JP2015033051A - Dynamic band allocation method, station-side device, computer program and pon system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To increase the number of customer premise-side devices which can be accommodated, while maintaining a service level such as throughput or delay time.SOLUTION: A dynamic band allocation method includes: allocating a band of uplink data which are received in a grant cycle in the future, for each customer premise-side device on the basis of reports collected from a plurality of customer premise-side devices; and granting a transmission time of the reports and the uplink data for the next time to the customer premise-side device, respectively. In the case where a first ONU defined as an ONU specialized for low-delay service applications and a second ONU defined as an ONU available for Internet browsing applications are included as the customer premise-side devices, this allocation method includes thinning processing for performing the band allocation based on the report of which the transmission source is the first ONU, in a predetermined cycle that is longer than the grant cycle.

Description

  The present invention relates to an upstream dynamic bandwidth allocation method suitable for a station-side device constituting a PON (Passive Optical Network) system, a computer program for causing a computer to execute this method, and a station-side device performing the method And a PON system including a station-side device that performs the method.

A station-side device, an optical fiber network configured to branch from an optical fiber connected thereto to a plurality of optical fibers via an optical coupler, and a home-side device connected to each end of the branched optical fiber A PON system has already been implemented.
The station-side device of the PON system dynamically allocates an upstream band in a time division manner to a plurality of home-side devices in order to prevent uplink signal interference.

Specifically, the station side device receives a bandwidth request control frame (report: also referred to as a request) in which the amount of data to be transmitted in the upstream direction from each home side device is recorded in advance. Based on the data amount (request value), a bandwidth to be allocated to each home-side apparatus is determined, and a control frame (gate) for notifying (granting) a transmission-permitted bandwidth is transmitted.
In the gate frame for this grant, the uplink transmission start time and the transmission permission length (time equivalent value) are recorded, so that each home-side device has a predetermined amount at the predetermined time recorded in the gate frame. Data can be sent in the upstream direction.

  As for how the station side device allocates bandwidth to the bandwidth request reports from a plurality of home side devices, that is, as a dynamic bandwidth allocation method in the uplink direction performed by the station side device, one home side device Each time a report arrives, distributed DBA (Dynamic Bandwidth Allocation) that allocates bandwidth to the home device as needed, and reports collected from multiple home devices are collected centrally. There is a centralized DBA that comprehensively allocates bandwidth to a plurality of home-side devices based on the above.

Of the above two types of dynamic bandwidth allocation methods, the centralized DBA, in other words, based on reports collected from a plurality of home-side devices, each home-side device received by the station-side device within the next grant period. The bandwidth allocation of the uplink data is comprehensively performed, and the station side device grants each home side device the transmission time for the next report and the uplink data.
According to this centralized DBA, it is possible to perform priority control for preferentially allocating a bandwidth to a high-priority home-side device that tends to lack a bandwidth within a grant cycle range of a predetermined cycle length, There is an advantage that the bandwidth can be guaranteed for each home device.

On the other hand, communication devices that perform various applications are connected to the home device, and the performance required for each application differs.
For example, an optical IP phone or the like requires a small rate but a low delay is required during a call. In Internet browsing applications, throughput is required, bandwidth efficiency is high, and a fair bandwidth allocation method is required. Therefore, various centralized DBAs described in Patent Documents 1 to 6 have been proposed so far in order to cope with differences in applications.

Japanese Patent No. 3734732 Japanese Patent No. 3768421 Japanese Patent No. 3768422 JP 2008-289202 A JP 2012-257341 A JP 2013-74427 A

By the way, it is expected that the cost per user can be reduced if the number of branches is increased as much as possible to accommodate more home-side devices in one PON system.
However, simply increasing the number of house-side devices accommodated in a conventional PON system that performs centralized DBA increases the number of upstream bursts and increases optical overhead as the number of home-side devices increases. There's a problem.

Further, if the number of house-side devices to be accommodated is simply increased, the bandwidth required for collecting reports and the like increases, so that the bandwidth allocated to user traffic is reduced and the service may be deteriorated.
On the other hand, if the grant period (DBA cycle) is increased in order to increase the bandwidth efficiency, the delay time also increases accordingly, which may affect applications such as optical IP phones that require low delay. come.

  In view of the above-described conventional problems, an object of the present invention is to increase the number of house-side devices that can be accommodated while maintaining service levels such as throughput and delay time.

According to the present invention, based on reports collected from a plurality of home-side devices, an uplink data band to be received in a future grant period is assigned to each home-side device, and the next-time report and the uplink data transmission time are allocated. A dynamic bandwidth allocation method for granting to each of the home side devices, wherein when the home side device includes a first ONU and a second ONU defined below, a bandwidth based on the report whose source is the first ONU The allocation is performed in a predetermined cycle longer than the grant cycle.
First ONU: ONU specialized for low-latency service applications
2nd ONU: ONU that can be used for Internet browsing

  According to the present invention, it is possible to increase the number of house-side devices that can be accommodated while maintaining service levels such as throughput and delay time.

1 is a connection diagram schematically showing a configuration example of a PON system. It is a block diagram which shows the internal function of a station side apparatus. It is a figure which shows the example of a report and grant frame structure. It is a sequence diagram showing a general centralized DBA. It is a sequence diagram which shows an example of the thinning process of the grant in a call state. It is a flowchart which shows the content of the allocation process by an allocation execution part. It is a sequence diagram which shows the allocation schedule in a DBA cycle. It is an image figure which shows the data storage state of the upstream buffer of a home side apparatus. It is a figure which shows the time allocation of the DBA cycle in centralized DBA of a comparative example.

<Outline of Embodiment of the Present Invention>
Hereinafter, an outline of embodiments of the present invention will be listed and described.
(1) The dynamic bandwidth allocation method according to the embodiment of the present invention allocates a bandwidth of uplink data received in a future grant period to each home device based on reports collected from a plurality of home devices. The present invention relates to a dynamic bandwidth allocating method for granting the next-side report and the uplink data transmission time to the home device.

In addition, in the dynamic bandwidth allocation method of this embodiment, when the first ONU and the second ONU defined below are included as the type of the home device, the bandwidth based on the report whose transmission source is the first ONU The allocation is performed in a predetermined cycle longer than the grant cycle.
First ONU: ONU specialized for low-latency service applications
2nd ONU: ONU that can be used for Internet browsing

As a specific example of the “first ONU”, for example, there is a “telephone ONU” used only for an IP telephone application. Of course, applications that require low delay are not limited to IP telephones, but include, for example, power control for power saving in home electrical devices.
Further, specific examples of the “second ONU” include, for example, a “network ONU” that is used only for browsing applications, and a “shared ONU” that performs both low-delay service applications such as IP telephones and browsing applications.

According to the dynamic bandwidth allocation method of the present embodiment, when the first ONU and the second ONU are included in the home side device, the bandwidth allocation based on the report with the transmission source being the first ONU is set to a predetermined cycle longer than the grant cycle. Therefore, bandwidth allocation for the first ONU is not performed in each grant period, and the number of bandwidth allocations can be minimized.
For this reason, even if the number of first ONUs is increased, a decrease in the bandwidth allocated to the second ONU can be suppressed, and the number of house-side devices that can be accommodated is increased while maintaining service levels such as throughput and delay time. be able to.

(2) In the dynamic bandwidth allocation method of the present embodiment, the bandwidth allocation based on the report whose transmission source is the second ONU is the first request value for low delay included in the report and the second request exceeding this It is preferable that it is a multiple request method allocation process executed for each grant period based on the request value.
In this case, efficient and fair bandwidth allocation to the second ONU can be maintained by the multiple request method allocation processing performed for each second ONU at each grant cycle.

(3) In the dynamic bandwidth allocation method according to the present embodiment, the predetermined period can be changed according to the service state of the first ONU, and the predetermined period when the service state of the first ONU is not in service. Is preferably set to a time length necessary for the first ONU to maintain the MPCP link with the station side device.
The reason is that, in the case of the first ONU that is not in service, it is sufficient if the MPCP link with the station side device can be maintained, and therefore the predetermined period may be extended to such an extent that no link disconnection occurs.

(4) Further, it is preferable that the predetermined period when the service state of the first ONU is in service is set to, for example, an uplink frame transmission period used for the low-delay service.
The reason is that in the case of the first ONU in service, it is sufficient if the uplink frame used for the low-delay service can be transmitted according to the rules, and therefore the predetermined period may be extended to about the transmission cycle.

  (5) However, since the transmission cycle of the upstream frame used for the low-delay service (for example, about 20 milliseconds in the case of a telephone frame) may be relatively long, when setting a predetermined period to the transmission cycle, A communication frame of another application (for example, sleep control using an extended MAC frame) may be forced to wait without being granted.

Therefore, the predetermined period when the service state of the first ONU is in service may be set to a time length equal to or shorter than the maximum delay time allowed for the low delay service.
The reason is that the maximum delay time allowed for the low-delay service is shorter than the transmission period of the upstream frame used for the low-delay service, so that the grant thinning process can be performed without affecting the operation of other applications. This is because it can be executed properly.

(6) In the dynamic bandwidth allocation method of the present embodiment, the time length for granting to the first ONU can be changed according to the service state of the first ONU, and the service state of the first ONU is not in service. The time length in this case is preferably set to a fixed length at least enough to transmit the report.
The reason is that, in the case of the first ONU that is not in service, it is sufficient if the MPCP link with the station side device can be maintained, and it is sufficient if a report necessary for maintaining the link can be transmitted.

(7) In the dynamic bandwidth allocating method of the present embodiment, the time length when the service state of the first ONU is in service is at least the amount that can transmit the report and the uplink frame used for the low-delay service. It is preferably set to a fixed length.
The reason is that, in the case of the first ONU in service, it is not sufficient to maintain the MPCP link by the report, and the service cannot be provided unless the uplink frame used for the low delay service is transmitted.

(8) In the dynamic bandwidth allocation method of the present embodiment, it is preferable that scheduling is performed from the front side in the order of region X, region Y, and region Z defined below in the grant period for performing bandwidth allocation of the first ONU.
Area X: Time area for assigning second ONU report Area Y: Time area for assigning second ONU uplink data Area Z: Time area for assigning first ONU report and unified burst of uplink data

  The reason is that if the unified burst of the first ONU is included in the region X, the end of the region X is delayed to the rear side, and the report of the second ONU cannot be collected by the next cycle, and the low-delay service for the second ONU This is because there is a possibility that it will not be possible to maintain.

(9) In the above case, the bandwidth allocation based on the report included in the region Z of the current grant period is performed simultaneously with the bandwidth allocation based on the report included in the region X of the next grant period. It is preferable to process.
The reason is that if the zone Z is scheduled last, the report of the first ONU cannot be collected in the next cycle, but if the report of the first ONU in the current cycle is processed together with the report of the second ONU in the next cycle, the process related to the first ONU This is because the delay is a delay of a grant period (for example, 600 μs), and the QoS of the low delay service of the first ONU can be maintained.

  (10) The computer program of this embodiment relates to a computer program that causes a computer to function as an allocation execution unit that executes the above-described dynamic bandwidth allocation method. Therefore, the computer program of this embodiment has the same operation and effect as the above-described dynamic bandwidth allocation method.

(11) The station side apparatus of this embodiment is related with the station side apparatus which performs the above-mentioned dynamic band allocation method. Therefore, the station side apparatus of this embodiment has the same operation effect as the above-mentioned dynamic band allocation method.
Note that, in the station side apparatus of the present embodiment, for example, for the transmission of the uplink data remaining in the grant period when the report of the first ONU is also granted each time in the same manner as the second ONU. It is particularly effective to execute a thinning process that does not grant each time for the first ONU under the condition that the remaining time is smaller than the total service bandwidth for the second ONU.

(12) The PON system of the present embodiment relates to a PON system including a station side device that executes the above-described dynamic band allocation method. Therefore, the PON system of the present embodiment has the same effects as the above-described dynamic band allocation method.
In the PON system of the present embodiment, for example, when the report of the first ONU is granted every time as in the second ONU, the remaining for transmission of the uplink data remaining in the grant period It is particularly effective to execute a thinning process that does not grant each time for the first ONU under the condition that the time is smaller than the largest RTT in the second ONU.

<Details of Embodiment of the Present Invention>
Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
[Overall configuration of PON system]
FIG. 1 is a connection diagram schematically showing a configuration example of a PON system.
As shown in FIG. 1, the PON system has a tree structure of an optical line terminal (OLT) 1 on the telecommunications carrier side and an optical network unit (ONU) 2 on the home side. It is a connection form (topology) connected by the PON line 8.

The optical line termination device 1 on the provider side is composed of an assembly in which a plurality of optical subscriber termination units (OSUs) 2 are housed in one casing, and a PON line 8 is connected to each termination panel. Has been.
In this embodiment, the optical line termination device on the telecommunications carrier side is referred to as a “station side device”, and the optical line termination device on the home side is referred to as a “home side device”. Further, “station side device” may be abbreviated as “OLT” and “home side device” may be abbreviated as “ONU”.

One optical fiber (main line) 5 connected to the station side device 1 has a plurality of optical fibers (branch lines) 7 branched from an optical coupler 6 which is a passive optical branching node, and a PON line 8 made of an optical fiber network. It is composed. The home apparatus 2 is connected to each of the optical fibers 7, 7... Branched from the optical coupler 6 of the PON line 8.
The station side device 1 is connected to the upper network 9. The home device 2 is connected to various communication devices 10A and 10B used by each user at home.

In FIG. 1, a total of five home-side devices 2 are illustrated, but in this embodiment, one or more optical couplers 6 branch at a branch number n (for example, n = 256) exceeding 128, and the number of branches. It is assumed that the home side device 2 corresponding to n is connected.
In the connection form shown in FIG. 1, only one optical coupler 6 is used. However, by arranging a plurality of optical couplers 6 in a column, the home-side devices 2 dispersed in a wide area are compared. It is also possible to connect to the station side device 1 using short optical fibers 5 and 7.

The PON system of this embodiment is composed of “symmetric 10G-EPON” in which the transmission rate of both upstream and downstream in the PON line 8 is 10 G (baud rate is 10.3125 Gbps).
In this case, the station side device 1 is composed of “10G symmetric OLT” capable of transmitting and receiving a 10 G optical signal to the optical fiber 5, and the home side device 2 can transmit and receive a 10 G optical signal to the optical fiber 7. It consists of “10G Symmetric ONU”.

Of course, “1GONU” in which both the upstream and downstream transmission rates are 1G (baud rate is 1.25 Gbps), and “asymmetric ONU” in which the downstream transmission rate is 10G and the upstream transmission rate is 10G are PON. It may be included in the system.
In this case, by adopting the station-side device 1 corresponding to the transmission rates of 1G and 10G in the upstream direction, a “multi-rate PON system” having a plurality of upstream transmission rates is obtained.

Uplink access control of the home side device 2 by the station side device 1 is basically performed in accordance with the communication system of 10G-EPON.
That is, the amount of data that the home device 2 wants to send is described in a report R (control frame for bandwidth request: also called “request”) in units of TQ (Time Quanta: 16 ns), for example, and gate G (grant transmission) In the control frame, the station side device 1 writes the transmission permission length and transmission start time in TQ units. The time on the PON line 8 is expressed by a counter that is incremented every 16 ns (= TQ) and is synchronized in the PON system.

As shown in FIG. 1, the PON system according to the present embodiment includes three types of ONUs 2A to 2C defined below.
“Telephone ONU 2A”: An ONU 2 to which the IP telephone 10A is connected, but communication devices for browsing the Internet such as the personal computer 10B are not connected.
The telephone ONU 2A is an ONU customized for use in replacing the conventional metal line fixed telephone with the IP telephone 10A (hereinafter also referred to as “telephone winding application”) at the user's home of the fixed telephone.

“ONU2B for network”: An ONU2 to which a communication device for browsing the Internet such as the personal computer 10B is connected but the IP telephone 10A is not connected.
The network ONU 2 </ b> B is an ONU specialized for browsing applications that is used to connect only the communication device 10 </ b> B for Internet browsing applications to the Internet via the PON.

“Shared ONU 2C”: The shared ONU 2C is an ONU 2 to which both an IP telephone 10A and a communication device for browsing the Internet such as a personal computer 10B are connected.
The shared ONU 2C is an ONU customized to support both a telephone winding application and an Internet browsing application.

In the station side apparatus 1 of this embodiment, the content of the allocation process performed with respect to the subordinate home side apparatus 2 is changed according to the types 2A to 2C of the home side apparatus 2 defined as described above.
Therefore, the telecommunications carrier determines in advance which type 2A to 2C of the home side device 2 the home side device 2 is the home side device 2 according to the application characteristics of the communication devices 10A and 10B connected to the home side device 2. Set to device 1. Accordingly, the station side device 1 stores in advance which type 2A to 2C of the home side device 2 the subordinate home side device 2 is associated with the MAC address or the like.

The telephone ONU 2A employs a “single request method” in which only one request value R1 is written in a report R for requesting a bandwidth to the station side device 1.
Further, the network ONU 2B and the shared ONU 2C (hereinafter collectively referred to as “ONU with network”) describe two request values R1 and R2 in a report R for requesting a bandwidth to the station apparatus 1. "Multiple request method" is adopted.

[Configuration of station side equipment]
FIG. 2 is a block diagram showing the internal functions of the station side device 1 of the present embodiment.
In FIG. 2, the station side apparatus 1 receives a downlink signal from the higher level network 9 for processing of a downlink signal from the higher level network 9 to the home side apparatus 2, and a buffer that temporarily stores the received downlink signal. 102 and a transmission unit 103 that transmits the downlink signal temporarily stored in the buffer 102 to the home apparatus 2.

In addition, the station side device 1 includes a receiving unit 104 that receives an upstream signal from the home side device 2 and a buffer 105 that temporarily stores the received upstream signal for processing of the upstream signal from the home side device 2 to the upper network 9. And a transmission unit 106 that transmits the upstream signal temporarily stored in the buffer 105 to the upper network 9.
Furthermore, the station side apparatus 1 includes a dynamic band allocation unit 107 that dynamically allocates the upstream band used by the subordinate home side apparatus 2 according to the request values R1 and R2 of the report R.

The dynamic band allocation unit 107 includes a request reception unit 108, a calculation unit 109, an allocation execution unit 110, a grant transmission unit 112, and a storage unit 113.
The storage unit 113 stores an identifier (identifier indicating one of the telephone 2A, the network 2B, or the shared 2C) that indicates the type of the home device 2 set by the communication carrier.
In addition, the storage unit 113 stores a predetermined reference table in which the minimum guaranteed bandwidth is described for each ONU 2B, 2C with respect to the ONUs 2B, 2C with a net.

Further, the storage unit 113 stores a computer program for causing the allocation execution unit 110 to execute information processing unique to the present embodiment, such as a centralized DBA, a thinning process, and a threshold value Th changing process described later. . The allocation execution unit 110 reads the program from the storage unit 113 and executes it to perform information processing on the program.
This computer program is fixed to a storage medium or transmitted / received as electronic data through a public network, and is distributed.

FIG. 3A is a diagram illustrating an example of a frame configuration of a report R transmitted from the home-side device 2, and FIG. 3B is a diagram illustrating an example of a frame configuration of the gate G transmitted from the station-side device 1. is there.
As shown in FIG. 3A, one report R can include two types of request values R1 and R2. These data amounts are represented by numerical values in TQ units. Of the two types of request values R1 and R2, the first request value R1 is a value representing the data amount corresponding to the minimum guaranteed bandwidth.

The second request value R2 is for describing the amount of data exceeding the first request value R1, and is the maximum amount of data that is desired to be transmitted in one grant period (maximum accumulated amount accumulated in the upstream buffer). Is described as the second request value R2.
In the present embodiment, the ONUs 2B and 2C with a net employ a multiple request method, and therefore the two request values R1 and R2 are described in the report R. On the other hand, since the telephone ONU 2A adopts a single request method, only one request value R1 is recorded in the report R.

As shown in FIG. 3B, the gate G of the station side device 1 has at least one set of transmission start time and transmission permission length (time equivalent value) permitted (granted) for one report R. It is written. In the example of FIG. 3B, up to four sets of “Grant # 1 to # 4” can be described.
When two time zones are granted to one gate G, the home side apparatus 2 divides the burst into each time zone and transmits it upstream (one gate and two grant method). When only the time zone is granted, one burst is transmitted in the time zone (1 gate 1 grant system).

The gate G of the station side device 1 includes a data area generally called a flag field (“Number of grants / Flags” in FIG. 3B).
This flag field is an identifier for identifying the type of gate frame transmitted by the station side device 1 by the home side device 2. For example, when the station-side device 1 causes the home-side device 2 to transmit the report R, the station-side device 1 sets a predetermined value other than 0 in this flag field. A flag field forcing the transmission of the report R is referred to as a “force report”.

Returning to FIG. 2, when the receiving unit 104 receives the report R in which the amount of uplink data requested by the home device 2 is written, the report R is sent to the request receiving unit of the dynamic bandwidth allocating unit 107 via the buffer 105. Received at 108 and passed to the calculation unit 109.
The calculation unit 109 calculates the “priority” necessary for the allocation process of the allocation execution unit 110. The priority is a reference when selecting the home device 2 that adopts the second request value R2 for the surplus bandwidth within the grant period that occurs after the assignment with the first request value R1. The calculation unit 109 calculates the priority based on, for example, a predetermined weighting or a minimum guaranteed bandwidth ratio.

The allocation execution unit 110 basically executes “centralized DBA” in which an upstream band is allocated every predetermined grant period (hereinafter also referred to as “DBA cycle”).
In the centralized DBA of this embodiment, for each ONU 2, priority is given to an allocation using a request value (first request value R1) that is equal to or less than a threshold value for guaranteeing low latency, and a surplus bandwidth within the grant period that occurs after this allocation. And a process of assigning request values (second request value R2) exceeding the threshold in order from the network ONU 2B having a higher degree (see FIG. 6).

  But the allocation execution part 110 changes the content of the allocation process by centralized DBA according to the classification | category 2A-2C of the transmission origin of the report R. FIG. In the present embodiment, the contents of processing changed according to the transmission source types 2A to 2C are summarized as follows.

"Grant thinning process"
The grant thinning process is a process performed by the allocation execution unit 110 when the transmission source of the report R is “ONU2A for telephone”.
That is, for the report R whose transmission source is “telephone ONU 2A”, the allocation execution unit 110 employs “thinning processing” in which bandwidth allocation is not performed in each grant period. In addition, when granting the uplink transmission of the telephone ONU 2A, the allocation execution unit 110 grants the uplink bandwidths having different fixed lengths L1 and L2 to the telephone ONU 2A according to the usage status of the telephone.

"Threshold change process"
The threshold value changing process is a process of changing the low delay guarantee threshold value Th (see FIG. 8) according to the use state of the telephone for the report R whose transmission source is “shared ONU2C”.
Specifically, when the shared ONU 2C is in a “non-calling state” (IP phone is not in service), the threshold Th for low delay allocation is set to a lower K1 (for example, the longest control frame length), and “ In the case of “calling state” (during IP telephone service), the threshold value Th for low delay allocation is set to a higher K2 (= K1 + telephone frame length).

The allocation execution unit 110 includes the determined transmission start time and the transmission permission length when the transmission start time and the transmission permission length corresponding to the time equivalent value are determined for each transmission source by the centralized DBA accompanied by the thinning process and the change process. A gate G is generated.
The generated gate G is sent to the buffer 102 by the grant transmitter 112 and sent to the corresponding home device 2 via the transmitter 103. The home device 2 that has received the gate G transmits an upstream optical burst signal at the designated transmission start time and transmission permission length (time).

[General centralized DBA]
As described above, the station side apparatus 1 of the present embodiment basically controls the upstream band for each ONU 2 by “centralized DBA”. First, an outline of a general centralized DBA will be described. FIG. 4 is a sequence diagram showing a general centralized DBA.

  In FIG. 4, it is assumed that time advances from the left side to the right side, and that there are three ONUs 2. Also, in FIG. 4, the grant period, which is a fixed-length operation period by the OLT 1, is represented by the symbol T, the current grant period is represented by the symbol Tc (subscript c is “current”), and the next grant cycle is Tn (subscript). The letter n is represented by “next”).

As shown in FIG. 4, in the centralized DBA, the OLT 1 receives the report R from the ONU 2 first in the current grant period Tc, and assigns the next cycle when each report R is received. Start the calculation.
Then, the OLT 1 generates three gates G describing the calculation results in the current grant period Tc, and transmits these gates G to the respective home-side devices 2 to report the next report R and data (upstream). User data) D bandwidth allocation is notified to each ONU 2.

That is, the OLT 1 comprehensively performs bandwidth allocation of the uplink data D received by the OLT 1 within the next grant period Tn based on the reports R collected from all ONUs 2 during the current grant period Tc, and reports the next time. The transmission time of R and uplink data D is granted to each ONU 2.
According to the centralized DBA, it is possible to perform bandwidth control that preferentially allocates a bandwidth to the ONU 2 having a high priority, which tends to be insufficient in bandwidth, within a predetermined DBA cycle (grant cycle). .

[Problems and solutions for general centralized DBAs]
By the way, if the number of branches is increased as much as possible to accommodate more ONUs 2 in one PON, the cost per user can be reduced.
Therefore, as described above, when a PON system is used for a telephone winding application in which a fixed telephone is replaced with an IP telephone 10A, the number of branches of the PON system can be increased as much as possible to accommodate more telephone ONUs 2A. This is desirable in terms of equipment costs.

However, if the number of ONUs 2 that perform centralized DBA is simply increased, the number of bursts in the upstream direction will increase and overhead will increase, and the bandwidth required for collection of reports D will also increase, so the bandwidth allocated to user traffic will decrease. Service may be degraded.
That is, in the centralized DBA, since the gate G to which the report R and the upstream data D band are individually assigned is normally generated, as shown by the broken lines in FIG. And the upstream data D become separate burst signals separated in time.

For this reason, overhead is required for each burst signal of the report R and the uplink data D, and rounding up of fractions in units of FCW (FEC Code Word) and TQ is also performed, so that the bandwidth utilization efficiency of uplink transmission is improved. Easy to be disturbed.
On the other hand, if the grant period is increased in order to increase the bandwidth efficiency, the delay time is increased by that amount, which may adversely affect communication of an application such as the IP telephone 10A that requires low delay.

  Therefore, in the present embodiment, in order to increase the number of ONUs 2 that can be accommodated in one PON system while maintaining the service level of throughput and delay time, the contents of the allocation process are determined according to the transmission source types 2A to 2C. I decided to change it.

Specifically, in the present embodiment, the network ONU 2B and the shared ONU 2C follow a centralized DBA that grants every time according to the multiple request method, but the telephone ONU 2A does not grant in each grant cycle. A thinning process is performed in which bandwidth allocation based on the report R is performed at a predetermined period longer than the grant period.
As a result, a large amount of bandwidth is allocated to traffic for network use, and the service level can be maintained even when the number of telephone ONUs 2A is increased.

That is, in the telephone ONU 2A, when the IP telephone 10A is in a non-calling state, almost no bandwidth is required and the influence of delay is small, and the link of MPCP (Multi-Point Control Protocol) or OAM (Operation Administration and Maintenance) As long as it is maintained, even if the collection period of the report R is somewhat extended, there is no significant effect.
On the other hand, when the IP telephone 10A call is started, data transmission / reception of a telephone frame starts and low delay is required. However, the bandwidth required for a call is not so large, and even if the grant is omitted up to the maximum delay allowed for the telephone frame, communication of the telephone frame can be performed periodically.

In this embodiment, the shared ONU 2C is also preferentially assigned the request value R1 below the threshold value Th for guaranteeing low delay to secure the low delay service in the same manner as the network ONU 2B. By changing the threshold Th according to the state, the bandwidth efficiency of the entire system is improved.
Specifically, the threshold Th for the non-calling state is set to the minimum amount for control communication, and the telephone usage is added to the threshold Th for the calling state to prevent excessive allocation to the net use in the non-calling state. .

Thus, in the station side apparatus 1 of this embodiment, the allocation execution part 110 changes the content of the dynamic band allocation based on centralized DBA according to the types 2A-2C of ONU2.
Also, the ONUs 2A and 2C that accommodate applications for telephony use a new report different from the conventional one depending on the types 2A and 2C of their own devices so that the OLT 1 can determine whether or not the IP telephone 10A is in a call state. The system and the rules for uplink transmission are adopted.

[Contents of thinning process]
The details of the thinning process will be described below after describing the report method of the telephone ONU 2A that is the target of the thinning process.

(Telephone ONU reporting method, etc.)
The telephone ONU 2A employs a “non-status report method”. This method is a method capable of transmitting the upstream data D stored in its own buffer even when the report R has not yet been declared.
Accordingly, when the OLT 1 grants an excessive amount of data, the telephone ONU 2A can perform upstream transmission even with the upstream data D that is not yet declared as the request value R1.

The telephone ONU 2A transmits the report frame and the upstream data frame simultaneously as the same burst signal. That is, the telephone ONU 2A transmits the report R and the upstream data D to the OLT 1 as a unified burst signal instead of the separated state shown in FIG.
The telephone ONU 2A includes two upstream queues. These upstream queues are divided into 1) for control messages (MPCP, OAM, etc.) and 2) for telephones, and the telephone ONU 2A reports the queuing buffer amount to each queue as it is.

(Details of thinning process)
When the allocation execution unit 110 acquires the report R of the telephone ONU 2A, the allocation execution unit 110 determines whether the status of the telephone ONU 2A is “non-calling state” or “calling state”. Grant “decimation” for each policy described.
The state determination of the telephone ONU 2A can be performed, for example, based on whether or not the request value is included in the telephone upstream queue described in the report R.

"Thinning processing in non-call state"
In the non-calling state, the allocation execution unit 110 executes a thinning process in which a fixed length L1 is allocated and granted only once in a period corresponding to N (N: natural number) DBA cycles.
The fixed length L1 is a time length that allows at least the report R to be transmitted, and a force report is attached to the grant G that notifies the fixed length L1. N is set to a length that does not cause at least MPCP timeout.

Here, since the MPCP timeout time is 1 second, if the DBA cycle is “Tdba”, the value of N may be set to a value that satisfies the inequality of N <1 / Tdba. For example, if Tdba = 600 μs, N <1666.
On the other hand, the allocation execution unit 110 monitors the report value of the telephone ONU 2A. When the report R in the control message queue is non-zero, the allocation execution unit 110 determines whether or not the request can be transmitted with the fixed length L1. .

The allocation execution unit 110 suspends the thinning process and grants a request for the next cycle if it cannot be transmitted with the fixed length L1. In this case, the allocation execution unit 110 basically makes a full answer (grant the request).
However, for example, there is a possibility that the telephone ONU 2A is a malicious ONU 2. Therefore, if the request value of the report R acquired from the telephone ONU 2A exceeds a preset upper limit bandwidth, a zero response (transmission) You may decide to grant a grant amount of zero.

On the other hand, if the request value in the telephone queue of report R is non-zero, allocation execution unit 110 regards the IP telephone 10A as having started a call, and transmits the report R containing the request value (telephone ONU 2A). ) Status to the call state.
The OLT 1 may have a function of monitoring the upstream communication of the telephone ONU 2A. In this case, when the upstream frame from the telephone class queue is buffered, the OLT 1 determines that the ONU 2A that is the transmission source of the upstream frame has shifted to the call state.

"Decimation process in a call state"
When granting to the telephone ONU 2A in a call state, the allocation execution unit 110 allocates and grants a fixed length L2 longer than the fixed length L1. This fixed length L2 is a time length during which at least a report frame and a telephone frame can be transmitted simultaneously.
Further, the allocation execution unit 110 performs a grant thinning process even in a call state. FIG. 5 is a sequence diagram illustrating an example of grant thinning-out processing in a call state.

As shown in FIG. 5, the allocation execution unit 110 allocates bandwidth to the ONU 2A during a period corresponding to M (M: natural number) DBA cycles immediately after the telephone frame has been raised from the telephone ONU 2A. Omitted.
Here, for example, in VoIP (Voice over IP), the data amount of the telephone frame during a call is about 226 bytes per frame (there may be 214 bytes, and there are some fluctuations depending on the presence or absence of a tag). There is a premise that the data is periodically transmitted at about 50 frames / second.

Therefore, assuming a telephone ONU 2A that employs VoIP, if the upstream transmission rate is 10 Gbps, an area in which bandwidth allocation of the telephone frame is omitted every 20 ms after the last reception of the telephone frame. Can be set.
On the other hand, after the (M + 1) th DBA cycle, the allocation execution unit 110 resumes bandwidth allocation for the telephone ONU 2A and prepares for telephone communication by the telephone ONU 2A.
In this way, it is possible to improve the efficiency of the upstream bandwidth while suppressing the bandwidth allocation delay for the telephone frame to approximately 1 DBA cycle.

"Variation of thinning-out processing in a call state"
In the above-described call state thinning-out process, when the telephone ONU 2A transmits a high priority frame (for example, an OAM frame) other than the telephone queue, the OAM frame is processed for M DBA cycles in the worst case. There is a risk of delay.
Since the OAM link timeout period is as long as 5 seconds, the OAM link will not be interrupted by the above delay, but in the case of an OAM application that requires relatively immediacy such as sleep control, it is about 20 ms. It can be a problem even if there is a delay.

Therefore, when the delay time allowed for the telephone frame is “Dmax”, the allocation execution unit 110 sets a period obtained by multiplying the maximum number (= “A”) that is at least Dmax / Tdba or less by the grant period. You may decide to adopt the thinning process of always granting once.
For example, if Dmax = 2000 μs and Tdba = 600 μs, 2000/600 = 3.333... And A = 3.

Therefore, if the grant for the telephone ONU 2A is performed at least once in the period of the 3 × DBA cycle, the maximum delay time of the telephone frame becomes 3 × 600 μs = 1800 μs, and the delay of the telephone frame falls within the allowable range. .
Therefore, in this way, it is possible to improve the efficiency of the upstream band by the thinning process while preventing the influence on other control communication such as OAM.

  If the uplink user frame (telephone frame) from the telephone ONU 2A is not received for a certain period of time, the allocation execution unit 110 regards the ONU 2A as having ended the call and returns the status to the non-call state. This monitoring period is at least the uplink transmission period (= 20 milliseconds) of the telephone frame.

[Centralized DBA of this embodiment]
Hereinafter, after describing the reporting method of the network ONU 2B and the shared ONU 2C, the details of the centralized DBA of this embodiment will be described with reference to FIGS.

(Network ONU reporting method, etc.)
The network ONU 2B employs a “status report method”. This method is a method in which uplink data D stored in its own buffer is not transmitted unless it is reported in a report R.
Therefore, the network ONU 2B transmits only the amount of data reported to the OLT 1 and does not perform upstream transmission until the OLT 1 grants an excessive amount of data until it reports as the report R.

The network ONU 2B includes two upstream queues. These upstream queues are divided into 1) for control messages (MPCP, OAM, etc.) and 2) for the other Internet. The priority order of the upstream queues is set so that the order is for control messages> for the Internet.
The network ONU 2B can include two request values R1 and R2 in the report R. One of them is a buffer amount (first request value R1) that is equal to or less than the threshold value Th for low delay allocation and has a good frame break, and the other is the total buffer amount (second request value R2).

(Shared ONU reporting method, etc.)
The shared ONU 2C also adopts the “status report method”. Therefore, until the shared ONU 2C transmits only the data reported to the OLT 1 and reports it as the report R, the OLT 1 does not perform upstream transmission even if the OLT 1 grants an excessive amount of data.

The shared ONU 2C includes three upstream queues. These upstream queues are divided into 1) for control messages (MPCP, OAM, etc.), 2) for telephone, and 3) for other Internet. The priority order of the upstream queue is set in the order of control message>phone> Internet.
The shared ONU 2C can also include two request values R1, R2 in the report R. One of them is a buffer amount (first request value R1) that is equal to or less than the threshold value Th for low delay allocation and has a good frame break, and the other is the total buffer amount (second request value R2).

(Details of centralized DBA of this embodiment)
FIG. 6 is a flowchart showing the contents of the allocation process by the allocation execution unit 110.
As shown in FIG. 6, the allocation execution unit 110 first sets a report R of the network ONU 2B and the shared ONU 2C (the telephone ONU 2A of the telephone ONU 2A) at the first part of the next grant period Tn in order to keep up with the next allocation calculation. The time required for collecting the report R of the unified burst R + D is allocated (step ST1 in FIG. 6).

Next, the allocation execution unit 110 allocates the remaining uplink data transmission time to each ONU 2 as a transmission time based on the first request value R1 (step ST2 in FIG. 6).
Specifically, the allocation execution unit 110 first allocates the bandwidth of the uplink data D1 for the first request value R1 of all the ONUs 2B and 2C with nets to the grant period Tn. The bandwidth of the upstream data D1 is arranged before the report R bandwidth.

Further, the allocation execution unit 110 is not granted in the current grant cycle Tc, but has fixed lengths L1 and L2 of the telephone ONU 2A that grants in the next grant cycle Tn (report R and uplink data indicated by broken lines in FIG. 6). If there are D singulated bursts R + D), that band is allocated within grant period Tn.
The band of the singulated burst R + D is arranged in front of the band of the upstream data D1 of the ONUs 2B and 2C with a net.

Note that the band of the unified burst R + D of the telephone ONU 2A may be arranged from the end of the next grant period Tn to the last.
If the next grant period Tn is a grant period that is not granted to the telephone ONU 2A by the above-described grant thinning process performed for the telephone ONU 2A, the band allocation of the above-described unified burst R + D is not performed.

Next, the allocation execution unit 110 has the highest priority calculated by the calculation unit 109 (see FIG. 2) when there is still a surplus (“surplus time” in FIG. 6) in the transmission time of the uplink data D. The network-attached ONUs 2B and 2C are selected (step ST3 in FIG. 6).
Then, the allocation execution unit 110 adopts the second request value R2 instead of the first request value R1 for the selected ONU 2B, 2C, and allocates the transmission time of the uplink data D2 corresponding to the second request value R2 ( Step ST3 in FIG. 6).

In this case, if the “surplus time” of the uplink data transmission time is longer than the time corresponding to the second request value R2 of the next allocation candidate ONU 2B, 2C, the uplink data transmission time still remains.
Therefore, the assignment execution unit 110 assigns the transmission time of the uplink data D2 corresponding to the second request value R2 to the ONUs 2B and 2C having the next highest priority in the above step ST3.

Then, the allocation execution unit 110 determines whether all the ONUs 2B and 2C have been allocated for the second request value R2 or the surplus time of the uplink data transmission time is the next allocation candidate ONU 2B and 2C. It is determined whether or not a time corresponding to the second request value R2 is reached (step ST4 in FIG. 6).
The allocation execution unit 110 repeatedly executes the process of step ST3 until the determination result is Yes, and ends the allocation process for the next grant period Tn when the determination result is Yes.

[Schedule within Grant Cycle]
FIG. 7 is a time chart showing an allocation schedule when attention is paid to one DBA cycle.
As described above, in the allocation process (FIG. 6) of the present embodiment, the band of the unified burst R + D of the telephone ONU 2A is arranged after the band of the upstream data D1 of the ONUs 2B and 2C with a net within the grant period Tn. It is like that.

For this reason, when the allocation process (FIG. 6) of this embodiment is executed, as shown in FIG. 7, an allocation schedule is formed in which the following areas X to Z are arranged in order from the front side in one DBA cycle.
Area X: Time area where reports R of ONUs 2B and 2C with nets gather Area Y: Time area where upstream data D1 and D2 of ONUs 2B and 2C with nets gather Area Z: Time area where unified bursts R + D of telephone ONUs 2A gather

  Further, in the present embodiment, when the collection of the report R of the ONUs 2B and 2C with the net is completed, the DBA process (allocation process in FIG. 6) based on the request values R1 and R2 of the report R of the ONU 2B and 2C with the net is executed. However, at this time, as shown in FIG. 7, it is determined whether or not there is a report R (specifically, a report R of the unified burst R + D included in the area Z) from the telephone ONU 2A one cycle before, If there is the report R, the report R of the telephone ONU 2A is simultaneously processed during the DBA processing.

  As described above, in the present embodiment, the reason why the region Z is arranged on the rear side of the DBA cycle is that the report R of the unified burst R + D of the telephone ONU 2A is handled in the same manner as the report R of the ONUs 2B and 2C with a net. As the number of ONUs 2A for telephones increases, the end of the area X is delayed to the rear side, and the reports R of the ONUs 2B and 2C with a network cannot be collected by the next cycle, and the delays of the ONUs 2B and 2C with a network are reduced. This is because the service may not be maintained.

  Further, in this embodiment, the reason why the report R from the telephone ONU 2A of the previous cycle is DBA processed together with the report R of the ONUs 2B and 2C with the net in the next cycle is that the bandwidth of the telephone ONU 2A is fixed length L1, Even if the number of telephone ONUs 2A increases, it can be easily calculated in the next cycle, and if the DBA process is performed together with the report of the ONUs 2B and 2C with the net in the next DBA cycle, This is because the processing delay is only a delay of DBA cycle (= 600 μs), and the QoS for telephone use can be maintained.

[About two types of request values]
FIG. 8 is a conceptual diagram showing the data accumulation state of the upstream buffer of the home side apparatus 2 (ONUs 2B and 2C with a network).
In FIG. 8, f1 to f6 indicate variable-length frames (in this embodiment, Ethernet frames having a variable-length range of 64 to 1522 bytes). In the example of FIG. 8, six variable-length frames f1 to f6 are accumulated in the upstream buffer, and Δ marks indicate the breaks (boundaries) between the frames f1 to f6.

As described above, the ONUs 2B and 2C with a network store the first and second request values R1 and R2 in one report R when requesting the amount of data to be transmitted upstream from the OLT 1, and the report R (FIG. 3 (a) )) Frame is transmitted to OLT1.
Therefore, the OLTs 2B and 2C with a net determine the request values R1 and R2 based on the accumulation state of the variable length frames f1 to f6 in the upstream buffer.

As shown in FIG. 8, the first request value R1 has a data amount that is equal to or smaller than a preset threshold Th and that is closest to the first threshold R1.
The threshold value Th indicates that the data amount for the minimum guaranteed bandwidth is the data amount equal to or smaller than this value. The shared ONU 2C according to the present embodiment changes the threshold Th according to its own status. A method for changing the threshold Th will be described later.

The second request value R2 corresponds to a break of the variable-length frame f6 that is equal to or less than the maximum data amount (total buffer amount in the example) that the ONUs 2B, 2C with a net wants to transmit in one grant cycle. The amount of data.
As described above, the first and second request values R1 and R2 both have a data amount that matches the delimiter (Δ mark in FIG. 6) of the variable length frames f2 and f6.

Therefore, even if the OLT 1 (allocation execution unit 110) directly adopts the request values R1 and R2 of the ONUs 2B and 2C with the net and performs the dynamic bandwidth allocation for generating the gate G, the variable length frames f1 to f6 that cannot be divided are obtained. These frames f1 to f6 can be efficiently arranged within the grant period T as a set unit.
For this reason, the occurrence of dead time in the form that the variable length frames f1 to f6 cannot be aligned and the frames f1 to f6 do not enter the grant period T is suppressed.

  However, at the final stage of bandwidth allocation for one grant period T, the variable-length frames f1 to f6 cannot be aligned, and a certain amount of wasted time is generated. There is only one home side device 2-4 that performs the allocation, and the length of the dead time can be suppressed to the maximum equivalent to the maximum size of the variable length frames f1 to f6 (in the case of an Ethernet frame, 1522 bytes). .

[Contents of threshold change processing]
Next, the threshold value changing process performed by the allocation execution unit 110 of the OLT 1 and each shared ONU 2C will be described.
The shared ONU 2C can determine its own status relating to the call of the IP telephone 10A based on a change in the state of the telephone queue.

Specifically, when the status is “non-calling state” and the status is “non-calling state”, the shared ONU 2C transitions the status to “calling state” when uplink data enters the telephone queue.
In addition, when the status is “calling state”, the shared ONU 2C sets the status to “non-calling” if the period when the uplink data amount in the telephone queue is zero continues for the uplink transmission period (= about 20 milliseconds) of the telephone frame. Transition to state.

The shared ONU 2C can change the threshold Th for low-latency allocation according to its own status regarding the call of the IP telephone 10A.
Specifically, when the status of the shared ONU 2C is “non-call state”, the low delay allocation threshold Th is set to K1 (= the longest control frame length). Further, when the status of the shared ONU 2C is “calling state”, the low delay allocation threshold Th is set to K2 (= K1 + telephone frame length).

The allocation execution unit 110 sets the status of the subordinate shared ONU 2C to either “non-calling state” or “calling state”, and for the low delay of the shared ONU 2C depending on which status is The threshold value Th is changed.
The allocation execution unit 110 sets the threshold Th for low delay allocation to K1 (= longest control frame length) for the shared ONU 2C whose status is “non-calling state”. If the telephone queue of the shared ONU 2C in the “non-call state” is reported as non-zero, the allocation execution unit 110 transitions the status of the shared ONU 2C to the “call state”.

  The allocation execution unit 110 sets the threshold Th for low delay allocation to K2 (= K1 + telephone frame length) for the shared ONU 2C whose status is “calling state”. The allocation execution unit 110 transitions the status of the shared ONU 2C to the “non-call state” when a report indicating that the telephone queue of the shared ONU 2C in the “calling state” is zero continues for a certain period of time. This monitoring period is at least longer than the uplink transmission cycle (= about 20 milliseconds) of the telephone frame.

The allocation execution unit 110 of the present embodiment also performs allocation processing by the multiple request method, which is equivalent to the network ONU 2B, for the shared ONU 2C. For this reason, the uplink data of the first request value R1 such as a telephone frame requested by the shared ONU 2C is granted every time in each DBA cycle.
For this reason, the shared ONU2C telephone frame in a call state is assigned a band in each grant period, and QoS for telephone use for the shared ONU2C is secured.

In the present embodiment, when the status of the shared ONU 2C is in a non-calling state, the threshold value Th for low delay allocation is set smaller than that in the calling state.
For this reason, when the status of the shared ONU 2C is in a non-calling state, there is an effect that the uplink bandwidth used by other ONUs 2 can be increased.

[Number of telephone ONUs accommodated]
By the way, in the PON system that contains both the telephone ONU 2A and the ONUs 2B and 2C with the net, the centralized DBA that grants in each DBA cycle (hereinafter referred to as the ONU 2B and 2C with the net) also for the report R of the telephone ONU 2A. , “Concentrated DBA of Comparative Example”).
That is, in the centralized DBA of the comparative example, the allocation execution unit 11 refers to the request values R1 and R2 of the reports R collected from all the ONUs 2 without distinguishing the usages of the ONUs 2 for telephones and networks. Centralized DBA to be performed.

FIG. 9 is a diagram illustrating time distribution of DBA cycles in the centralized DBA of the comparative example.
Here, when the number of telephone ONUs 2A is “U” and the number of ONUs 2B and 2C with a network is “V”, the centralized DBA of the comparative example has (U + V) report R as shown in FIG. Arranged in front of the DBA cycle. This time domain is the time required to receive the burst for the report R from all the ONUs 2A to 2C that can be connected to the OLT 1, and the remaining time can receive the burst for the upstream data D of each ONU 2A to 2C. It will be time to do.

In the centralized DBA of the comparative example, when the following “condition A” is satisfied, service levels such as throughput and delay time cannot be maintained for some of the network ONUs 2B and 2C.
“Condition A”: the remaining time (time T2 in FIG. 9) obtained by subtracting the time (time T1 in FIG. 9) after assigning the report R of all ONUs 2A to 2C that can be connected to the OLT 1 from the DBA cycle is the ONU 2B with net , 2C must be less than the total bandwidth to be serviced.

The reason is that, in the centralized DBA of the comparative example that grants to all ONUs 2A to 2C every time, if the above condition A is satisfied, the ONUs 2B and 2C with a net can be used for the desired net even if the remaining time T2 is fully used. This is because the service cannot be provided.
The above condition A can be expressed by an inequality as follows.
_{U ≧ (Tdba-Σ V BRi} -Σ V (Ki · BWi · Tdba + OHi)) / BR ... (1)
In addition, the definition of the parameter in the said inequality (1) is as enumerating below.

U: Number of ONUs for telephone V: Number of ONUs with network BWi: Average data bandwidth (Mbps) to be serviced by ONUi with network
Tdba: DBA cycle (μs)
Ki: Conversion weight from the bit domain to the time domain (a constant determined by the uplink rate).
OHi: ONUi optical overhead (μs)
BRi: burst size for report of ONUi with net (μs)
BR: Burst size for report of telephone ONU (μs)

Thus, in the centralized DBA of the comparative example, when the number U of telephone ONUs 2A increases to the extent that the inequality (1) is satisfied, the network ONUs 2B and 2C cannot maintain a desired service level.
In this regard, in the present embodiment, the allocation execution unit 110 executes a thinning process for the telephone ONU 2A that does not allocate bandwidth in each DBA cycle, so the report R of the telephone ONU 2A is included in time T1 in FIG. Even if the number U of telephone ONUs 2A is increased to the extent that the above inequality (1) is satisfied, the network ONUs 2B and 2C can appropriately maintain a desired service level.

In other words, the centralized DBA of this embodiment is effective when it is desired to increase the number U of telephone ONUs 2A to the extent that the above inequality (1) is satisfied in the centralized DBA of the comparative example.
By performing the above-described thinning process on the telephone ONU 2A, the number of telephone ONUs 2A can be increased while maintaining the service level such as the throughput and delay time of the network ONUs 2B and 2C. Accordingly, the total number (U + V) of ONUs 2 accommodated in the PON system can be increased while maintaining fairness.

On the other hand, in the centralized DBA of the comparative example, even when the following “condition B” is satisfied, service levels such as throughput and delay time cannot be maintained for some of the network ONUs 2B and 2C.
“Condition B”: The remaining time (time T2 in FIG. 9) obtained by subtracting the time (time T1 in FIG. 9) after assigning the report R of all ONUs 2A to 2C that can be connected to OLT 1 from the DBA cycle is connected to OLT 1. It must be less than the RTT (Round Trip Time) of the furthest ONU 2B, 2C with a net.

The reason for this is that in the centralized DBA of the comparative example that grants to all the ONUs 2A to 2C every time, when the above condition B is satisfied, the area where the reports R of the ONUs 2B and 2C with nets are arranged (time T1 in FIG. 9) is the DBA cycle. This is because the MPCP handshake cannot be performed within one cycle with the farthest ONUs 2B and 2C with the net.
The above condition B can be expressed by an inequality as follows.
U ≧ (Tdba−RTTmax−ΣBRi) / BR (2)
In addition, the definition of the parameter in the said inequality (2) is as enumerating below.

U: Number of telephone ONUs Tdba: DBA cycle (μs)
BRi: burst size for report of ONUi with net (μs)
BR: Burst size for report of telephone ONU (μs)
RTTmax: The largest RTT among ONUs with a net

Thus, in the centralized DBA of the comparative example, when the number U of telephone ONUs 2A increases to the extent that the inequality (2) is satisfied, the network ONUs 2B and 2C cannot maintain a desired service level.
In this regard, in the present embodiment, the allocation execution unit 110 executes a thinning process for the telephone ONU 2A that does not allocate bandwidth in each DBA cycle, so the report R of the telephone ONU 2A is included in time T1 in FIG. Even if the number of telephone ONUs 2A is increased to the extent that the inequality (2) is satisfied, the network ONUs 2B and 2C can appropriately maintain a desired service level.

In other words, the centralized DBA of this embodiment is effective when it is desired to increase the number U of telephone ONUs 2A to the extent that the above inequality (2) is satisfied in the centralized DBA of the comparative example.
By performing the above-described thinning process on the telephone ONU 2A, the number of telephone ONUs 2A can be increased while maintaining the service level such as the throughput and delay time of the network ONUs 2B and 2C. Accordingly, the total number (U + V) of ONUs 2 accommodated in the PON system can be increased while maintaining fairness.

[Other variations]
The embodiments disclosed herein are illustrative of the present invention and are not limiting. The scope of the present invention is shown not by the above-described embodiment but by the scope of claims for patent, and includes all modifications within the scope and meaning equivalent to the scope of claims for patent.

For example, in the above-described embodiment, “telephone ONU 2A” is exemplified as the ONU 2 specialized for the low-delay service by assuming the IP phone as the low-delay service provided by the ONU 2.
Of course, examples of the low-delay service include not only IP telephones but also applications such as low-delay communication for power control. Assuming these low-delay communications, the ONU 2A dedicated to the low-delay service is not necessarily limited to telephone use.

In the above-described embodiment, the storage unit 113 (FIG. 3) of the station-side device 1 stores the types 2A to 2C and the minimum guaranteed bandwidth of all the home-side devices 2, but reports from the home-side device 2 The types 2A to 2C and the minimum guaranteed bandwidth may be acquired by R, and the storage contents in the storage unit 113 may be updated as needed.
In this way, it is not necessary for the communication carrier to set such information in the storage unit 113 of the station side device 1.

1: Station side equipment (OLT)
2: Home unit (ONU)
2A: Telephone ONU (first ONU)
2B: ONU for the network (second ONU)
2C: Shared ONU (second ONU)
5: Optical fiber (main line) 6: Optical coupler 7: Optical fiber (branch line)
8: PON line 9: Host network 10A: IP telephone (communication equipment) 10B: PC (communication equipment)
101: receiving unit 102: buffer 103: transmitting unit 104: receiving unit 105: buffer 106: transmitting unit 107: dynamic bandwidth allocating unit 108: request receiving unit 109: calculating unit 110: allocation executing unit 112: grant transmitting unit 113: Storage

Claims (12)

Based on reports collected from a plurality of home-side devices, a bandwidth of uplink data received in a future grant period is allocated to each home-side device, and the next-time report and the uplink data transmission time are assigned to the home-side device. A dynamic bandwidth allocation method that grants to
In the case where the first ONU and the second ONU defined below are included as the types of the home-side devices, the bandwidth allocation based on the report whose transmission source is the first ONU is performed at a predetermined cycle longer than the grant cycle. Bandwidth allocation method.
First ONU: ONU specialized for low-latency service applications
2nd ONU: ONU that can be used for Internet browsing   Bandwidth allocation based on the report whose transmission source is the second ONU is executed for each grant period based on the first request value for low delay included in the report and the second request value exceeding the first request value. The dynamic bandwidth allocation method according to claim 1, wherein the allocation process is a multiple request method. The predetermined period can be changed according to the service state of the first ONU,
3. The predetermined period when the service state of the first ONU is not in service is set to a time length necessary for the first ONU to maintain an MPCP link with a station side device. The dynamic bandwidth allocation method described in 1. The predetermined period can be changed according to the service state of the first ONU,
The dynamic period according to any one of claims 1 to 3, wherein the predetermined period when the service state of the first ONU is in service is set to a transmission period of an uplink frame used for the low-delay service. Bandwidth allocation method. The predetermined period can be changed according to the service state of the first ONU,
The predetermined period when the service state of the first ONU is in service is set to a time length equal to or less than a maximum delay time allowed for the low-delay service. The dynamic bandwidth allocation method described in 1. The length of time for granting to the first ONU can be changed according to the service state of the first ONU.
The dynamic time according to any one of claims 1 to 5, wherein the time length when the service state of the first ONU is not in service is set to a fixed length at least capable of transmitting the report. Bandwidth allocation method. The length of time for granting to the first ONU can be changed according to the service state of the first ONU.
The time length when the service state of the first ONU is in service is set to a fixed length that can transmit at least the report and an uplink frame used for the low-delay service. The dynamic bandwidth allocation method according to any one of the preceding claims. 8. The dynamic bandwidth according to claim 1, wherein scheduling is performed from the front side in the order of region X, region Y, and region Z defined below in the grant period for performing bandwidth allocation of the first ONU. Assignment method.
Area X: Time area for assigning second ONU report Area Y: Time area for assigning second ONU uplink data Area Z: Time area for assigning first ONU report and unified burst of uplink data   9. The operation according to claim 8, wherein the bandwidth allocation based on the report included in the region X of the current grant period is processed simultaneously with the bandwidth allocation based on the report included in the region X of the next grant period. Bandwidth allocation method. Based on reports collected from a plurality of home-side devices, a bandwidth of uplink data received in a future grant period is allocated to each home-side device, and the next-time report and the uplink data transmission time are assigned to the home-side device. A computer program that causes a computer to function as an allocation execution unit that grants to
In the case where the first ONU and the second ONU defined below are included as the types of the home-side devices, thinning is performed in which a bandwidth allocation based on the report whose transmission source is the first ONU is performed at a predetermined cycle longer than the grant cycle. A computer program that includes processing.
First ONU: ONU specialized for low-latency service applications
2nd ONU: ONU that can be used for Internet browsing Based on reports collected from a plurality of home-side devices, a bandwidth of uplink data received in a future grant period is allocated to each home-side device, and the next-time report and the uplink data transmission time are assigned to the home-side device. An allocation execution unit that grants to
A station side device comprising: a first ONU and a receiving unit that receives the report from the second ONU defined below;
The remaining time for transmission of the uplink data remaining in the grant period is the total of the service bandwidth for the second ONU when the report of the first ONU is granted each time as in the second ONU. Under smaller conditions,
The said allocation execution part is a station | side apparatus which performs the thinning | decimation process which performs the band allocation based on the said report whose transmission origin is said 1st ONU in the predetermined period longer than the said grant period.
First ONU: ONU specialized for low-latency service applications
2nd ONU: ONU that can be used for Internet browsing A PON system including a station side device and a plurality of home side devices connected to the station side device by a PON line,
The station side device allocates a bandwidth of uplink data received in a future grant period to each home device based on reports collected from a plurality of home devices, and transmits the report and the uplink data for the next time. An allocation execution unit that grants time to each of the home-side devices;
The first ONU and the second ONU defined below are included as the types of the home side devices,
The RTT having the longest remaining time for transmission of the uplink data remaining in the grant period when the grant is made each time for the report of the first ONU as in the second ONU. Under conditions smaller than (Round Trip Time)
The allocation execution unit is a PON system that executes a thinning process for performing bandwidth allocation based on the report whose transmission source is the first ONU in a predetermined cycle longer than the grant cycle.
First ONU: ONU specialized for low-latency service applications
2nd ONU: ONU that can be used for Internet browsing

Images