WO2018090580A1

WO2018090580A1 - Method and apparatus for sensing optical access network service stream and computer storage medium

Info

Publication number: WO2018090580A1
Application number: PCT/CN2017/084225
Authority: WO
Inventors: 白晖峰; 王东山; 王立城; 宋彦斌; 赵冲; 刘全春
Original assignee: 北京智芯微电子科技有限公司
Priority date: 2016-11-17
Filing date: 2017-05-12
Publication date: 2018-05-24
Also published as: CN106375136A; CN106375136B

Abstract

The embodiments of the present invention relate to a method and apparatus for sensing an optical access network service stream and a computer storage medium, the method comprising: receiving service stream data and extracting characteristic parameters of the service stream data; normalizing the characteristic parameters to obtain a characteristic set of the service stream; and determining a type of classification to which the service stream belongs according to the characteristic set of the service stream and a preset simplified echo state network model.

Description

Optical access network service flow sensing method, device and computer storage medium

Cross-reference to related applications

The present application is based on a Chinese patent application filed on Jan. 17, 2016, the entire disclosure of which is hereby incorporated by reference.

Technical field

The present invention relates to the field of service sensing technologies, and in particular, to an optical access network service flow sensing method, apparatus, and computer storage medium.

Background technique

Optical access network refers to the use of optical fiber as the main transmission medium to realize the information transmission function of the access network. Passive Optical Network (PON) is the main form of optical access network. The main components of the PON system are the Optical Line Terminal (OLT) and the Optical Network Unit (ONU). The OLT is connected to the service node and connected to the user through the ONU.

As the optical access network services become more complex, in order to obtain better quality of service (QoS) guarantees, relevant network behaviors are implemented for service identification and classification, and the premise and basis of end-to-end QoS of services are further improved. When analyzing service performance, it is often necessary to know the traffic characteristics of a single service, and the statistical characteristics of concurrent flows carried by the network, which are used to guide the formulation and implementation of traffic engineering strategies, and thus the service flow perception has emerged. Traffic flow sensing is a higher-level traffic monitoring method. It classifies and identifies data packets according to different service flow definitions, and performs corresponding resource optimization scheduling to improve the optical access network's ability to support multiple services.

In the service flow sensing technology of the optical access network, the pattern recognition algorithm based on the characteristics of the traffic flow starts from To an increasingly important role, and the performance of pattern recognition algorithms directly affect the accuracy and efficiency of business perception. The Echo State Network (ESN) algorithm is a new neural network algorithm that can be used for pattern recognition. The ESN uses a reserve pool consisting of randomly sparsely connected nodes (neurons) as a hidden layer for high-dimensional, non-linear representation of the input. The generation process of the reserve pool is independent of the training process of the echo state network. Therefore, the linear method is needed to train the weight of the reserve pool to the output layer, so that the training process of the network is simplified, and the global optimality of the weight determination is ensured. Good generalization ability avoids the problems that the training algorithm existing in the traditional neural network is complex and easy to fall into the local minimum. The above advantages make the echo state network have great application potential in traffic awareness.

As the field of echo state network applications becomes more and more complex, and the real-time requirements of applications continue to increase, its hardware implementation is receiving more and more attention. The traditional reserve pool consists of a large number of nodes. This complex physical topology formed by a large number of node interconnections requires high hardware implementation techniques. Therefore, the traditional reserve pool calculation is mostly based on software, which seriously restricts the processing speed of the echo state network algorithm; in the high-speed optical access network, it is difficult to guarantee the real-time performance of the service recognition.

The information disclosed in this Background section is only intended to provide an understanding of the general background of the invention, and should not be construed as an admission

Summary of the invention

In view of this, the embodiments of the present invention are directed to a method, a device, and a computer storage medium for the optical access network service identification, so as to quickly determine the classification type to which the service flow belongs, and improve the efficiency of the service flow perception of the optical access network.

To solve the above technical problem, an embodiment of the present invention provides an optical access network service flow sensing method, including: receiving service flow data, extracting feature parameters of the service flow data, and normalizing the feature parameters. Processing to obtain a feature set of the service flow; The feature set of the service flow and the preset simplified echo state network model determine the classification type to which the service flow belongs.

In a possible implementation manner, the feature parameter includes: a packet length P _SIZE (i), a packet arrival interval P _INTERVAL (i), and a service duration P _DUR (i), where the feature parameter is Performing a normalization process to obtain a feature set of the service flow includes:

According to the formula

Calculate the feature set U(i) of the traffic flow, where P _{SIZE_MAX} is the maximum packet length of the statistics, P _{INTERVAL_MAX} is the maximum packet arrival interval, and P _{DUR_MAX} is the maximum service duration.

In a possible implementation, the determining, according to the feature set of the service flow and the preset simplified echo state network model, determining a classification type to which the service flow belongs includes: inputting a feature set of the service flow into the The preset simplified echo state network model calculates an output sample y(n); and determines a classification type to which the service flow belongs according to the calculated output sample y(n).

In a possible implementation manner, in the preset simplified echo state network model, the reserve pool structure of the echo state network model is set to a ring topology formed by N unit nodes, where N is the reserve pool The total number of nodes in the pool; the nodes in the reserve pool are generated according to the preset dynamic equations.

In a possible implementation manner, the preset dynamic equation is

The generating the nodes in the reserve pool according to the preset dynamic equation includes: integrating the dynamic equation to obtain a generation formula of the node x(i)

Where P is the average generation rate of the node, α is the excitation coefficient, δ is the node death rate, θ is the node spacing, T is the total distance between the node x(N) and the node x(0), and N is the total node of the reserve pool. The number, τ is the preset time period constant, t is the time variable, 0 < t < τ.

In order to solve the above technical problem, the embodiment of the present invention provides, in another aspect, an optical access network service flow sensing device, including: a parameter extraction module, configured to receive service flow data, and extract feature parameters of the service flow data. a parameter processing module configured to perform normalization processing on the feature parameter to obtain a feature set of the service flow; and a classification determining module configured to perform a network model according to the feature set of the service flow and a preset simplified echo state network model Determining the classification type to which the service flow belongs.

In a possible implementation manner, the feature parameter includes: a packet length P _SIZE (i), a packet arrival interval P _INTERVAL (i), and a service duration P _DUR (i), where the parameter processing module is configured to : according to the formula

In a possible implementation, the classification determining module includes: a calculation submodule configured to input a feature set of the service flow into the preset simplified echo state network model, and calculate an output sample y(n); Determining a submodule configured to determine a classification type to which the service flow belongs according to the calculated output sample y(n).

In a possible implementation manner, the classification determining module is further configured to determine a preset simplified echo state network model, wherein, in the preset simplified echo state network model, the echo state network model is reserved The pool structure is set to a ring topology composed of N unit nodes, where N is the total number of nodes in the reserve pool; nodes in the reserve pool are generated according to a preset dynamic equation.

In a possible implementation manner, the preset dynamic equation is

The generating, by the classification determining module, the nodes in the reserve pool according to the preset dynamic equation includes: integrating the dynamic equation to obtain a generating formula of the node x(i)

In order to solve the above technical problem, an embodiment of the present invention provides, in yet another aspect, a computer storage medium, wherein the computer storage medium stores computer executable instructions for performing the implementations described above. The optical access network service flow sensing method described in the example.

An optical access network service flow sensing method and apparatus, and a computer storage medium, receive service flow data, extract feature parameters of service flow data, and obtain a feature set of a service flow according to the feature parameter, according to the feature set And the preset simplified echo state network model can quickly determine the classification type to which the service flow belongs, and improve the efficiency of the service flow perception of the optical access network. Moreover, the simple ring topology and dynamic equations are combined to form nodes. On the one hand, the complexity of the traditional echo state network model is reduced. On the other hand, the dynamic equations are used to generate nodes in the simplified echo state network model to maintain the operation accuracy.

Other features of the present invention will become apparent from the following detailed description of exemplary embodiments Aspects will become clear.

DRAWINGS

The accompanying drawings, which are incorporated in FIG

Figure 1 shows a schematic structural diagram of a current echo state network;

FIG. 2 is a flowchart of a method for sensing traffic flow of an optical access network according to Embodiment 1 of the present invention;

FIG. 3 is a flowchart of another optical access network service flow sensing method according to Embodiment 2 of the present invention.

4 is a schematic structural diagram of a simplified echo state network model in Embodiment 2 of the present invention;

5 is a hardware implementation block diagram of an optical access network service flow sensing method (based on a simplified echo state network) according to Embodiment 3 of the present invention;

FIG. 6 is a schematic structural diagram of an optical access network service flow sensing device according to Embodiment 4 of the present invention;

FIG. 7 is a schematic structural diagram of another optical access network service flow sensing apparatus according to Embodiment 5 of the present invention.

detailed description

The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings, but it is understood that the scope of the present invention is not limited by the specific embodiments.

The technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. It is a partial embodiment of the invention, and not all of the embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative efforts are within the scope of the present invention. Unless otherwise stated It is to be understood that the term "comprises" or "comprises" or "comprises" or "comprises" or "the" Or other components.

The word "exemplary" is used exclusively herein to mean "serving as an example, embodiment, or illustrative." Any embodiment described herein as "exemplary" is not necessarily to be construed as preferred or preferred.

In addition, numerous specific details are set forth in the Detailed Description of the invention in the Detailed Description. Those skilled in the art will appreciate that the invention may be practiced without some specific details. In some instances, methods, means, and components that are well known to those skilled in the art are not described in detail in order to facilitate the invention.

The following describes the principle of the current Echo State Network (ESN):

The structure of the current echo state network is shown in Figure 1. It consists of an input layer, a reserve pool, and an output layer, which are randomly generated, large-scale, sparse connections (SD usually maintains 1%-5% connections, SD is the recursive structure of the nodes in the reserve pool that are connected to each other as a percentage of the total nodes. Assuming that the echo state network consists of K input units, N reserve pool processing units, and L output units, the basic equations of the echo state network are the following equations (1) and (2):

x(n+1)=f(W ⁱⁿ u(n+1)+Wx(n)+W ^back y(n)) (1)

y(n+1)=f ^out (W ^out u(n+1)+Wx(n+1)+W ^back y(n)) (2)

Where u(n), x(n), and y(n) are the input variables, state variables, and output variables of the ESN, respectively; f and f ^out are the activation function vectors of the reserve pool processing unit and the output unit, respectively. W ⁱⁿ represents the connection weight between the input unit and the reserve pool processing unit, W represents the connection weight between the internal processing units of the reserve pool, W ^back represents the connection weight between the output layer and the reserve pool, and W ^out is the reserve pool and The connection weight of the output unit. In addition, W ⁱⁿ , W and W ^back remain unchanged after initialization, so there is no need to obtain training; and W ^out needs to be obtained through training.

In the training of ESN, the sample data obtains W ^out by using the randomly generated weight matrix W ⁱⁿ and W ^back to stimulate the reserve pool processing unit to minimize the training mean square error by linear regression.

The basic principle of the echo state network classification and recognition algorithm is shown in the following formula (3):

Equation (3) is used to train the initialized echo state network model, where n represents only different samples, not time. In the process of classification training, the input samples must be kept unchanged until the state variables of the reserve pool tend to be stable, so that the difference between the results of the two iterations is the smallest, that is, the input sample u(n+1) is kept unchanged, and finally The (i)th iteration is consistent with the (i-1)th iteration, and an echo state network model is generated.

The training process of the ESN model is as follows:

Step 1: Initialization of the ESN. Set the size of the reserve pool (that is, the total number of nodes in the reserve pool), the internal connection weight matrix and other parameters; according to these parameters, the W ⁱⁿ , W and W ^back must also be initialized;

Step 2: Select the training sample set. Since random events are inevitable in the data collection process, the existence of abnormal data is inevitably caused. Therefore, abnormal data needs to be identified, and normal data is selected to form a training sample set;

Step 3: Form the network status. The state of the echo state network is updated from the initialization state, and the relevant values of the current state of the echo state network need to be saved after each round of updating;

Step 4: ESN training. The training process of the ESN is to obtain the output weight matrix W ^out according to the input and output training sample pairs.

Example 1

FIG. 2 is a flowchart of a method for sensing traffic flow of an optical access network according to an embodiment of the present invention. As shown in FIG. 2, the method may include: step S201, step S202, and step S203.

Step S201: Receive service flow data, and extract feature parameters of the service flow data.

Step S202: Perform normalization processing on the feature parameters to obtain characteristics of the service flow. set.

Step S203: Determine, according to the feature set of the service flow and the preset simplified echo state network model, a classification type to which the service flow belongs.

An optical access network service flow sensing method according to an embodiment of the present invention receives service flow data, extracts feature parameters of service flow data, and obtains a feature set of the service flow according to the feature parameter, according to the feature set and a preset simplified echo. The state network model can quickly determine the classification type to which the service flow belongs, and improve the efficiency of the service flow perception of the optical access network.

Example 2

FIG. 3 is a flowchart of another optical access network service flow sensing method according to an embodiment of the present invention. As shown in FIG. 3, the method may include: step S301, step S302, step S303, and step S304.

Step S301: Receive service flow data, and extract feature parameters of the service flow data.

In a possible implementation manner, the feature parameters include: a packet length P _SIZE (i), a packet arrival interval P _INTERVAL (i), and a service duration P _DUR (i). The data packet arrival interval is: the average time interval in which the data packets of the same service flow arrive continuously; the service duration is: the duration from the first data packet to the last data packet of the same service flow.

Step S302: Perform normalization processing on the feature parameters to obtain a feature set of the service flow.

In a possible implementation, the normalizing the feature parameters to obtain the feature set of the service flow includes:

According to formula (4)

Calculating the feature set U(i) of the service flow;

Among them, P _{SIZE_MAX} is the maximum packet length of the statistics, P _{INTERVAL_MAX} is the maximum packet arrival interval, and P _{DUR_MAX} is the maximum service duration.

The service flow sensing method according to the embodiment of the present invention is based on a preset simplified echo state network model, which is essentially a mapping of service features to service types, and the essence thereof is to determine the classification of decision attributes (service types) according to condition attributes (service characteristics). Identification mechanism. The optical access network system performs service optimization scheduling according to the classification and identification result. The mechanism is divided into service flow feature extraction and Simplified Echo-State-Network (S-ESN) training (same as the ESN training process described above) and S-ESN decides three processes.

For each accessed traffic flow, its characteristic parameters are extracted for the received data stream, including: packet length PSIZE(i), packet arrival interval PINTERVAL(i), and service duration PDERR(i). The feature parameters are normalized according to formula (4) to avoid overfitting, thereby obtaining a feature set U(i) describing the traffic flow.

Step S303: Input a feature set of the service flow into the preset simplified echo state network model, and calculate an output sample y(n).

The embodiment of the invention proposes a simplified echo state network (S-ESN) model. In order to reduce the complexity of the reserve pool in the echo state network, the reserve pool structure in the existing echo state network model is simplified into N unit nodes. The ring topology is constructed, and the S-ESN model is shown in Figure 4.

While simplifying the existing echo state network model, in order to maintain the reserve pool operation accuracy, the embodiment of the present invention introduces a dynamic equation with rich dynamic features to generate nodes in the reserve pool.

In a possible implementation manner, in the preset simplified echo state network model as shown in FIG. 4,

The reserve pool structure of the echo state network model is set to a ring topology composed of N unit nodes, where N is the total number of nodes in the reserve pool; nodes in the reserve pool are generated according to a preset dynamic equation.

The preset dynamic equation is:

The generating the nodes in the reserve pool according to the preset dynamic equation includes:

Integrating the dynamic equation to obtain the formula for generating the node x(i):

Where P is the average generation rate of the node (such as the value of 19.8), α is the excitation coefficient (if the value is 1), δ is the node death rate (such as the value of 0.8), and θ is the node spacing (if the fixed value is 0.2) . T is the total distance between node x(N) and node x(0); N is the total number of nodes in the reserve pool. τ is a preset time period constant, t is a time variable, 0 < t < τ.

Step S304: Determine, according to the calculated output sample y(n), the classification type to which the service flow belongs.

The feature set U(i) of the service flow (as an input sample) is input into the trained S-ESN, and the output sample y(n) is calculated by calculating the formula (1) and the formula (2), and according to the output sample y(n) Determine the classification type to which the service belongs, and different output samples correspond to different classification types.

An optical access network service flow sensing method according to an embodiment of the present invention receives service flow data, extracts feature parameters of a service flow, and obtains a feature set according to the feature parameter, according to the feature set and a preset simplified echo state network model. The classification type of the service flow can be quickly determined, and the efficiency of the service flow perception of the optical access network is improved. Moreover, the simple ring topology and dynamic equations are combined to form nodes. On the one hand, the complexity of the traditional echo state network model is reduced. On the other hand, the dynamic equations are used to generate nodes in the simplified echo state network model to maintain the operation accuracy.

Example 3

FIG. 5 shows an optical access network industry based on a simplified echo state network according to an embodiment of the present invention. The hardware implementation block diagram of the flow-aware method is as shown in FIG. 5. In the implementation manner of the optical access network service sensing mechanism, the embodiment of the present invention designs a master according to the master-slave architecture between the OLT and the ONU in the PON system. The S-ESN service sensing mechanism of the slave system is composed of an "S-ESN main module" and a plurality of "S-ESN sub-modules".

1) S-ESN main module: running in the OLT device, mainly responsible for the initialization and training of the S-ESN to form a well-trained S-ESN model. The OLT then broadcasts the trained S-ESN model information to each ONU.

2) S-ESN sub-module: Runs in the ONU device, and forms an S-ESN model according to the S-ESN information broadcast by the OLT. Then, the S-ESN sub-modules in each ONU work independently to perform service flow sensing. The S-ESN sub-module extracts the characteristic parameters of each service flow and performs normalization processing, and inputs the S-ESN model to perform the operation to obtain the classification and recognition result, that is, the S-ESN decision; then the scheduling module in the ONU performs the service according to the classification identification result. Optimize scheduling.

Example 4

FIG. 6 is a schematic structural diagram of an optical access network service flow sensing device according to an embodiment of the present invention. As shown in FIG. 6, the device includes:

The parameter extraction module 61 is configured to receive service flow data, and extract feature parameters of the service flow data.

The parameter processing module 62 is configured to perform normalization processing on the feature parameters to obtain a feature set of the service flow;

The classification determining module 63 is configured to determine, according to the feature set of the service flow and the preset simplified echo state network model, a classification type to which the service flow belongs.

In the embodiment of the present invention, the parameter extraction module 61, the parameter processing module 62, and the classification determining module 63 in the optical access network service flow sensing device may be used by the optical access network service flow sensing device in practical applications. Or a central processing unit (CPU) in the device where the optical access network service flow sensing device is located, and a digital signal processor (DSP, Digital Signal) Processor, Micro Control Unit (MCU) or Field-Programmable Gate Array (FPGA).

It should be understood by those skilled in the art that the functions of the modules in the optical access network service flow sensing device of this embodiment can be understood by referring to the related description of the optical access network service flow sensing method described in Embodiment 1.

An optical access network service flow sensing device according to an embodiment of the present invention receives service flow data, extracts feature parameters of a service flow, and obtains a feature set according to the feature parameter, according to the feature set and a preset simplified echo state network model. The classification type of the service flow can be quickly determined, and the efficiency of the service flow perception of the optical access network is improved.

Example 5

FIG. 7 is a schematic structural diagram of an optical access network service flow sensing device according to an embodiment of the present invention. As shown in FIG. 7, the device includes:

In a possible implementation manner, the feature parameters include: a packet length P _SIZE (i), a packet arrival interval P _INTERVAL (i), and a service duration P _DUR (i),

As an implementation manner, the parameter processing module 62 is specifically configured to:

According to the formula

Calculate the feature set U(i) of the traffic flow,

In a possible implementation manner, the classification determining module 63 includes:

The calculation sub-module 631 is configured to input the feature set of the service flow into the preset simplified echo state network model, and calculate an output sample y(n);

The determining sub-module 632 is configured to determine a classification type to which the service flow belongs according to the calculated output sample y(n).

In a possible implementation manner, the classification determining module 63 is further configured to:

Determining a preset simplified echo state network model;

Wherein, in the preset simplified echo state network model, the reserve pool structure of the echo state network model is set to a ring topology formed by N unit nodes, where N is the total number of nodes in the reserve pool; The nodes in the reserve pool are generated according to preset dynamic equations.

In a possible implementation manner, the preset dynamic equation is

The generating, by the classification determining module 63, the nodes in the reserve pool according to the preset dynamic equation includes:

Integrating the dynamic equation to obtain the formula for generating the node x(i)

Where P is the average generation rate of nodes, α is the excitation coefficient, δ is the node extinction rate, θ is the node spacing, T is the total distance between node x(N) and node x(0); N is the total node of the reserve pool number. τ is a preset time period constant, t is a time variable, 0 < t < τ.

In the embodiment of the present invention, the parameter extraction module 61, the parameter processing module 62, the classification determination module 63, and the sub-modules included in the classification determination module 63 in the optical access network service flow sensing device are all in practical applications. It can be implemented by a CPU, a DSP, an MCU, or an FPGA in the device where the optical access network service flow sensing device or the optical access network service flow sensing device is located.

It should be understood by those skilled in the art that the functions of the modules in the optical access network service flow sensing device of this embodiment can be understood by referring to the related description of the optical access network service flow sensing method described in Embodiment 2.

An optical access network service flow sensing device according to an embodiment of the present invention receives service flow data, extracts feature parameters of a service flow, and obtains a feature set according to the feature parameter, according to the feature set and a preset simplified echo state network model. The classification type of the service flow can be quickly determined, and the efficiency of the service flow perception of the optical access network is improved. Moreover, the simple ring topology and dynamic equations are combined to form nodes. On the one hand, the complexity of the traditional echo state network model is reduced. On the other hand, the dynamic equations are used to generate nodes in the simplified echo state network model to maintain the operation accuracy.

Embodiments of the present invention also describe a computer storage medium storing one or more programs, the one or more programs being executable by one or more processors to implement the following steps:

Receiving service flow data, and extracting characteristic parameters of the service flow data;

Normalizing the feature parameters to obtain a feature set of the service flow;

Determining, according to the feature set of the service flow and the preset simplified echo state network model, a classification type to which the service flow belongs.

As an implementation manner, the feature parameters include: a packet length P _SIZE (i), a packet arrival interval P _INTERVAL (i), and a service duration P _DUR (i),

The one or more programs may be executed by one or more processors to implement normalization processing on the feature parameters to obtain a feature set of the service flow:

According to the formula

Calculate the feature set U(i) of the traffic flow,

In one embodiment, the one or more programs may be executed by the one or more processors to determine the service according to a feature set of the service flow and a preset simplified echo state network model. Steps for the classification type to which the stream belongs:

Inputting a feature set of the service flow into the preset simplified echo state network model, and calculating an output sample y(n);

The classification type to which the service flow belongs is determined according to the calculated output sample y(n).

As an implementation manner, in a preset simplified echo state network model, the reserve pool structure of the echo state network model is set to a ring topology formed by N unit nodes, where N is the total number of nodes in the reserve pool. Generate nodes in the reserve pool based on preset dynamic equations.

The preset dynamic equation is

As an embodiment, the dynamic equation is integrated to obtain a formula for generating a node x(i)

It should be understood by those skilled in the art that the functions of the programs in the computer storage medium of the present embodiment can be understood by referring to the related description of the optical access network service flow sensing method described in the embodiments.

In the several embodiments provided by the present application, it should be understood that the disclosed apparatus and method may be implemented in other manners. The device embodiments described above are merely illustrative. For example, the division of the unit is only a logical function division. In actual implementation, there may be another division manner, such as: multiple units or components may be combined, or Can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, or direct coupling, or communication connection of the components shown or discussed may be indirect coupling or communication connection through some interfaces, devices or units, and may be electrical, mechanical or other forms. of.

The units described above as separate components may or may not be physically separated, and the components displayed as the unit may or may not be physical units, that is, may be located in one place or distributed to multiple network units; Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of the embodiment.

In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may be separately used as one unit, or two or more units may be integrated into one unit; The unit can be implemented in the form of hardware or in the form of hardware plus software functional units.

One of ordinary skill in the art can understand that all or part of the steps of the above method embodiments are implemented. The foregoing may be completed by a program instruction related hardware, where the foregoing program may be stored in a computer readable storage medium, and when executed, the program includes the steps of the foregoing method embodiment; and the foregoing storage medium includes: mobile storage A device that can store program code, such as a device, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

Alternatively, the above-described integrated unit of the present invention may be stored in a computer readable storage medium if it is implemented in the form of a software function module and sold or used as a standalone product. Based on such understanding, the technical solution of the embodiments of the present invention may be embodied in the form of a software product in essence or in the form of a software product stored in a storage medium, including a plurality of instructions. A computer device (which may be a personal computer, server, or network device, etc.) is caused to perform all or part of the methods described in various embodiments of the present invention. The foregoing storage medium includes various media that can store program codes, such as a mobile storage device, a ROM, a RAM, a magnetic disk, or an optical disk.

The foregoing description of the specific exemplary embodiments of the present invention has The description is not intended to limit the invention to the precise forms disclosed. The embodiments were chosen and described in order to explain the particular embodiments of the invention Choose and change. The scope of the invention is intended to be defined by the claims and their equivalents.

The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separate, and the components displayed as units may or may not be physical units, ie may be located A place, or it can be distributed to multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the embodiment. Those of ordinary skill in the art can understand and implement without deliberate labor.

Industrial applicability

In the embodiment of the present invention, the service flow data is received, the feature parameters of the service flow data are extracted, and the feature set of the service flow is obtained according to the feature parameter, and then the light is according to the feature set and the preset simplified echo state network model. The access network service flow sensing mode can quickly determine the classification type to which the service flow belongs, and improve the service flow sensing efficiency of the optical access network. The embodiment of the present invention adopts a simple ring topology and a dynamic equation to combine to generate a node. On the one hand, the complexity of the traditional echo state network model is reduced, and on the other hand, the dynamic equation is used to generate nodes in the simplified echo state network model to maintain the operation. Accuracy.

Claims

An optical access network service flow sensing method, where the optical access network service flow sensing method includes:

Receiving service flow data, and extracting characteristic parameters of the service flow data;

Normalizing the feature parameters to obtain a feature set of the service flow;

Determining, according to the feature set of the service flow and the preset simplified echo state network model, a classification type to which the service flow belongs.
The optical access network service flow sensing method according to claim 1, wherein the characteristic parameters comprise: a packet length P SIZE (i), a packet arrival interval P INTERVAL (i), and a service duration P DUR (i) ),

Performing normalization on the feature parameters to obtain a feature set of the service flow includes:

According to the formula
Calculate the feature set U(i) of the traffic flow,

Among them, P SIZE_MAX is the maximum packet length of the statistics, P INTERVAL_MAX is the maximum packet arrival interval, and P DUR_MAX is the maximum service duration.
The optical access network service flow sensing method according to claim 1 or 2, wherein the determining, according to the feature set of the service flow and the preset simplified echo state network model, determining a classification type to which the service flow belongs includes :

Inputting a feature set of the service flow into the preset simplified echo state network model, and calculating an output sample y(n);

The classification type to which the service flow belongs is determined according to the calculated output sample y(n).
The optical access network service flow sensing method according to claim 1 or 2, wherein In the preset simplified echo state network model,

Setting the reserve pool structure of the echo state network model to a ring topology formed by N unit nodes, where N is the total number of nodes in the reserve pool;

Generate nodes in the reserve pool based on preset dynamic equations.
The optical access network service flow sensing method according to claim 4, wherein the preset dynamic equation is

The generating the nodes in the reserve pool according to the preset dynamic equation includes:

Integrating the dynamic equation to obtain the formula for generating the node x(i)

Where P is the average generation rate of the node, α is the excitation coefficient, δ is the node death rate, θ is the node spacing, T is the total distance between the node x(N) and the node x(0), and N is the total node of the reserve pool. The number, τ is the preset time period constant, t is the time variable, 0 < t < τ.
An optical access network service flow sensing device, where the optical access network service flow sensing device includes:

a parameter extraction module, configured to receive service flow data, and extract feature parameters of the service flow data;

a parameter processing module, configured to perform normalization processing on the feature parameter to obtain a feature set of the service flow;

The classification determining module is configured to determine, according to the feature set of the service flow and the preset simplified echo state network model, a classification type to which the service flow belongs.
The optical access network traffic flow sensing device according to claim 6, wherein the characteristic parameters comprise: a packet length P SIZE (i), a packet arrival interval P INTERVAL (i), and a service duration P DUR (i) ),

The parameter processing module is configured to:

According to the formula
Calculate the feature set U(i) of the traffic flow,

Among them, P SIZE_MAX is the maximum packet length of the statistics, P INTERVAL_MAX is the maximum packet arrival interval, and P DUR_MAX is the maximum service duration.
The optical access network service flow sensing device according to claim 6 or 7, wherein the classification determining module comprises:

a calculation submodule configured to input a feature set of the service flow into the preset simplified echo state network model, and calculate an output sample y(n);

Determining a submodule configured to determine a classification type to which the service flow belongs according to the calculated output sample y(n).
The optical access network service flow sensing device according to claim 6 or 7, wherein the classification determining module is further configured to:

Determining a preset simplified echo state network model;

In the preset simplified echo state network model,

Setting the reserve pool structure of the echo state network model to a ring topology formed by N unit nodes, where N is the total number of nodes in the reserve pool;

Generate nodes in the reserve pool based on preset dynamic equations.
The optical access network traffic flow sensing device according to claim 9, wherein the preset dynamic equation is

The classification determining module is further configured to:

Integrating the dynamic equation to obtain the formula for generating the node x(i)

Where P is the average generation rate of the node, α is the excitation coefficient, δ is the node death rate, θ is the node spacing, T is the total distance between the node x(N) and the node x(0), and N is the total node of the reserve pool. The number, τ is the preset time period constant, t is the time variable, 0 < t < τ.
A computer storage medium having stored therein computer executable instructions for performing the method of any one of claims 1 to 5.