross per switch 14, cross per switch control circuit 13, cross per switch interface card (cross per switch interface) It consists of CIC) array 15 and network interface card (NIC) array 16.

Fig. 4 (a) shows the structure of the network interface card 11a. It consists of an optical transceiver 41, a serial-to-parallel converter (SERDES) 42, and a medium control chip (MAC) 43. A serial optical signal input from an optical fiber cable (not shown) is converted into a serial electric signal by the optical transceiver 41, and then converted into a parallel signal by the serial-parallel converter (SERDS) 42. This parallel signal is interpreted for each packet by the media control chip (MAC) 43 and sent to the crossbar switch interface card 12a. Also, the signal in packet units sent from the cross-parsing switch 12a is converted into a parallel signal by the medium control chip (MAC) 43, and the serial-to-parallel converter (SERDE S ) The optical signal is converted into a serial electric signal by the optical transceiver, and further converted into an optical serial signal by the optical transceiver 41, and transmitted to an optical fiber cable (not shown). The network interface card 16a has a structure similar to that of the network interface card 11a.

Note that a packet is a unit of information transmission, and includes a destination address and a source address in addition to the information entity. The unit of transmission of this information may be called a cell, but these are almost synonymous.

In the above description, the medium control chip (MAC) 43 has a function of distinguishing between a packet and a non-packet part (idle signal). Further, the medium control chip (MAC) 43 can identify the source address and the destination address for each packet. Statistical multiplexing (compression-decompression) is realized by combining this function of the medium control chip (MAC) 43 with the function of the cross-par switch.

By using the cross-persistent method, it is possible to construct a statistical multiplexing apparatus that functions for a transmission channel having a higher transmission rate than other methods, for example, a shared memory method.

Fig. 4 (b) shows the structure of the cross-par switch interface card 12a. It consists of a first-in first-out memory (FIFO) 44, a media control chip (MAC) 43, a serial-to-parallel converter (S ERD ES) 42, and a copper wire transceiver 46. The packet unit signal sent from the network interface card 11a is stored in a first-in first-out memory (FIFO) 44 and then sent to a medium control chip (MAC) 43 to be converted into a parallel signal. The serial signal is converted into a serial signal by a serial-to-parallel converter (SERDES) 42, and further converted by a copper wire transceiver 46 into a serial electric signal that can be transmitted several meters over a copper wire, and sent to the cross-par switch 14. . The amount of memory stored in the first-in first-out memory (FIFO) 44 is sent to the crossbar switch control circuit 13 via a signal line 45. In addition, the serial electric signal sent from the crossbar switch 14 is transmitted through the transceiver 46 for copper wires, the serial-parallel converter (S ERD ES) 42, and the medium control chip (MAC) 43 in the opposite direction to the above, and the FIFO is first-in first-out. Stored in memory (FIFO) 44. The signal of each packet stored in the first-in first-out memory (FIFO) 44 is stored in the network interface. Sent to one face card 1 la. The cross-switch switch-in face card 15a has the same structure as the cross-bar switch interface card 12a.

Returning to FIG. 3, the operation of the statistical multiplexing device 1a will be described. The signals from the eight-channel independent signal transmission channel array pass through a network interface card (NIC) array 11, a cross-par switch interface card (CIC) array 12, and a crossbar switch 14. Sent to From the crossbar switch interface card (CIC) array 12, the signal amount stored in each crossbar switch interface card (CIC) is sent to the crossover switch control circuit 13. The crossbar switch control circuit 13 changes the connection pattern of the crossover switch according to the remaining signal amount stored in each crossbar interface card (CIC). The change of the connection pattern of the cross per switch is performed for each packet unit described above. Note that the packet length is not always a fixed value. The extraction of the packet (removal of the idle signal) and detection of the source address and the destination address are performed by the medium control chip (MAC) 43. The detection result of the medium control chip (MAC) 43 is sent to the crossbar switch control circuit 13 to change the connection pattern of the crossover switch.

Fig. 5 shows the state of the most basic control. This is a control method in which a cross-per-switch interface card 12a and a cross-bar switch interface card 12b are alternately connected to a cross-bar switch 15a. One evening, the signal 17 from the sparse switch interface card 12a is sent to the crossover switch interface card 15a, and the other time, the signal 1 from the crossbar switch interface card 1 2b is sent. 8 is sent to the crossbar switch face card 15a. The cross-par switch control circuit 13 compares the signal remaining in the cross-switch interface card 12a with the signal remaining in the cross-per switch interface card 12b. 5 Control cross-par switch 14 to connect to a. Then, the remaining signal of the connected crossbar interface card, for example, 12a, decreases. It does not increase at least. On the other hand, In the case of a new cross-peripheral interface card 1 2b, the signal level is increasing. When the remaining signal level of the crossbar switch-in switch card 1 2 b exceeds the remaining signal level of the cross-par switch-in switch card 1 2 a, the cross-par switch control circuit 13 is switched to the crossbar switch. The crossbar switch 14 is controlled so that the switch interface 12b is connected to the crossover switch interface 15a.

This control will be described with reference to FIG. Fig. 6 (a) Network interface-This is the signal on the face card 11a side. An idle signal 20 (Idle) is placed between packet 21 (DATA1) and packet 22 (DATA2). Existing. FIG. 6 (b) shows the signal on the side of the network interface card 11b, and the idle signal is also transmitted between the packet 23 (DATA 3) and the packet (DATA 4). Issue 20 (Idle) exists. By dynamically rearranging these two signals, a signal containing a high concentration of packets is formed as shown in Fig. 6 (c).

For the sake of simplicity, we have considered the case where two crossbar switch-in face cards 12a or 12b are alternately connected to a crossbar switch interface card 15a. In practice, the cross-parse switch control circuit 13 checks the remaining signal amount in the cross-bar switch interface array 12 and dynamically changes the connection pattern to the cross-bar switch interface array 15. This is shown in Figure 7. Fig. 7 (a) shows the connection of the D-subswitch 14 at a certain moment. This pattern changes at the next moment as shown in Fig. 7 (b). The signal multiplexed (compressed) as described above is restored (expanded) by the statistical multiplexer 1b. FIG. 8 is a diagram showing the internal configuration of the statistical multiplexer 1b. Network interface card (NIC) array 15, crossbar switch interface (CIC) array 16, cross-par switch 14, cross-switch controller 13, cross-bar switch interface It consists of a card (CIC) array 12 and a network interface card array (CIC) 11. The signal transmitted after being compressed is restored to the original signal by a process opposite to that of the statistical multiplexer 1a. FIG. 9 (a) shows a signal received by, for example, the network interface card 15a. This signal is distributed by the cross switch control circuit 13 to one of the cross switch interface card (CIC) arrays 12 for each destination of each packet. Fig. 9 (b) shows the signals received by the crossbar switch-in face card (CIC) 12a, and Fig. 9 (c) shows the signals received by the cross-parse switch interface card (CIC) 12b. In summary, the opposite operation from that of Fig. 6 is being performed.

For convenience of explanation, the statistical multiplexer 1a shown in FIG. 1 has been described as operating only for compression and the statistical multiplexer 1b as operating only for decompression. Both devices la and 1b have compression and decompression functions, and are capable of bidirectional signal transmission.

[Second embodiment]

The second embodiment will be described with reference to FIGS. 10 to 12. As described above, traffic congestion may cause a situation 8 that exceeds the maximum transmission capacity 7 (in this case, the total transmission capacity of the wavelength multiplexed signal path 4) shown in Fig. 2 (b).

FIG. 10 explains how to deal with such a situation 8. The traffic 51 exceeding the maximum transmission capacity 7 is stored in the memory and then sent as the delay traffic 52. This is also performed in the first embodiment by storing the data in the first-in first-out memory (FIFO) 44 in the crossbar switch interface card (CIC) array 12 at the time of compression. However, the traffic may be congested so that the first-in-first-out memory (FIFO) 44 provided in each cross-switch interface card (CIC) array 12 cannot be stored.

FIG. 11 shows a second embodiment of the statistical multiplexer. In this embodiment, even when the traffic is congested as described above, the occurrence of memory overflow can be more robustly prevented.

In FIG. 11, a major difference from the first embodiment is that a crossbar switch interface card (0 CIC) 31a to 31b for dealing with overflow is provided. In Fig. 12, in addition to traffic 32a to 32d, If there is still more traffic 33a to 33b, and there is not enough memory in the corresponding switch interface card 12, traffic 33a to 33b Is controlled so that it flows through the o-barf mouth-to-mouth cross-par switch interface card (OCIC) 3 la or 3 1 b. Cross-Purpose Switch Interface Card (OCIC) for handling flow 1 3a or 3 lbs has a larger first-in, first-out memory than Cross-Parent Switch Interface Card (CIC) 12 If the transmission capacity is exceeded, traffic can be absorbed for a long time. Crossover Switch Card for Overflow-Switch Interface Card (OCIC) The traffic stored in 3 la to 3 lb passes through the crossbar switch 14 when congestion is resolved, and the crossover switch interface card (CIC) array 15 Sent to the side.

Fig. 12 shows a crossover switch interface card for handling overflow.

(OCIC) shows the internal structure of 3la to 31b. Copper wire transceiver 46a, serial to parallel converter (S ERD ES) 42a, medium control chip (MAC) 3a, large capacity first in first out memory (FI FO) 47, medium control chip (MAC) 43b, serial parallel It consists of a converter (SERDES) 42b and a copper transceiver 46b. The signal from the output side of the cross-par switch 14 (the cross-bus switch interface array 15 side) is converted to a parallel signal by the copper wire transceiver 46a and serial-to-parallel converter (S ERDES) 42a. You. This parallel signal is interpreted for each packet by the medium control chip (MAC) 43a, and then stored in the large capacity re-start destination memory (FIFO) 47. Then, the medium control chip (MAC) 43b, serial-parallel converter (S ERDES) 42b, and copper wire transceiver 46b pass through the input side of the cross-par switch 14 (cross-bus switch interface power array 1). 2).

[Modification of Embodiment]

As described above, for the sake of explanation, the speeds of the signal transmission channel array 5 and the four wavelength-multiplexed signal transmission lines 4 are assumed to be the same. However, this speed can be different. For example, signal transmission channel array 5 has eight gigabit Ethernet (Code rate: 1.25 Gb ps), and the wavelength-division multiplexed signal 4 may be four 0 C—12 (code rate: 622 Mb ps). Due to the difference in coding format, the information rate of Gigabit Ethernet is 1.0 Gb ps, and the information rate of OC-12 is about 622 Mb ps. When compared at the information rate, the two do not have a simple integer ratio relationship, but efficient transmission can be performed even in such a case through a statistical multiplexing device.

Although the case of statistical multiplexing (compression-decompression) of an 8-channel signal to a 4-channel signal has been described, the number of channels is for convenience of explanation, and any number of channels can be used. Usually, the number of channels after compression is smaller than the number of channels before compression, but the present invention is not necessarily limited to such a case. For example, it is also possible to statistically multiplex 2-channel Gigabit Ethernet (code rate 1.25 Gb ps, information rate l Gbps) into 8-channel OC-3 (code rate = information rate = 155 Mb ps). It is included in the scope of the present invention. In this case, two Gigabit Ethernet signals with an information rate of 2 Gbps are compressed and sent to a 0C-3 transmission line array with an information rate of 1.22 Gbps. It can send a new standard signal (gigabit Ethernet) to the existing network (OC-3).

Alternatively, the transmission speeds of the transmission channels after compression may be uneven. For example, two Gigabit Ethernets may be multiplexed into one 0C-12 and four 0C-3 and sent. In this case, too, it is valuable in that signals of the new network standard can be transmitted to the existing network. In other words, the equipment group that has already been installed can be used effectively.

Further, the transmission speeds of the transmission channels before compression may be uneven. The transmission channel group before compression may be a mixture of Fast Ethernet and Gigabit Ethernet with a transmission speed of 100 Mbps. This is also economically valuable in terms of effective use of existing resources.

In order to change the speed of the connected transmission channel non-uniformly, for example, in the first embodiment (FIG. 3), the network interface card 11a and the crossbar switch interface card 12a are used. You can change the speed with. K The speed of the cross-parse interface card (CIC) connected to the lossy switch 13 must be uniform. However, the transmission speed of the crossbar switch and the speed of the network interface card 12a can be set separately. The network interface card 1 la and the crossover switch interface card 1 2 a are connected by a bus, and the first-in first-out memory (FIF 0) 44 is mounted on the crossbar switch interface card 12 a Therefore, buffering can be performed for different transmission rates.

In addition, although the description has been given of the case where the statistical multiplexing is performed in combination with the wavelength multiplexing, this is just for convenience of description. A parallel channel may be constructed by arranging optical fibers, and statistical multiplexing (compression-decompression) may be performed on the parallel channel. Alternatively, a parallel channel made of copper wire instead of optical fiber may be constructed. Further, wavelength multiplexing may be further performed for each of the parallel channels. Further, the statistical multiplexing (compression-decompression) of the present invention can be applied to a medium in which the band is divided in a form such as time division multiple access. Industrial applicability

According to the present invention, it is possible to extract (compress) only valid packet portions from a plurality of signal channel sources, transmit and receive through a plurality of transmission channels, and then restore (decompress) a compressed signal. Efficiency can be improved. In addition, existing transmission channels with different transmission speeds can be used effectively.

Claims

The scope of the claims

1. In a statistical multiplexing apparatus having an uncompressed transmission channel group composed of a plurality of transmission channels and a compressed transmission channel group composed of a plurality of transmission channels,

The sum of the transmission capacities of the uncompressed transmission channels is greater than the sum of the transmission capacities of the compressed channels;

Extracting means for extracting a packet portion from a signal sent from the uncompressed transmission channel group, and first relocation means for dynamically relocating the extracted packet to the compressed transmission channel group. A statistical method comprising: compressing a signal by reducing a ratio of an idle signal in a signal transmitted to the compressed transmission channel by the extraction means and the first rearrangement means. Multiplexer.

2. The statistical multiplexing apparatus according to claim 1, wherein a destination detecting unit that detects a destination of the bucket from a dynamically rearranged signal transmitted from the compressed channel group, and a destination corresponding to the uncompressed channel. A statistical multiplexing apparatus comprising: a second rearrangement unit for rearranging the packet in a channel; and expanding the signal by the destination detection unit and the second rearrangement unit.

3. The statistical multiplexing device according to claim 1, wherein the uncompressed channel and the compressed channel have different transmission speeds.

4. The statistical multiplexing device according to claim 1, wherein the speeds of transmission channels included in the uncompressed channel group are not uniform.

5. The statistical multiplexing device according to claim 1, wherein the speeds of transmission channels included in the compressed channel group are not uniform.

6. The statistical multiplexing apparatus according to claim 1, further comprising a crossbar switch.

7. The statistical multiplexing apparatus according to claim 6, wherein traffic is bypassed to an internal channel equipped with a large-capacity memory at the time of traffic congestion.