JP2011049674A - Power saving method using shared buffer in communication network and communication node - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a power saving method in a communication network where a plurality of communication nodes are connected through a communication path to a first communication node. <P>SOLUTION: The second communication node is operated in a sleep mode only for the sleep time specified by a sleep message and operated in a normal mode after the lapse of the sleep time. The first communication node includes two kinds of shared buffers for storing communication data through the second communication node, a sleep state flag for indicating the presence/absence of the second communication node being operated in the sleep mode, a scheduler for reading the communication data from the shared buffers and outputting them to a communication path, and a distributing means in which buffer storage is performed according to the operation of the second communication node in the normal mode or the sleep mode. Also, a scheduler determines the communication data stored in which buffer are read by referring to the sleep state flag. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a power saving method and a communication node capable of suppressing power consumption in a communication network.

  As a method for realizing power saving in the communication network, there is a method in which power is supplied only to the minimum necessary functional blocks and other functional blocks are suspended during a period when the communication node does not transmit and receive data. A state in which the functional block is suspended is generally called a sleep mode, and a function for suspending a specific functional block is called a sleep function. The operation in the sleep mode is referred to as “sleeping”. For example, Patent Literature 1 discloses a method for realizing power saving by shifting to a sleep mode during a period in which data transmission is stopped by flow control between communication nodes.

  Further, the above flow control can be performed for each logical line, not between the connected communication nodes. For example, Patent Literature 2 discloses a method for realizing flow control for each logical line called a VLAN (Virtual Local Area Network).

  Furthermore, as a flow control method, there is also known a method of stopping data transmission by designating priority, instead of stopping all data transmission from the communication node. For example, Non-Patent Document 1 discloses a flow control method in which priority can be specified.

  Also in the optical access network, a method for realizing power saving by giving a sleep function to a communication node has been studied. For example, Non-Patent Document 2 describes an optical access network called EPON (Ethernet Passive Optical Network) between an OLT (Optical Line Terminal) and an ONU (Optical Network Unit). A procedure for exchanging messages and transitioning an ONU to sleep mode is disclosed.

  Here, in the power saving method using the sleep function, there is a problem of how to handle the communication data when the communication data passing through the sleeping communication node is generated.

  If the generated communication data may be discarded, that is, communication data that does not cause a problem even if the communication node does not receive normally, no special consideration is required. In other words, data may be transmitted to the communication node without going to sleep, and communication data may be discarded without being normally received because the communication node is sleeping. Alternatively, when another communication node (management node) connected to the communication node manages the sleep state and the sleep state is in the sleep state, the management node may implement a method of discarding communication data addressed to the communication node. it can.

  However, communication data that cannot be discarded may occur due to the application or communication service. In this case, there is a need for a method for handling communication data that passes through the sleeping communication node without being discarded.

  In order to solve the above problem, a method of storing data in a communication node (parent node) to which a communication node (child node) having a sleep function is connected is known. For example, communication data passing through a sleeping child node is accumulated on the parent node side to which the child node is connected, and communication data is transmitted and received only at an appropriate timing when the child node shifts from the sleep mode to the normal mode. The method is known.

  FIG. 9 shows a configuration of a communication network for realizing a power-saving communication network that avoids discarding of communication data that passes through a sleeping child node using the above-described conventional technology. In FIG. 9, a plurality of child nodes 3 are connected to the parent node 13 via the communication path 2. The parent node 13 is connected to the upper network 5, and each child node 3 is connected to the lower network 4.

(Sleep procedure)
FIG. 10 shows a procedure in which the parent node 13 shifts the child node 3-i to the sleep mode. First, the parent node 13 gives an instruction to transmit a control message to the child node 3-i, and the child node 3-i transmits a control message in response to the instruction.

  Next, the parent node 13 determines whether or not to shift the child node 3-i to the sleep mode. This determination is made based on the presence / absence of downlink communication data from the parent node 13 to the child node 3-i and the presence / absence of uplink communication data from the child node 3-i to the parent node 13.

  The presence or absence of downlink communication data can be identified by observing the flow rate of downlink communication data from the upper network 5 to the lower network 4-i in the parent node 13. On the other hand, the presence / absence of uplink communication data can be identified based on a control message transmitted from the child node 3-i to the parent node 13. The control message is transmitted to notify the parent node 13 of the communication data amount accumulated in the child node 3-i. The parent node 13 makes this determination for each child node.

  When the parent node 13 determines that the child node 3-i should be shifted to the sleep mode, the parent node 13 transmits a sleep message to the child node 3-i. The sleep message includes a time during which the child node 3-i sleeps (sleep time = STi).

  The child node 3-i reads the sleep time STi included in the received sleep message and immediately shifts to the sleep mode. At this time, power supply to all the functions of the child node is not stopped, but at least a timer for managing the sleep time is kept operating.

  When the sleep time STi elapses after receiving the sleep message, the child node 3-i starts immediately and shifts to the normal mode. The parent node 13 gives an instruction so that the child node 3-i transmits a control message at the time when the child node 3-i shifts to the normal mode. In response to the instruction, the child node 3-i transmits a control message indicating the presence / absence of the accumulated data amount or a control message for requesting the transition to the sleep mode to the parent node 13. Similar to the above, the parent node 13 determines whether or not to shift the child node 3-i to the sleep mode.

  By repeating this series of procedures and shifting the child node to the sleep mode at an appropriate time, power saving can be achieved.

(Parent node configuration)
FIG. 11 shows a configuration of a parent node according to the prior art. With reference to FIG. 11, a procedure for storing communication data passing through a sleeping child node without discarding the communication data will be described.

  The parent node 13 includes a buffer 15 for each child node and for each priority. That is, as shown in FIG. 11, the buffer 15 includes a buffer (15-1H to 15-3H) for storing high-priority communication data and a buffer (15-1L to 15-3L) for storing low-priority communication data. Have

  Communication data received by the parent node 13 from the upper network 5 is distributed to the buffer 15 by the distribution unit 14 and stored. The distribution unit 14 identifies the child node through which the communication data passes and the priority from the communication data, and accumulates them in the corresponding buffer 15. For example, if the communication data has high priority via the child node 3-1, the distribution unit 14 stores the communication data in the buffer 15-1H.

  The schedulers 16 and 17 read communication data from the buffer 15 and transmit the read communication data to the communication path 2. The scheduler 16 includes a scheduler 16-H that reads communication data from the high priority buffer and a scheduler 16-L that reads communication data from the low priority buffer.

  The scheduler 16 controls the reading of communication data with reference to the management table 18 that manages the operation state of the child node. Specifically, the communication data is not read from the buffer 15 in which the communication data passing through the child node 3 in sleep is stored, and only the buffer 15 in which the communication data passing through the child node 3 in the normal mode is stored. Read communication data from. In the example of FIG. 11, the management table 18 indicates that the child nodes 3-1 and 3-2 are sleeping, so the scheduler 16 -H reads communication data from only the buffer 15-3 H, and the scheduler 16 -L reads communication data from buffer 15-3L only. In the other buffer 15, the communication data remains accumulated because the child node through which it passes is sleeping.

  The scheduler 17 reads communication data with complete priority. That is, as long as there is communication data to be read by the scheduler 16-H, only the communication data read by the scheduler 16-H is read. The communication data read by the scheduler 16-L is read only when there is no communication data read by the scheduler 16-H.

  The parent node 13 manages the operation state of each child node 3 by using the management table 18. By transmitting the sleep message, the parent node 13 immediately records the corresponding child node in the management table 18 as sleeping, and records the child node in the management table 18 as the normal mode after the notified sleep time has elapsed. The management table 18 is updated.

  As described above, even in the prior art, it is possible to configure a power-saving communication network without discarding communication data passing through a sleeping child node.

JP 2008-263281 A JP 2003-224574 A

IEEE802.1Qbb Draft0.2 http://www.ieee802.org/3/av/public/2008_09/3av_0809_mandin_4.pdf

  However, when a buffer for storing data is provided as described above, the necessary buffer capacity can be excessively large. For example, if the number of child nodes is 32, the priority type is 4, and the maximum amount of data that can be stored during the sleep mode is 10 kilobytes, the parent node is 32 × 4 × 10 kilobytes only for realizing power saving. It must have a 1280 kilobyte buffer. This is clearly overkill.

  Data that passes through a child node that is sleeping does not always arrive at the parent node but arrives with a certain probability. For example, it is stochastically that the data passing through the sleeping child node arrives at the parent node at a maximum of about four devices, and the priority of data that must be accumulated without being discarded Is only one type, the required buffer capacity is 4 × 1 × 10 kilobytes = 40 kilobytes, which is 1/32 of the above capacity.

  On the other hand, a method using a shared buffer is also conceivable. For example, communication data is accumulated in a FIFO (First In First Out) queue as a shared buffer in the parent node, and the parent node reads only the data passing through the child node in the normal mode and passes through the child node in the sleep mode. Data is stored in a buffer.

  However, in the above method, although it is a FIFO queue, it is necessary to read data that passes through the child node in the normal mode before data that passes through the child node in the sleep mode, and management is complicated.

  In other words, in the conventional method, in order to accumulate the data passing through the child node in the sleep mode without being discarded, it is necessary to provide the parent node with an excessive buffer or to perform complicated buffer management. Providing these in the parent node is not only expensive, but may not reduce the power consumption of the entire communication network.

  The present invention has been made in view of the above-described circumstances, and reduces the buffer capacity necessary to avoid discarding communication data via a communication node that is sleeping when realizing a power saving method in a communication network. An object of the present invention is to provide a method for managing a buffer by simple means to avoid discarding communication data and such a communication node.

  In order to solve the above problems, the present invention provides a power saving method in a communication network in which a plurality of second communication nodes are connected to a first communication node via a communication path. The second communication node operates in a sleep mode that suppresses power consumption of the second communication node for the sleep time specified by the received sleep message, and after the specified sleep time has elapsed, And a means for operating in a normal mode that does not suppress power consumption. The first communication node includes two types of shared buffers for storing communication data passing through the second communication node, buffer A and buffer B, and the second communication node operating in the sleep mode. A sleep state flag indicating presence / absence, a scheduler that reads the communication data from the buffer A and the buffer B, outputs the communication data to the communication path, and communication data that passes through the second communication node that operates in the normal mode. Distribution means for accumulating in the buffer B communication data stored in A and passing through the second communication node operating in the sleep mode. The sleep time is specified for each second communication node so that each second communication node shifts from the sleep mode to the normal mode at the same time, and the sleep time Includes the time at each second communication node that is advanced by the time required for propagation of communication data from the first communication node to each of the second communication nodes via the communication path from the elapsed time. . In the power saving method according to the present invention, the distribution means stores the received communication data in the buffer A when the received communication data is communication data that passes through the second communication node operating in the normal mode. Is stored in the buffer B when the communication data passes through the second communication node operating in the sleep mode, and the scheduler has the sleep state flag at least one second communication node in the sleep mode. When the communication data stored only from the buffer A is read when indicating that it is operating, and when the sleep state flag indicates that all the second communication nodes are operating in the normal mode Reading communication data accumulated from the buffer A and the buffer B. To.

  Moreover, the 1st communication node in the communication network which concerns on this invention is provided. The first communication node is connected to a plurality of second communication nodes via a communication path, and receives communication data via the plurality of second communication nodes from another network. The second communication node operates in a sleep mode that suppresses the power consumption of the second communication node for the sleep time specified by the received sleep message, and after the specified sleep time has elapsed, the power consumption It operates in the normal mode that does not suppress The first communication node includes two types of shared buffers for storing communication data passing through the second communication node, buffer A and buffer B, and the second communication node operating in the sleep mode. A sleep state flag indicating presence / absence, and a scheduler that reads out the communication data from the buffer A and the buffer B and outputs the communication data to the communication path, and the sleep state flag indicates that at least one second communication node is in a sleep mode. When the communication data stored only from the buffer A is read and the sleep state flag indicates that all the second communication nodes are operating in the normal mode. A scheduler for reading out communication data accumulated from the buffer A and the buffer B; In order to store the communication data passing through the second communication node operating in the normal mode in the buffer A and storing the communication data passing through the second communication node operating in the sleep mode in the buffer B. Distribution means. The sleep time is specified for each second communication node so that each second communication node shifts from the sleep mode to the normal mode at the same time, and the sleep time Includes the time at each second communication node that is advanced by the time required for propagation of communication data from the first communication node to each of the second communication nodes via the communication path from the elapsed time. It is characterized by that.

  According to the present invention, a buffer capacity required by using a shared buffer to avoid discarding communication data via a sleeping communication node is reduced, and a power saving communication system is realized by a simple buffer management method. be able to.

It is a figure which shows the structure of the communication network in 1st Example of this invention. It is a sequence diagram which shows the procedure which transfers to sleep mode in 1st Example of this invention. It is a sequence diagram which shows the mode of the message exchange between the communication nodes in 1st Example of this invention. It is a figure which shows the management table and sleep state flag in 1st Example of this invention. It is a figure which shows the structure of the parent node in 1st Example of this invention. It is a figure which shows the structure of the communication network in 2nd Example of this invention. It is a figure which shows the mode of the message transfer between the communication nodes in the 2nd Example of this invention. It is a figure which shows the mode of the message transfer between the communication nodes in the 3rd Example of this invention. It is a figure which shows the structure of the communication network in a prior art. It is a sequence diagram which shows the procedure which transfers to sleep mode in a prior art. It is a figure which shows the structure of the parent node in a prior art.

  The best mode for carrying out the present invention will be described below with reference to the drawings. In addition, what is given the same number in a figure shall have the same characteristic.

(Network configuration)
A first embodiment of the present invention will be described below. FIG. 1 shows the configuration of a communication network in this embodiment. A plurality of child nodes 3 are connected to the parent node 1 via the communication path 2.

  In the present embodiment, an example is shown in which only three child nodes (3-1 to 3-3) are connected to the parent node 1. However, in the present invention, the number of child nodes is not limited. The same applies to the other embodiments. That is, for simplicity of explanation, only three child nodes are illustrated, but the number of child nodes is not limited to this.

  In FIG. 1, the communication path 2 connecting the parent node 1 and the plurality of child nodes 3 is not particularly limited. That is, the communication path 2 may be an IP (Internet Protocol) network configured by a router or the like, or a layer 2 network configured by an Ethernet (registered trademark) switch. Further, the parent node is a radio base station, A wireless network having a node as a wireless terminal may be used. Alternatively, a PON (Passive Optical Network) may be used in which a parent node is an OLT, a child node is an ONU, and a parent node and a plurality of child nodes are connected by point-multipoint connection using an optical fiber and an optical splitter. Alternatively, a combination of these may be used.

  The parent node 1 is connected to the upper network 5 and the child nodes 3 are connected to the lower network 4 respectively. Communication data from the upper network 5 to the lower network 4 is transferred via the parent node 1 and the communication path 2 via the appropriate child node 3. In this specification, communication data transferred from the upper network 5 to the lower network 4 is called downlink communication data, and communication data transferred in the opposite direction is called uplink communication data.

(Sleep procedure)
FIG. 2 shows a procedure for shifting the child node 3 to the sleep mode. This is almost the same as the procedure in the prior art shown in FIG. Hereinafter, a procedure for shifting the child node 3-i to the sleep mode will be described. First, the parent node 1 gives an instruction to transmit a control message to the child node 3-i, and the child node 3-i transmits a control message in response to the instruction.

  Next, the parent node 1 determines whether or not to shift the child node 3-i to the sleep mode. This determination is made based on the presence / absence of downlink communication data from the parent node 1 to the child node 3-i and the presence / absence of uplink communication data from the child node 3-i to the parent node 1. The presence or absence of downlink communication data can be identified by observing the flow rate of downlink communication data from the upper network 5 to the lower network 4-i in the parent node 1. On the other hand, the presence / absence of uplink communication data can be identified based on a control message transmitted from the child node 3-i to the parent node 1. The control message is transmitted to notify the parent node of the communication data amount stored in the child node 3-i. The parent node 1 makes this determination for each child node.

  In the present invention, the control message is not limited to a format for notifying the amount of accumulated data. A form in which the child node requests to enter the sleep mode may be used. Further, the above determination when the child node is shifted to the sleep mode may be made based on the presence / absence of communication data for each priority.

  When the parent node 1 determines that the child node 3-i should be shifted to the sleep mode, the parent node 1 transmits a sleep message to the child node 3-i. The sleep message includes a time during which the child node 3-i sleeps (sleep time = STi).

  The child node 3-i reads the sleep time STi included in the received sleep message and immediately shifts to the sleep mode. At this time, power supply to all the functions of the child node is not stopped, but at least a timer for managing the sleep time is kept operating.

  When the sleep time STi elapses after receiving the sleep message, the child node 3-i starts immediately and shifts to the normal mode. The parent node 1 gives an instruction so that the child node 3-i transmits a control message at the time when the child node 3-i shifts to the normal mode. In response to the instruction, the child node 3-i transmits, to the parent node 1, a control message indicating the presence or absence of the accumulated data amount or a control message for requesting a transition to the sleep mode. Similarly to the above, the parent node 1 determines whether or not to shift the child node 3-i to the sleep mode.

  By repeating this series of procedures and shifting the child node to the sleep mode at an appropriate time, power saving can be achieved.

(Synchronous sleep)
Next, a procedure for giving an instruction so that the parent node 1 shifts to the sleep mode for each child node 3 will be described with reference to FIGS.

  First, the parent node 1 gives a transmission instruction to the child node so that the time for receiving the control message from each child node 3 is concentrated as close as possible (S1).

  Next, the parent node 1 determines whether to shift to the sleep mode for each child node (S2). The child node to be shifted to the sleep mode is updated by writing information to that effect in the management table 6 provided in the parent node 1. Here, the child nodes 3-1 and 3-2 are shifted to the sleep mode.

  Furthermore, a sleep message is sequentially transmitted toward the child nodes 3-1 and 3-2 to be shifted to the sleep mode (S 3). At this time, as shown in FIG. 3, the sleep times ST1 and ST2 are set so that the child nodes 3-1 and 3-2 shift to the normal mode at the same time.

  Details will be described below. First, it is assumed that the propagation times from the parent node 1 to the child nodes 3-1 to 3-3 are D1 to D3, respectively. Further, it is assumed that the parent node sleeps the child node from time Ts to time Ta, and the length of the time is Ta−Ts = ST. Further, when the propagation time through the communication path 2 is not constant, the maximum propagation time is assumed in advance, and the value can be set to D1 to D3.

  As shown in FIG. 4A, the parent node 1 sets the sleep state flag 7 included in the parent node 1 at time Ts, and records that there is a child node operating in the sleep mode.

  The parent node 1 sequentially transmits a sleep message toward the child nodes 3-1 and 3-2. In this case, the actual transmission times are assumed to be Ts + ΔT1 and Ts + ΔT2, respectively. The time at which the sleep message arrives at the child nodes 3-1 and 3-2 is the time obtained by adding the respective propagation times. That is, if it is a child node 3-1, it will arrive at time Ts + ΔT1 + D1.

  The parent node 1 determines that the child node has shifted from the sleep mode to the normal mode at time Ta, resets the sleep state flag 7 and simultaneously updates the management table 6 as shown in FIG. Record that all child nodes are operating in normal mode. However, even if the parent node 1 transmits communication data toward the child node 3 at the time Ta, the above-described propagation times D1 to D3 are required until the communication data actually arrives at the child node. That is, the time at which the communication data transmitted from the parent node to the child nodes 3-1 to 3-3 at the time Ta arrives at the child node is Ta + D1, Ta + D2, and Ta + D3, respectively. Since the power saving effect is greater when operating in the sleep mode for as long as possible, the child node is put to sleep until the time when communication data can actually arrive.

That is, the sleep time to be notified to each of the child nodes 3-1 and 3-2 is (the time when communication data from the parent node can actually arrive) − (the time when the sleep message can be received). Is optimal. For example, the sleep time STi notified to the child node 3-i is:
STi = (Ta + Di) − (Ts + ΔTi + Di) = ST−ΔTi
It becomes. That is, this means a value obtained by subtracting the time ΔTi required for actually transmitting the sleep message from the predetermined sleep time ST. As a result, the child node 3 that has been put to sleep can shift to the normal mode at the same time.

(Accumulation procedure)
Next, a processing procedure for accumulating communication data passing through the sleeping child node in the parent node will be described with reference to FIG.

  The parent node 1 includes a buffer A and a buffer B as two types of shared buffers. In the buffer A, the communication data passing through the child node 3 in the normal mode is temporarily stored until the communication data is read out. The buffer B stores communication data that passes through the child node 3 in the sleep mode so that the communication data is not discarded even if the corresponding child node is in the sleep mode. Furthermore, the two types of shared buffers are provided only for the high priority communication data, and the low priority communication data is provided only for the buffer A, and the buffer B is not provided. In FIG. 5, the high-priority buffer A is indicated as buffer A-H, the high-priority buffer B is indicated as buffer B-H, and the low-priority buffer A is indicated as buffer A-L. Each buffer is not an individual buffer for each child node but a shared buffer.

  The buffer capacity of the buffer B-H is the amount of communication data that can reach the parent node at the same time as the communication data passing through the sleeping child node. Since the probability of occurrence of such communication data at the same time is low and the buffer stores only communication data with high priority, it can be said that the necessary capacity is sufficiently small.

  In addition, the parent node 1 includes a management table 6 that manages the operation mode for each child node, and a sleep state flag 7 that indicates the presence or absence of a child node that is sleeping.

  A procedure of processing in which the parent node 1 accumulates communication data will be described. First, the distribution unit 8 included in the parent node 1 identifies the child node 3 through which the communication data passes based on the information described in the communication data. Here, the information described in the communication data may be a VLAN tag identifier or an IP (Internet Protocol) header address. The distribution unit 8 also identifies the priority of the communication data based on the VLAN tag or IP header information. In this embodiment, the VLAN tag or the IP header is used as the identifier, but the present invention is not limited to these. The format is not limited as long as the information can identify the child node through which the communication data passes. Further, the format for identifying the priority is not limited.

  Next, the distribution unit 8 refers to the management table 6 and confirms the current state of the child node 3 through which the communication data passes. If the corresponding child node is operating in the normal mode and the communication data has high priority, the communication data is stored in the buffers A-H. If the corresponding child node is operating in the normal mode and the communication data has a low priority, the communication data is stored in the buffers A-L. On the other hand, if the corresponding child node is sleeping and the communication data has high priority, the communication data is accumulated in the buffer B-H. If the corresponding child node is sleeping and the communication data has low priority, the communication data is discarded.

  As described above, the sleeping child node 3 shifts to the normal mode at the same time. For this reason, when storing communication data in the buffer B, it is not necessary to distinguish each child node 3 that passes through it, and it is not necessary to adjust the storage order, so that the required buffer capacity can be reduced in this embodiment. And can be processed easily.

  In the present embodiment, the priority has been described as having two levels of high and low. However, the present invention is not limited to the number of priorities. Finer priorities (such as 4 levels) may be set, or no priorities may be set. In this embodiment, when the child node is in the sleep mode, only the high priority communication data is stored in the buffer B, and the low priority communication data is discarded. However, in the present invention, the parent node 1 may include a buffer B corresponding to all priorities set, or a buffer B corresponding to only the highest priority as in the present embodiment. May be. In other words, the priority with which the buffer B can be provided is a design parameter for the operator and is not limited.

(Reading procedure)
Next, a procedure of processing in which the parent node 1 reads communication data from the buffer and transmits the communication data to the child node 3 to the communication path 2 will be described with reference to FIG.

  The parent node 1 includes a scheduler 9-H and a scheduler 10. The scheduler 9-H refers to the sleep state flag 7, and when the sleep state flag 7 is set, the scheduler 9-H considers that there is a child node 3 in sleep, and the communication stored in the buffer B-H. Data is not read, but only communication data stored in the buffers A-H is read. Further, the scheduler 10 transmits the communication data read by the scheduler 9-H to the communication path 2. At that time, as long as there is communication data in the high priority buffer A-H, the scheduler 10 transmits the communication data read from the buffer A-H to the communication path 2 and accumulates it in the buffer A-H. Only when there is no more communication data, the scheduler 10 reads the communication data from the buffer A-L and transmits it to the communication path 2.

  On the other hand, when the sleep state flag 7 is reset, the scheduler 9-H assumes that all the child nodes 3 are operating in the normal mode, and both the buffer A-H and the buffer B-H. The communication data is read out from the scheduler 10, and the scheduler 10 transmits the read communication data to the communication path 2. In this embodiment, reading is performed according to the following procedure. First, the scheduler 9-H reads communication data stored in the buffer B-H. As long as there is communication data stored in the buffer BH, reading from the buffer BH is continued. The scheduler 9-H reads communication data from the buffer A-H only when the buffer B-H is empty. The scheduler 9-H continues reading from the buffer A-H as long as the buffer B-H is empty and there is communication data in the buffer A-H.

  The scheduler 10 reads communication data with complete priority. That is, the scheduler 10 reads communication data from the buffer A-L only when both the buffer B-H and the buffer A-H are empty.

  As described above, the sleeping child node 3 shifts to the normal mode at the same time. Therefore, when the sleep state flag 7 is reset, all the child nodes 3 are operating in the normal mode. Therefore, when the scheduler 9-H reads the communication data from the buffer B-H, the communication data It is not necessary to check whether or not the child node 3 through which is passed is in the sleep state, and can be processed easily.

  In the present embodiment, the read priority order by the scheduler 10 is set to the complete priority. However, the present invention is not limited to this. Further, the priority order of reading between the buffer A and the buffer B by the scheduler 9 is set to the complete priority of the buffer B in this embodiment, but is not particularly limited in the present invention. The embodiment may alternately read from the buffer A and the buffer B, and the reading order is a design parameter for the operator and is not limited.

  With the above configuration, it is possible to implement the present invention that can reduce the buffer capacity for avoiding discarding of communication data that passes through the communication node that is sleeping and that can easily manage the buffer.

  A second embodiment of the present invention will be described below. The configuration of the communication network in this embodiment is shown in FIG. The first embodiment and the second embodiment are different in the points described below.

  As shown in FIG. 6, a concentrator node 12 is connected between the parent node 11 and the upper network 5. The parent node 11 has the synchronous sleep management function described in the first embodiment, such as shifting the plurality of child nodes 3 to the sleep mode. The concentrator node 12 has a shared buffer function for accumulating communication data passing through the sleeping child nodes. The parent node 11 notifies the concentration node 12 of a notification message 19 including the identifier of the sleeping child node and the sleep time.

  In the present embodiment shown in FIG. 6, the parent node 11 and the concentrating node 12 are directly connected, but the present invention is not limited to this configuration. If the concentrating node 12 can receive the notification message 19 transmitted from the parent node 11, a communication network may be configured between the parent node 11 and the concentrating node 12.

  In the present embodiment, the parent node 11 has the sleep message transmission function in the first embodiment. On the other hand, the concentrator node 12 is similar to the parent node 1 in the first embodiment in that the buffer A-H, the buffer B-H, the buffer A-L, the management table 6, the sleep state flag 7, and the distribution data are stored. A unit 8, a scheduler 9-H, and a scheduler 10 are provided.

  As described with reference to FIG. 2, the parent node 11 determines whether to shift the child node 3 to the sleep mode based on the presence / absence of uplink / downstream communication data for each child node 3. The parent node 11 sets a sleep time and transmits a sleep message to the child node 3 to be shifted to the sleep mode. At that time, the parent node 11 notifies the concentration node 12 of a notification message 19 including the identifier of the child node 3 that transmitted the sleep message and the sleep time STc.

  FIG. 7 is a diagram illustrating message exchange between communication nodes in the present embodiment. The sleep time STc notified to the concentrating node 12 is adjusted and notified as shown in FIG. 7 (M1 in the figure). For simplicity of explanation, only the child node 3-1 is shown.

In this embodiment, STc is adjusted as follows. If the predetermined sleep time is ST (= Ta−Ts), ΔTc is the time required to notify the concentration node 12 from the parent node 11, and Dc is the propagation time between the parent node 11 and the concentration node 12, the sleep time STc is
STc = ST−ΔTc−2Dc
It becomes.

  On the other hand, the concentrating node 12 updates the management table 6 based on the identifier of the child node 3 that has transmitted the sleep message, notified from the parent node 11. Further, the concentrating node 12 sets the sleep state flag 7 based on the notified sleep time STc. The concentrated node 12 resets the sleep state flag 7 when the time STc after reception of the notification message 19 has elapsed, and updates the management table 6 on the assumption that all the child nodes 3 have shifted to the normal mode.

  As a result, even if the concentrating node 12 transmits communication data immediately after the sleep state flag 7 is reset, the communication data arrives at the parent node 11 at time Ta. Further, since the communication data is transferred to the child node 3 and arrives immediately after the child node 3 shifts to the normal mode, the child node 3 can normally receive the communication data.

  6 is a procedure for storing communication data in the buffers AH, BH, and AL, and a procedure for the scheduler 9-H and the scheduler 10 to read communication data from the shared buffer. This is the same as the first embodiment.

  As described above, the present invention can be implemented even if the function of buffering communication data and the function of transmitting a sleep message are separated into the concentrating node 12 and the parent node 11, respectively.

  A third embodiment of the present invention will be described below. The second embodiment and the third embodiment differ in the points described below.

  The parent node 11 in this embodiment notifies the concentration node 12 only of the identifier of the child node 3 that has transmitted the sleep message. Unlike the second embodiment, the parent node 11 does not notify the concentrating node 12 of the sleep time. The parent node 11 notifies the concentration node 12 that the child node 3 shifts to the normal mode at the time Ta when the child node 3 shifts to the normal mode. Specifically, as shown in FIG. 8, the propagation time required for the round trip between the parent node 11 and the concentrating node 12 is earlier than the time Ta at which the parent node 11 determines that the child node 3 has shifted to the normal mode. At time (Ta-2Dc), the parent node 11 notifies the concentrating node 12 that the normal mode is to be changed.

  The concentrator node 12 updates the management table 6 and sets the sleep state flag 7 when notified from the parent node 11 of the identifier of the child node 3 that has shifted to the sleep mode. On the other hand, when notified that the child node 3 has shifted to the normal mode, the concentrating node 12 updates the management table 6 and resets the sleep state flag 7 on the assumption that all the child nodes 3 have shifted to the normal mode. To do.

  In the present invention, the form in which the parent node 11 notifies the concentration node 12 of the transition to the normal mode is not limited. A direct information bit indicating the transition to the normal mode may be included in the notification message 19 or indirectly by notifying the information bit indicating that no child node 3 is shifted to the sleep mode. May be shown.

  As described above, the present invention can be implemented without the parent node 11 notifying the concentrating node 12 of the sleep time.

  As described above, according to the present invention, when realizing a power saving method in a communication network, the buffer capacity necessary to avoid discarding communication data via a communication node in sleep is reduced, and It is possible to provide a method for managing a buffer by simple means and such a communication node.

1, 11, 13 Parent node 2 Communication path 3 Child node 4 Lower network 5 Upper network 6, 18 Management table 7 Sleep state flag 8, 14 Distribution unit 9, 10, 16, 17 Scheduler 12 Concentration node 15 Buffer 19 Notification message

Claims (10)

A power saving method in a communication network in which a plurality of second communication nodes are connected to a first communication node via a communication path,
The second communication node is
Means for operating in a sleep mode for suppressing power consumption of the second communication node for a sleep time specified by a received sleep message;
Means for operating in a normal mode that does not suppress power consumption after elapse of the designated sleep time;
The first communication node is
Buffer A and buffer B, which are two types of shared buffers for storing communication data passing through the second communication node;
A sleep state flag indicating whether or not the second communication node is operating in the sleep mode;
A scheduler that reads the communication data from the buffer A and the buffer B and outputs the communication data to the communication path;
Distribution for storing communication data passing through the second communication node operating in the normal mode in the buffer A and storing communication data passing through the second communication node operating in the sleep mode in the buffer B Means,
With
The sleep time is designated for each second communication node such that each second communication node shifts from the sleep mode to the normal mode at the same time,
The simultaneous period is advanced from the time when the sleep time elapses by the time required for communication data to propagate from the first communication node to each of the second communication nodes via the communication path. Including the time at the second communication node,
The method
The distribution means stores in the buffer A when the received communication data is communication data that passes through the second communication node that operates in the normal mode, and the received communication data operates in the sleep mode. Storing the data in the buffer B when the communication data passes through
The scheduler reads communication data stored only from the buffer A when the sleep state flag indicates that at least one second communication node is operating in a sleep mode, and the sleep state flag Reading out the communication data accumulated from the buffer A and the buffer B when the second communication node indicates that all of the second communication nodes are operating in the normal mode. Method. The buffer A, the buffer B, and the scheduler are provided for each priority by the number of priorities identified by the communication data received by the first communication node,
The distribution means distributes and stores the communication data in a buffer A or a buffer B provided for each priority according to the priority indicated by the communication data.
The power saving method according to claim 1. The buffer A, the buffer B, and the scheduler are provided for each of one or more specific priorities that are smaller than the number of priorities identified by the communication data received by the first communication node,
The distribution unit discards the communication data when the communication data corresponds to a priority other than the specific priority and the second communication node through which the communication data passes operates in a sleep mode. ,
The power saving method according to claim 2. The sleep time specified by the sleep message is
4. The power saving method according to claim 3, wherein the time is set by subtracting a time required for actually transmitting the sleep message from a predetermined time. The communication data includes a VLAN (Virtual Local Area Network) tag,
The power saving method according to claim 4, wherein the distribution unit identifies an identifier of a second communication node through which the communication data passes and a priority of the communication data included in the VLAN tag. The communication data includes an IP (Internet Protocol) header,
The power saving method according to claim 4, wherein the distribution unit identifies an identifier of a second communication node through which the communication data passes and a priority of the communication data included in the IP header. The first communication node is
Means for setting the sleep time for operating the second communication node in the sleep mode for each second communication node;
The power saving method according to claim 5, further comprising: means for transmitting the sleep message including the sleep time to the second communication node. The communication network further includes a third communication node connected between the first communication node and the second communication node,
The third communication node is
Means for setting the sleep time for operating the second communication node in the sleep mode for each second communication node;
Means for transmitting the sleep message including the sleep time to the second communication node;
The means for notifying the first communication node of an identifier of each second communication node that is the destination of the transmitted sleep message and the sleep time. Power saving method. The communication network further includes a third communication node connected between the first communication node and the second communication node,
The third communication node is
Means for setting the sleep time for operating the second communication node in the sleep mode for each second communication node;
Means for transmitting the sleep message including the sleep time to the second communication node;
Means for notifying the first communication node of the identifier of each second communication node that is the destination of the transmitted sleep message;
The power saving method according to claim 5 or 6, comprising means for notifying the first communication node that the sleep time has elapsed. A first communication node in a communication network,
The first communication node is connected to a plurality of second communication nodes via a communication path and receives communication data via the plurality of second communication nodes from another network;
The second communication node operates in a sleep mode that suppresses power consumption of the second communication node for a sleep time specified by the received sleep message, and after the specified sleep time has elapsed, Operates in normal mode without power suppression,
The first communication node is
Buffer A and buffer B, which are two types of shared buffers for storing communication data passing through the second communication node;
A sleep state flag indicating whether or not the second communication node is operating in the sleep mode;
A scheduler that reads the communication data from the buffer A and the buffer B and outputs the communication data to the communication path, and the sleep state flag indicates that at least one of the second communication nodes is operating in a sleep mode. If the communication data stored only from the buffer A is read and the sleep state flag indicates that all the second communication nodes are operating in the normal mode, the buffer A and the buffer A Reading out the communication data accumulated from B, a scheduler,
In order to store the communication data passing through the second communication node operating in the normal mode in the buffer A and storing the communication data passing through the second communication node operating in the sleep mode in the buffer B And the sorting method of
With
The sleep time is designated for each second communication node such that each second communication node shifts from the sleep mode to the normal mode at the same time,
The simultaneous period is advanced from the time when the sleep time elapses by the time required for communication data to propagate from the first communication node to each of the second communication nodes via the communication path. A first communication node in a communication network, comprising a time in a second communication node.

Images