CN103560978B - The method and apparatus of optical access network Bandwidth Dynamic Allocation - Google Patents

Abstract

The present invention relates to a method and device for dynamically allocating bandwidth in an optical access network. It overcomes the shortcomings of the existing TDM‑PON dynamic bandwidth allocating mechanism, such as poor real-time performance and being limited to a single OLT, and proposes an optical access oriented A centralized flexible bandwidth dynamic allocation method and device for global network access. The device of the present invention is composed of a control layer and managed equipment, wherein the control layer includes a communication agent module, a bandwidth resource analysis module, and a strategy decision module. The method for dynamically allocating bandwidth of the present invention monitors network traffic in a loop, and if there is a busy hotspot in the network, a dynamic bandwidth allocating mechanism is executed. The bandwidth dynamic allocation mechanism includes a bandwidth dynamic optimization mechanism and a bandwidth dynamic restoration mechanism. Compared with the prior art, the present invention performs cyclic detection and adopts joint dynamic bandwidth scheduling technical measures of the entire network, so that the bandwidth resource management of the optical access network is intelligentized, the bandwidth utilization rate is greatly improved, and the operation cost of the access network is saved. , improving the user experience of the access network.

Description

Method and device for dynamic bandwidth allocation of optical access network

technical field

The present invention relates to a method and device for dynamic bandwidth allocation in an optical access network, in particular to a centralized flexible bandwidth dynamic allocation method and device for combining multiple optical line terminals in a time-division multiplexed passive optical network.

Background technique

With the continuous advancement of "triple play" and the rapid development of FTTx, flexible allocation of ultra-large bandwidth is an inevitable trend in the development of optical access networks. At the same time, higher requirements are put forward for intelligent management of optical access networks. How to improve resource utilization through intelligent control of the access network is of great significance to the full popularization of optical access networks.

The optical access network is a multi-point-to-point topology in the upstream direction. In the time-division multiplexing passive optical network (TDM-PON), the upstream channel adopts time-division multiplexing (TDM) to share bandwidth resources. TDM-PON The system uses a dynamic bandwidth allocation mechanism (DBA) to improve the system uplink bandwidth utilization and ensure fairness and quality of service (QoS). Bandwidth is allocated flexibly and efficiently. The ONU sends a bandwidth request to the OLT, indicating the amount of data to be sent, and the OLT processes the bandwidth requests from all ONUs, and sends authorization messages to each ONU according to these requests and pre-established policies. The ONU receives the authorization information from the OLT and sends data according to the window size. In addition, in order to ensure the fairness among ONUs and prevent the ONU with too much data from monopolizing the entire bandwidth for a long time, the OLT limits the size of the ONU transmission window.

The dynamic bandwidth allocation mechanism realizes the flexible configuration of the local bandwidth of the optical access network according to the bandwidth requirements of the terminal and the pre-established strategy. However, this dynamic bandwidth allocation mechanism inside the OLT cannot meet the flexible configuration requirements of the entire optical access network bandwidth. On the one hand, the traditional dynamic bandwidth allocation mechanism relies on manual pre-established parameter strategies, which can only achieve a certain range of flexibility, poor real-time performance, and cannot dynamically meet the terminal bandwidth requirements; on the other hand, the traditional dynamic bandwidth allocation mechanism is limited to Within a single OLT, there is a lack of flexible configuration of bandwidth between multiple OLTs. When the bandwidth requirements of multiple ONUs within the same PON increase, the traditional dynamic bandwidth allocation mechanism cannot meet the high bandwidth requirements due to the limitation of the total upstream bandwidth of the OLT.

Contents of the invention

The technical problem to be solved by the present invention is to overcome the disadvantages of the existing TDM-PON dynamic bandwidth allocation mechanism, such as poor real-time performance and being limited to a single OLT, and propose a centralized flexible bandwidth dynamic system oriented to the overall situation of the optical access network. Dispensing method and apparatus.

The device for dynamically allocating bandwidth in the optical access network of the present invention consists of a control layer and a managed device (including the OLT and the metropolitan convergence layer equipment SR/BRAS connected thereto), wherein the control layer includes a communication agent module, a bandwidth resource analysis Module, policy decision-making module, bandwidth resource analysis module collects real-time traffic information of the underlying equipment and analyzes it, and sends the analyzed traffic conditions of the entire network to the policy decision-making module. Formulate a bandwidth optimization strategy. The communication agent module is responsible for protocol analysis and communication with the managed device. The information flow between the control layer and the managed device is bidirectional. The managed device reports traffic status information to the The management device issues bandwidth optimization policies.

The method for dynamic allocation of optical access network bandwidth of the present invention is as follows:

Step 1: Initialization. Allocate the initial bandwidth of all devices, and the initial bandwidth value is the guarantee of the fairness of bandwidth allocation.

Step 2: Monitor network traffic. If there is a busy hotspot in the network, a dynamic bandwidth allocation mechanism is implemented. The bandwidth dynamic allocation mechanism includes a bandwidth dynamic optimization mechanism and a bandwidth dynamic recovery mechanism.

The bandwidth dynamic optimization mechanism involves steps 3 and 4. There are three options: only include step 3; only include step 4; include both step 3 and step 4.

Step 3: Execute the OLT internal bandwidth optimization strategy. Optimize bandwidth resource allocation among ONUs under the same OLT.

Step 4: Execute the joint bandwidth optimization strategy of multiple OLTs. Optimize the allocation of bandwidth resources between OLTs under the same MAN access switch.

If the bandwidth optimization mechanism cannot meet the bandwidth demand of the busy hotspot, the bandwidth recovery mechanism will be implemented. The bandwidth recovery mechanism involves steps 5 and 6. Corresponding to the foregoing bandwidth dynamic optimization mechanism, there are three solutions: only step 5 is included; step 6 is only included; step 5 and step 6 are included.

Step 5: Implement OLT internal bandwidth recovery strategy. By adjusting the internal bandwidth resource allocation of the OLT, the bandwidth allocation value of the busy hotspot is forcibly restored within the initial bandwidth value to ensure the fairness of bandwidth allocation.

Step 6: Implement multi-OLT joint bandwidth recovery strategy. By adjusting the bandwidth resource allocation between multiple OLTs, the bandwidth allocation value of the busy hotspot is forcibly recovered within the initial bandwidth value to ensure the fairness of bandwidth allocation.

Step 7: Update the allocated bandwidth of the device according to the bandwidth values determined in steps 3 to 6, and then return to step 2.

Compared with the prior art, adopting the method and device of the present invention introduces joint bandwidth adjustment of multiple OLTs, and performs loop detection and adjustment. Due to the adoption of the joint dynamic bandwidth scheduling technology measures of the entire network, the bandwidth resource management of the optical access network is intelligentized, the bandwidth utilization rate is greatly improved, the operation cost of the access network is saved, and the user experience of the access network is improved.

Description of drawings

Figure 1 is a traditional optical access network network management architecture.

Fig. 2 is a device for dynamically allocating bandwidth in an optical access network.

Fig. 3 is a device for dynamically allocating bandwidth with equipment as the direct management object in the optical access network.

Fig. 4 is a device for dynamically allocating bandwidth with a distributed network management server as the direct management object in the optical access network.

Fig. 5 is a method for dynamically allocating bandwidth to the optical access network and the converging layer equipment of the metropolitan area network connected thereto.

FIG. 6 is a method for dynamically allocating bandwidth within the OLT.

Fig. 7 is a method for dynamically allocating bandwidth jointly by multiple OLTs.

detailed description

Figure 1 shows the network management architecture of traditional optical access network, including OLT, multiple ONUs connected to each OLT, metropolitan area network aggregation layer equipment (SR and BRAS) connected to OLT, and network management system. The traditional centralized network management has a single function and can only realize one-way information flow, that is, collect basic information such as network topology, and does not have the function of dynamically configuring network status. Moreover, the traditional network management architecture is distributed, and the metropolitan area network and the access network are independent. This leads to the limitations of traditional network management, which cannot dynamically allocate network resources globally for the entire access network.

Figure 2 is a device for dynamic allocation of bandwidth in the optical access network, including the control layer and managed equipment, where the managed equipment includes OLT and the metropolitan convergence layer equipment (SR and BRAS) directly connected to the OLT, and the control layer includes Communication agent module, bandwidth resource analysis module, strategy decision module. The control layer is integrated in the server and connected with the management interface of the managed device. A variety of control protocols can be selected between the control layer and the managed device, such as SNMP protocol, OpenFlow protocol, etc. The bandwidth resource analysis module collects and analyzes the real-time traffic information of the underlying devices, and sends the analyzed traffic conditions of the entire network to the policy decision-making module. The policy decision-making module formulates bandwidth optimization strategies according to the traffic distribution of the entire network and the bandwidth allocation of the entire network. The communication agent module is responsible for communication with managed devices, protocol analysis, etc. The information flow between the control layer and the managed device is bidirectional: the managed device reports traffic status information to the control layer, and the control layer issues bandwidth optimization policies to the managed device. The three modules of the control layer can be integrated into a single controller, or can be separated to form a control cluster. As other schemes of the present invention embodying the same idea, the communication agent module can also be integrated with existing managed devices, and can also be integrated with existing distributed network management systems, which can adapt to different networking environments and shield the underlying hardware information. Provide an abstract interface for the upper-layer control unit to realize unified control and management of complex heterogeneous access networks.

The device control layer of the present invention collects the real-time status information of the whole network, and then sends the adjustment strategy to the corresponding equipment. Through the polling mechanism, the dynamic allocation of network resources is realized, and the resource utilization rate is improved. It solves the limitations of the traditional network management architecture and the characteristics of poor real-time performance.

The management object of the device for dynamic allocation of bandwidth in the optical access network proposed by the present invention can be equipment or a network management server. The former has a simple network structure and high reliability.

Fig. 3 is a device for realizing dynamic allocation of bandwidth with equipment as direct management object in the optical access network. In this embodiment, the SNMP protocol is used between the communication agent module and the managed device. At this time, the access network device has a management interface that can recognize the SNMP protocol. For access network devices that support other protocols (such as the OpenFlow protocol), a communication proxy module using other protocols (such as the OpenFlow protocol) is required.

One way to implement other modules of the control layer (bandwidth resource analysis module, policy decision-making module) is to use embedded software; another method is to use the network operating system (Network OS) and implement it by developing application software on the operating system. The network operating system provides unified south and north interfaces, the south interface is connected to the communication agent module, and communicates with the device through the communication agent module; the network operating system provides a unified north interface for the application layer, and the unified north interface can realize the flexibility of the application layer Expansibility, simplifying the development of network functions, this open architecture improves the scalability and reliability of the device of the present invention.

Figure 4 is a device for realizing bandwidth dynamic allocation in an optical access network with a distributed network management server as the direct management object. This implementation method draws on the existing traditional network management architecture. The communication agent module communicates with the distributed network management server of the access network and the distributed network management server of the metropolitan area network in two directions, and directly reads and modifies the network status database on the server side. The operating system reads the network status database on the server side, and through analysis and decision-making, issues bandwidth optimization configuration, and then updates the network configuration of the managed device through the network management server to realize dynamic bandwidth adjustment. The communication interface between the communication agent module and the network management server is the same as the communication protocol used by the communication interface between the network management client and the network management server.

Each step of the method of the present invention is completed through the cooperation of the above-mentioned communication agent module, bandwidth resource analysis module, and policy decision-making module. The main steps (steps 3 to 6) of the method embodiment are realized by the policy decision-making module. The monitoring traffic in the method embodiment ( Step 2) is implemented through the bandwidth resource analysis module, and device bandwidth configuration (steps 1 and 7) is implemented through the communication agent module.

Fig. 5 is a method for dynamically allocating bandwidth of optical access network equipment and connected metropolitan area aggregation equipment, and the working process includes 7 steps.

Step 1: Set initial bandwidth for all devices. Here, "initial bandwidth" is defined: it is the user's contracted bandwidth, or the operator's minimum guaranteed bandwidth to the user.

Step 2: The control layer periodically polls to detect the traffic distribution. If there is a busy hotspot in the network, the bandwidth dynamic allocation mechanism is triggered, otherwise, the polling continues. Here, a "busy threshold" is preset, and when the ratio of the actual traffic of the ONU (or OLT) to its allocated bandwidth is greater than the threshold, it is judged that the ONU (or OLT) is in a busy state.

The bandwidth dynamic optimization mechanism is divided into OLT internal bandwidth optimization strategy (step 3) and multiple OLT joint bandwidth optimization strategy (step 4). There are three options: only include step 3; only include step 4; include both step 3 and step 4.

Step 3: The OLT internal bandwidth optimization strategy only optimizes the dynamic bandwidth allocation parameters of each ONU in a single OLT, reduces the allocated bandwidth of idle ONUs, and improves the allocated bandwidth of busy ONUs;

Step 4: The multi-OLT joint bandwidth optimization strategy optimizes the bandwidth allocation between multiple OLTs, that is, the upstream bandwidth allocation of OLTs, reduces the allocated bandwidth of idle OLTs, and increases the allocated bandwidth of busy OLTs.

When the bandwidth optimization mechanism executes the "step 3 and step 4" scheme, a best embodiment is to execute step 3 first, if the sum of the free bandwidth available to each ONU under the same OLT is greater than the bandwidth requirement, then pass If the operation in step 3 has satisfied the bandwidth requirement, step 4 will not be executed; if the sum of available free bandwidth of each ONU belonging to the same OLT is less than the bandwidth requirement, step 4 needs to be executed.

If the bandwidth optimization mechanism cannot meet the bandwidth demand of the busy hotspot, which indicates that the idle bandwidth has been exhausted, the bandwidth recovery mechanism is implemented to ensure the fairness of bandwidth allocation. For example, there are user A and user B. When A's bandwidth requirement is greater than its initial bandwidth and B's bandwidth requirement is smaller than its initial bandwidth, the above bandwidth optimization process will increase A's allocated bandwidth and decrease B's allocated bandwidth. But as time changes, when B's bandwidth demand becomes larger, if there is no other free bandwidth, B can only enjoy services lower than the initial bandwidth, which is unfair, so it is necessary to make B's actual bandwidth within the initial bandwidth range Complement within.

Define "required bandwidth", that is, the part where the bandwidth demand of the user's actual traffic exceeds the allocated bandwidth.

Define "remaining bandwidth", that is, in the above example, the part of the bandwidth obtained by user A through optimization that is higher than the initial value. Surplus bandwidth is achieved by reducing the bandwidth of other users with low demand. When the user's demand is very large, the allocated bandwidth should at least meet their respective initial bandwidth to ensure fairness.

The bandwidth recovery mechanism is divided into OLT internal bandwidth recovery strategy (step 5) and multi-OLT joint bandwidth recovery strategy (step 6). Corresponding to the aforementioned bandwidth optimization mechanism, there are three schemes: only step 5 is included; only step 6 is included Step; includes steps 5 and 6.

Step 5: The OLT internal bandwidth recovery strategy adjusts the allocated bandwidth of each ONU in a single OLT according to the initial bandwidth value, and restores the bandwidth value of the busy hot spot to the initial bandwidth value;

Step 6: Multi-OLT joint bandwidth recovery strategy adjusts the allocated bandwidth of each OLT according to the initial bandwidth value, and restores the bandwidth value of the busy hotspot to the initial bandwidth value. Finally, the control layer sends a policy message to update the allocated bandwidth of the device.

When the bandwidth recovery mechanism executes the "steps 5 and 6" scheme, a preferred embodiment is to execute step 5 first, and skip step 6 if the bandwidth requirement can be met in step 5. If step 5 is not satisfied, go to step 6.

When the bandwidth optimization mechanism in this method includes steps 3 and 4, the corresponding bandwidth recovery mechanism should include steps 5 and 6. Only through step 5 may not be able to restore to the initial value. For example, there are three ONUs 1.1, 1.2, and 1.3 in 1#OLT, and three ONUs 2.1, 2.2, and 2.3 in 2#OLT. During the time period t0~t1, ONU2.1~2.3 has a large bandwidth requirement, while ONU1.1~ 1.3 Bandwidth requirements are relatively small. At this time, the multi-OLT joint bandwidth optimization strategy will be implemented, resulting in the ONU bandwidth in 1#OLT being lower than the initial bandwidth. Then, if the bandwidth requirements of ONU1.1~1.3 also increase during the time period t1~t2, obviously the bandwidth optimization strategy does not work, and the OLT internal bandwidth restoration strategy cannot meet the requirements, because none of the three ONUs under 1#OLT The remaining bandwidth can only meet the requirements through the multi-OLT joint bandwidth recovery strategy.

FIG. 6 shows a method for dynamically allocating bandwidth within an OLT in an optical access network. The contents of Steps 3.1~3.3 and Steps 5.1~5.2 are explained in detail below.

When a certain ONU in the same OLT is found to be in a busy state during the detection, then perform step 3.1: find all idle ONUs under the same OLT. Here, an "idle threshold" is preset, and when the ratio of the actual traffic of the ONU to its allocated bandwidth is smaller than the threshold, it is judged that the ONU has idle bandwidth. If there is no free ONU, bandwidth adjustment requirements cannot be met, and other bandwidth allocation strategies continue to be implemented;

When there is an idle ONU, perform step 3.2: According to the required bandwidth of the above "busy ONU" and the idle bandwidth of the idle ONU, determine the new allocated bandwidth of each ONU through the bandwidth scheduling algorithm until the required bandwidth of the busy ONU is met. There are many kinds of bandwidth scheduling algorithms: For example, the bandwidth is simply adjusted according to the order of idle bandwidth from large to small; Sequentially adjust (for example, from large to small, small to large, or random) the idle bandwidth of each ONU; if the bandwidth adjustment rate is less than 1, the bandwidth of each idle ONU is proportionally controlled according to the bandwidth adjustment rate, and each ONU The adjusted bandwidth of is equal to the bandwidth adjustment rate multiplied by its idle bandwidth. The new allocated bandwidth of each ONU is obtained by considering the adjusted value on the basis of the original bandwidth.

Step 3.3, judge the status of the busy ONU after optimization. If the busy problem is solved, go to step 7; otherwise, it indicates that the bandwidth adjustment requirement cannot be met, and other bandwidth allocation strategies need to be continued. Note that the processing of step 3.2 completes the bandwidth optimization strategy but does not necessarily meet the bandwidth requirements. For example, the traffic/allocated bandwidth of an (assuming the busy threshold is 0.7), after optimization, 2M bandwidth is dispatched to the ONU, and the new ratio is 9M/12M=0.75, which is still > the busy threshold. At this point, other bandwidth allocation strategies need to be implemented.

When the adjusted ONU changes from the idle state to the busy state, its bandwidth allocation may be smaller than the initial bandwidth, thus affecting the fairness of bandwidth allocation. In order to ensure the fairness of bandwidth allocation, the OLT internal bandwidth recovery strategy is introduced, including steps 5.1 to 5.2.

When the OLT internal bandwidth allocation strategy fails (that is, after the OLT internal bandwidth optimization strategy is executed, the idle bandwidth of other ONUs is exhausted and the actual bandwidth X of the busy ONU cannot be fully met), perform step 5.1: determine the bandwidth allocation of the busy ONU Whether the amount is less than the initial bandwidth, if yes, go to step 5.2, otherwise go to step 7.

Step 5.2: First, determine the bandwidth required by the busy ONU whose real-time bandwidth is smaller than the initial bandwidth to achieve recovery within the initial value, then count other ONUs whose real-time bandwidth is greater than the initial bandwidth and calculate their remaining bandwidth, adjust the bandwidth according to the bandwidth scheduling algorithm, and update Corresponding to the allocated bandwidth of the ONU, the required bandwidth of the busy ONU can be met. For example, if X is the required bandwidth, and the same as the ONU bandwidth data in the previous example, X=9/0.7-10≈2.9, it is necessary to implement a bandwidth restoration strategy to restore the ONU allocated bandwidth to 12.9M. For another example, take the bandwidth Y restored to the initial value as the required bandwidth, assuming that the initial bandwidth is 15M, then the allocated bandwidth of the ONU is 10M<initial bandwidth 15M, Y=5M, and the bandwidth restoration strategy needs to be implemented to restore the allocated bandwidth of the ONU to 15M . Other embodiments also include using min(X, Y) as the required bandwidth. There are many specific bandwidth recovery algorithms: for example, simply calculate and adjust the bandwidth according to the order of the remaining bandwidth from large to small; or calculate the bandwidth adjustment rate (equal to the required bandwidth divided by the sum of the remaining The bandwidth of the ONU whose bandwidth is greater than the initial bandwidth is proportionally controlled, and the adjusted bandwidth of the adjusted ONU is equal to the bandwidth adjustment rate multiplied by its remaining bandwidth. This not only ensures the full utilization of network resources, but also does not affect the fairness of resource allocation.

FIG. 7 shows a method for dynamically allocating joint bandwidth of multiple OLTs to the convergence layer equipment of the MAN connected to the OLT. The method of multi-OLT joint bandwidth dynamic allocation is the same as the method of OLT internal bandwidth dynamic allocation in Figure 6. The difference is that the management object is multiple OLTs. The content of steps 4.1 to 4.3 and steps 6.1 to 6.2 will be described in detail below.

Step 4.1: Find an idle OLT under the same BRAS/SR. If there is no idle OLT, the bandwidth adjustment requirements cannot be met, and continue to implement other bandwidth allocation strategies;

When there is an idle OLT, perform step 4.2: calculate the required bandwidth of the busy OLT, calculate the sum of the idle bandwidth of the idle OLT, and update the allocated bandwidth of the corresponding busy OLT according to the bandwidth scheduling algorithm. There are many specific bandwidth scheduling algorithms, such as simply counting and adjusting bandwidth according to the order of idle bandwidth from large to small, or first calculating the bandwidth adjustment rate (equal to the required bandwidth divided by the sum of idle bandwidth), if the bandwidth adjustment rate is greater than 1, then directly Statistically adjust the bandwidth according to the order of idle bandwidth from large to small. If the bandwidth adjustment rate is less than 1, then calculate and adjust the bandwidth according to the bandwidth adjustment rate (the adjusted bandwidth of each OLT is equal to the bandwidth adjustment rate multiplied by its idle bandwidth).

Step 4.3, judge the status of the busy OLT after optimization. If the busy problem is solved, go to step 7; otherwise, it indicates that the bandwidth adjustment requirement cannot be met, and other bandwidth allocation strategies need to be continued.

In order to ensure the fairness of bandwidth allocation, a multi-OLT joint bandwidth restoration strategy is introduced. When multi-OLT joint bandwidth allocation fails, step 6.1: when the real-time bandwidth is less than the initial bandwidth, then perform step 6.2, otherwise perform step 7.

Step 6.2: First calculate the actual required bandwidth X of the busy OLT, and the required bandwidth Y restored to the initial value, then count the OLTs whose real-time bandwidth is greater than the initial bandwidth, and calculate the remaining bandwidth restored to the initial value, expressed as Y or min( X, Y) is the required bandwidth, and the allocated bandwidth of the corresponding OLT is calculated and updated according to the bandwidth scheduling algorithm. Like the scheduling algorithm of the bandwidth allocation strategy, there are many specific bandwidth recovery algorithms: for example, simply calculate and adjust the bandwidth according to the order of the remaining bandwidth from large to small; or first calculate the bandwidth adjustment rate (equal to the required bandwidth divided by the sum of the remaining bandwidth), According to the bandwidth adjustment rate, the bandwidth of each OLT whose allocated bandwidth is greater than the initial bandwidth is proportionally controlled, and the adjusted bandwidth of the adjusted OLT is equal to the bandwidth adjustment rate multiplied by its remaining bandwidth. This not only ensures the full utilization of network resources, but also does not affect the fairness of resource allocation. Then go to step 7.

Regarding the 7th step, in the embodiment of Fig. 5～Fig. 7, the control object of bandwidth allocation mechanism in OLT is the bandwidth attribute of OLT, and the multiple ONUs that OLT constraint is connected with it; The control object of multi-OLT joint bandwidth allocation mechanism is SR /BRAS bandwidth attribute, SR/BRAS restricts multiple OLTs connected to it. The bandwidth obtained by the OLT or the reduced bandwidth in the joint bandwidth allocation mechanism of multiple OLTs is the total upstream bandwidth allocated to each OLT by the SR/BRAS, and the OLT should allocate the bandwidth of the internal ONU according to the total upstream bandwidth in real time.

The solution of the present invention does not limit how the total upstream bandwidth of the OLT is allocated to each ONU. In special cases, the sum of bandwidth allocated by OLT to each ONU is less than the total upstream bandwidth allocated to OLT by BRAS/SR, resulting in "redundant bandwidth" of OLT. Based on the teaching of the bandwidth dynamic allocation method shown in Figures 5 to 6 in this manual, implementers can also be inspired to make effective use of this part of the redundant bandwidth (when it exists, it can be regarded as the idle bandwidth of the ONU).

Claims (10)

1. A device for realizing bandwidth dynamic allocation in an optical access network, comprising a control layer and a managed device, the managed device comprising an OLT and a metropolitan convergence layer device directly connected to the OLT, the metropolitan convergence layer Layer equipment includes SR and BRAS, characterized in that, The control layer includes a communication agent module, a bandwidth resource analysis module, and a policy decision module; The bandwidth resource analysis module collects and analyzes the real-time traffic information of the underlying equipment, and sends the analyzed traffic conditions of the entire network to the policy decision module; The policy decision-making module formulates bandwidth optimization strategies according to the distribution of network traffic and the allocation of network bandwidth; The communication agent module includes message parsing and conversion, protocol stack functions, and communicates with managed devices; The information flow between the control layer and the managed device is bidirectional. The managed device reports traffic status information to the control layer, and the control layer issues bandwidth optimization policies to the managed device. 2. The device for realizing bandwidth dynamic distribution in the optical access network as claimed in claim 1, further comprising a distributed network management server, and the distributed network management server includes an access network distributed network management server, a metropolitan area network Distributed network management server, The communication agent module conducts two-way communication with the managed device through the network management server, The control layer reads the network status database on the server side, issues bandwidth optimization configurations through analysis and decision-making, and then updates the network configuration of managed devices through the network management server to realize dynamic bandwidth adjustment. The communication interface between the communication agent module and the network management server uses the same communication protocol as the communication interface between the network management client and the network management server. 3. The device for realizing dynamic allocation of bandwidth in the optical access network according to claim 1, characterized in that, the control layer communication agent module is integrated with the managed equipment. 4. The device for realizing dynamic allocation of bandwidth in an optical access network according to claim 2, characterized in that, the communication agent module of the control layer is integrated with the network management server. 5. The device for realizing dynamic allocation of bandwidth in the optical access network according to any one of claims 1 to 4, wherein the network operating system is used to implement the bandwidth resource analysis module of the control layer with application software on the operating system , Policy decision-making module; the network operating system provides a unified south and north interface, the south interface is connected with the communication agent module, and the north interface is used to realize the extension of the application layer. 6. A method for realizing dynamic allocation of bandwidth in an optical access network, for the device described in claims 1 to 5, characterized in that it comprises the following steps: Step 1: Initialize, allocate the initial bandwidth of all devices; Step 2: Monitor network traffic and find busy hotspots; Step 3: Execute the OLT internal bandwidth optimization strategy to optimize the bandwidth resource allocation between ONUs under the same OLT; Step 4: Execute the multi-OLT joint bandwidth optimization strategy to optimize the bandwidth resource allocation between OLTs under the same MAN access switch; Step 5: Implement the OLT internal bandwidth recovery strategy, and by adjusting the OLT internal bandwidth resource allocation, the bandwidth allocation value of the busy hotspot is forcibly restored within the initial bandwidth value range to ensure the fairness of bandwidth allocation; Step 6: Execute the multi-OLT joint bandwidth recovery strategy. By adjusting the bandwidth resource allocation between multiple OLTs, the bandwidth allocation value of the busy hotspot is forcibly restored within the initial bandwidth value range to ensure the fairness of bandwidth allocation; Step 7: Update the allocated bandwidth of the device according to the bandwidth value determined in steps 3 to 6, and then return to step 2; The relationship between steps 2 to 6 in the above steps is: If the second step finds a busy hotspot in the network, then implement a dynamic bandwidth allocation mechanism, including a dynamic bandwidth optimization mechanism and a dynamic bandwidth recovery mechanism; if no busy hotspot in the network is found in the second step, then perform loop detection on the network traffic; The bandwidth dynamic optimization mechanism involves steps 3 and 4, and there are three options: only include step 3; only include step 4; include steps 3 and 4; If the bandwidth optimization mechanism cannot meet the bandwidth requirements of busy hotspots, the bandwidth recovery mechanism will be implemented; The bandwidth recovery mechanism involves steps 5 and 6. Corresponding to the bandwidth dynamic optimization mechanism above, there are three schemes for the bandwidth recovery mechanism: only step 5 is included; step 6 is included only; steps 5 and 6 are included. 7. realize the method for bandwidth dynamic distribution in optical access network as claimed in claim 6, described bandwidth dynamic optimization mechanism comprises the 3rd step and the 4th step, it is characterized in that, Execute step 3 first. If the sum of available free bandwidth of each ONU belonging to the same OLT is greater than the bandwidth requirement, the bandwidth requirement has been met through the operation of step 3, and step 4 will not be executed; if the ONUs belonging to the same OLT If the total free bandwidth available to each ONU under an OLT is less than the bandwidth requirement, step 4 needs to be performed. 8. realize the method for bandwidth dynamic distribution in optical access network as claimed in claim 7, described bandwidth dynamic recovery mechanism carries out the 5th step and the 6th step, it is characterized in that, First perform step 5, and skip step 6 if step 5 can meet the bandwidth requirement; if step 5 cannot meet the requirement, perform step 6. 9. The method for dynamically distributing bandwidth in the optical access network according to any one of claims 6 to 8, wherein said OLT internal bandwidth optimization strategy comprises the 3.1st step, the 3.2nd step, the 3.3rd step; The OLT internal bandwidth recovery strategy includes the 5.1st step and the 5.2nd step, wherein Step 3.1: Find all idle ONUs under the same OLT. When there are idle ONUs, perform step 3.2; if there are no idle ONUs, bandwidth adjustment requirements cannot be met, and other bandwidth allocation strategies continue to be implemented; Step 3.2: Determine the new allocated bandwidth of each ONU according to the required bandwidth of the busy ONU and the idle bandwidth of the idle ONU until the required bandwidth of the busy ONU is met; Step 3.3: Judging the state of the busy ONU after optimization, if the busy problem is solved, execute step 7, otherwise it indicates that the bandwidth adjustment requirements cannot be met, and continue to implement other bandwidth allocation strategies; Step 5.1: Determine whether the bandwidth allocation of the busy ONU is less than the initial bandwidth, if yes, execute step 5.2, otherwise skip step 5.2; Step 5.2: Determine the new allocated bandwidth of each corresponding ONU according to the required bandwidth of the busy ONU and the remaining bandwidth of other ONUs. 10. The method for dynamically distributing bandwidth in an optical access network according to any one of claims 6 to 8, wherein the multi-OLT joint bandwidth optimization strategy comprises the 4.1st step, the 4.2th step, and the 4.3rd step; Described many OLT joint bandwidth restoration strategy comprises the 6.1st step, the 6.2nd step; Wherein Step 4.1: Find all idle OLTs under the same BRAS/SR. When there is an idle OLT, perform step 4.2; if there is no idle OLT, the bandwidth adjustment requirements cannot be met, and continue to implement other bandwidth allocation strategies; Step 4.2: According to the required bandwidth of the busy OLT and the idle bandwidth of the idle OLT, determine the new allocated bandwidth of each OLT until the required bandwidth of the busy OLT is met; Step 4.3: Determine the state of the busy OLT after optimization, and execute step 7 if the busy problem is solved, otherwise it indicates that the bandwidth adjustment requirements cannot be met, and continue to implement other bandwidth allocation strategies; Step 6.1: Determine whether the bandwidth allocation of the busy OLT is less than the initial bandwidth, if yes, execute step 6.2, otherwise execute step 7; Step 6.2: Determine the new allocated bandwidth of each corresponding OLT according to the required bandwidth of the busy OLT and the remaining bandwidth of other OLTs.