CN115914071A - Message transmission analysis method combining SRv6 with k nearest neighbor algorithm - Google Patents

Abstract

The invention discloses a message transmission analysis method combining SRv6 with a k nearest neighbor algorithm. The method comprises the following steps: s11, an address identifier generating module is initially allocated to the Srv6 network node; and S12, constructing a k-nearest neighbor algorithm model, an early warning model and generating an intelligent distribution identification module. The method combines the SRv6 message environment with the optimal path method in the SRv6 network, and improves the accuracy and flexible configuration of data transmission by adopting different distribution algorithms for an SRv6 message routing distribution mechanism.

Description

A Message Delivery Analysis Method Combining SRv6 with k-Nearest Neighbor Algorithm

technical field

The invention belongs to the field of network technology and security, and in particular relates to a message delivery analysis method combining SRv6 with a k-nearest neighbor algorithm.

Background technique

With the rapid development of computer technology, information network has become an important guarantee for social development. In the era of cloud-network integration, flexible and agile network service capabilities directly affect the competitiveness of operators. SR (SegmentRouting) is the core of source routing technology. One, SRv6 is the application of SR technology in IPv6 networks. The emergence of SRv6 is a huge innovation. It combines SDN technology to enable programmable networks, which provides innovative solutions for basic network services and value-added network services in the cloud network era. soil.

Chinese invention patent CN201911329833.1 discloses a SID allocation method and device based on an SRv6 network, including: obtaining at least one network segment, each network segment including at least one IPv6 address; dividing each network segment into different types and mutually A plurality of SID segments that do not overlap; determine the corresponding allocation method according to the type of each SID segment; allocate each SID segment to the routing information of the SR device according to the corresponding SID segment. SID to generate the routing table. However, this invention only guarantees the uniqueness of the SID, and cannot ensure the optimal path.

Chinese invention patent CN202010346913.4 discloses a SRv6-based data processing method and related equipment, including: the controller generates a segment identifier, the segment identifier includes position information, instructions, and a path identifier, and the path identifier is used for Indicating the forwarding path of the SRv6 message between network devices in the SR network; the controller sends the segment identifier to the network device, so that the network device forwards the SRv6 message according to the segment identifier. But this invention can't carry out the intelligent distribution of address identification.

Contents of the invention

Aiming at the deficiencies of the prior art, the present invention provides a message delivery analysis method combining SRv6 with the k-nearest neighbor algorithm. This method highlights the position of artificial intelligence in the SRv6 network in which the optimal path method is combined with the SRv6 message environment. Different allocation algorithms are used for the SRv6 packet routing allocation mechanism to improve the accuracy of transmitted data and flexible configuration.

In order to achieve the above object, the present invention adopts the following technical solutions:

A message delivery analysis method combining SRv6 k-nearest neighbor algorithm, comprising the following steps:

S11, Srv6 network node initial allocation address identifier generation module;

S12. Construct a k-nearest neighbor algorithm model, an early warning model and generate an intelligent allocation identification module;

Described step S12 comprises the following steps:

S121. Obtain the failure rate of all next-hop SR nodes associated with the initiating node through the k-nearest neighbor algorithm model, and then predict the alarm probability of all the next-hop SR nodes through the early warning model;

S122. Determine the optimal next-hop SR node by calculating the failure rate and alarm probability, until the optimal SR node for each hop to the destination SR node is obtained and connected in series to form an optimal address identifier;

S123. If the optimal address identifies each hop of the SR node is better than the initial address identifies each hop of the SR node, update the Argument initial address identification of the SRH part of the IPv6 extension header.

The SRv6 is a network forwarding technology, SR refers to the Segment Routing technology, v6 refers to native IPv6, and SRv6 is exactly IPv6+Segment Routing; the initial allocation identifier is all SIDs of all path Segment Lists from the current SR node to the target SR node, Connected together from bottom to top, with ### intervals in the middle, the initial address identification generated; the optimal address identification is through the construction of k-nearest neighbor algorithm model and early warning model, adding AI fault prediction and intelligent path optimization , and replace the initial allocation identifier after comparison and selection, it provides a new solution for complex service message forwarding in a super-large network, reduces resource consumption for message forwarding, and improves transmission data accuracy and flexible configuration.

The SR cuts the message forwarding path into different segments, and inserts segment information into the message at the starting point of the path. The intermediate node only needs to forward according to the segment information carried in the message. Such a path Segmentation, called "Segment", and identified by SID (Segment Identifier, segment identifier);

As a further description of the message delivery analysis method of SRv6 combined with the k-nearest neighbor algorithm of the present invention, the step S11 includes the following steps:

S111. Obtain all path Segment Lists from the current SR node to the target SR node;

S112, extract all the SIDs of the Segment List, connect them in series from bottom to top, and use ### as an interval in the middle to generate an initial address identifier;

S113. Before the current SR node transmits to the next-hop SR node, put the initial address identifier into the Argument of the SRH part of the IPv6 extension header.

The Segment List is when the current SR node in the SRv6 network performs IPv6 message transmission to the target SR node, all paths automatically assigned by the networking routing to automatically reach the target SR node, and it is stored in the SRv6 message protocol SID; The IPv6 extension header SRH is mainly composed of Locator, Function, and Argument, wherein, Locator=accessibility of location information, Function=definition of service function, Argument=enhancement; the address identification is convenient for locating communication objects.

As a further illustration of the message delivery analysis method of SRv6 of the present invention in conjunction with k nearest neighbor algorithm, described step S121 k nearest neighbor algorithm comprises the following steps:

S1211. Prepare data, and arrange the data according to dimensions;

S1212. Calculate the distance from the test sample point to each other sample point;

S1213. Sort each distance, and then select K points with the smallest distance;

S1214. Compare the categories to which the K points belong, and classify the test sample points into the category with the highest proportion among the K points according to the principle that the minority obeys the majority.

The k-nearest neighbor algorithm is used for data mining and classification. The so-called K-nearest neighbors means the K nearest neighbors. It means that each sample can be represented by its closest K neighbors. The nearest-neighbor algorithm is to combine data The method by which each record in the collection is classified.

As a further description of the message delivery analysis method of SRv6 combined with the k-nearest neighbor algorithm of the present invention, the k-nearest neighbor algorithm model formula is:

In the formula: x ₁ x ₂ ... x _n is the n-dimensional data of sample X; y ₁ y ₂ ... y _n is the n-dimensional data of sample Y; X is time, Y is the network delay in milliseconds (ms); d(x, y) is the distance between samples X and Y, and in the present invention is an associated node whose 24-hour delay of the current network node is <= 50 milliseconds.

As a further description of the message delivery analysis method of SRv6 combined with the k-nearest neighbor algorithm of the present invention, the early warning model is constructed using a Markov chain, and the formula is:

X(k+1)=X(k)×P

In the formula: X(k) represents the state vector of the trend analysis and prediction object at the time t=k, P represents the one-step transition probability matrix, and X(k+1) represents the state of the trend analysis and prediction object at the time t=k+1 vector.

As a further description of the message delivery analysis method of SRv6 combined with the k-nearest neighbor algorithm of the present invention, the calculation method of the failure rate and alarm probability in step S13 is a weighted average. The weighted average is used to calculate the original data in a reasonable proportion.

The present invention has the following beneficial effects:

When the IPv6 node in the SRv6 network is connected to the network, the present invention uses the k-nearest neighbor algorithm to obtain the probability of faulty nodes in all network nodes that may be the first hop, thereby screening out healthy network nodes, and then uses the Markov chain The alarm data is used as the analysis basis to analyze the failure probability of each Sid network node and assign a probability value. By adding AI fault prediction and intelligent path optimization, it provides a new solution for complex business message forwarding in a super-large network , and reduce the resource consumption of message forwarding, while improving the accuracy of transmitted data and flexible configuration.

Description of drawings

FIG. 1 is a flow chart of the overall solution of the message delivery analysis method of SRv6 combined with the k-nearest neighbor algorithm of the present invention.

FIG. 2 is a flow chart of the initial allocation address identifier generating module in FIG. 1 .

Fig. 3 is a flow chart of the module of constructing the k-nearest neighbor algorithm model, the early warning model and generating the intelligent distribution identification in Fig. 1 .

Figure 4 is a structural diagram of the SRv6 protocol.

Fig. 5 is a flowchart of the k-nearest neighbor algorithm.

Fig. 6 is a schematic diagram of the principle of the k-nearest neighbor algorithm model.

FIG. 7 is an overall framework diagram of the message delivery analysis method of SRv6 combined with the k-nearest neighbor algorithm of the present invention.

Detailed ways

The present invention will be further described below in conjunction with accompanying drawing.

A message delivery analysis method combining SRv6 with the k-nearest neighbor algorithm, as shown in Figure 1, comprising the following steps:

S11, Srv6 network node initial allocation address identifier generation module;

As shown in Figure 2, it specifically includes the following steps:

S111. Obtain the Segment List of all paths from the current SR node to the target SR node; when the current SR node in the SRv6 network transmits IPv6 packets to the target SR node, the network routing will automatically assign all the paths to the target SR node, referred to as Segment List, which is stored in the SRv6 message protocol SID;

S112, extract all the SIDs of the Segment List, connect them in series from bottom to top, and use ### as an interval in the middle to generate an initial address identifier;

S113. Before the current SR node transmits to the next-hop SR node, put the initial address identifier into the Argument of the SRH part of the IPv6 extension header.

S12. Construct a k-nearest neighbor algorithm model, an early warning model and generate an intelligent allocation identification module;

As shown in Figure 3, it specifically includes the following steps:

S121. Obtain the failure rate of all next-hop SR nodes associated with the initiating node through the k-nearest neighbor algorithm model, and then predict the alarm probability of all the next-hop SR nodes through the early warning model;

S122. By calculating the failure rate and the alarm probability, determine the optimal next-hop SR node until the optimal SR node of each hop to the destination SR node is obtained and connected in series to form an optimal address identifier; The calculation method of rate and alarm probability is weighted average;

S123. If the optimal address identifies each hop of the SR node is better than the initial address identifies each hop of the SR node, update the Argument initial address identification of the SRH part of the IPv6 extension header.

Described SRv6 protocol structure as shown in Figure 4, adds SRH (Segment Routing Header) extension header in IPv6 routing extension header, and this extension header designates the explicit path of an IPv6, stores the Segment List information of IPv6, and Segment List is to segment and A forwarding path obtained by orderly arrangement of network nodes; the programmability of SRv6 comes from the use of SRv6 SID128bit, SRv6 Segment defines the network instructions in SRv6 network programming, indicates where to go, how to go, and identifies the ID of SRv6 Segment Known as SRv6 SID, SRv6 SID is a 128-bit value in the form of an IPv6 address, consisting of three parts: Locator, Function, and Arguments; SRv6 Segment structure Locator: has a positioning function and provides IPv6 routing capabilities, and packets are implemented through this field Addressing and forwarding. In addition, the route corresponding to the Locator can also be aggregated; Function: used to express the forwarding action to be executed by the device instruction, and different forwarding behaviors are expressed by different Functions; Arguments: optional field, which is for the Function Supplement is the parameter corresponding to the instruction when it is executed. These parameters may contain streams, services or any other related information; each segment of SRv6 is 128bit, which can be flexibly divided into multiple segments, and the function and length of each segment can be customized. It has flexible programming capability, that is, the business can be edited; SRv6 has more powerful network programming capability through the above programming space, which can better meet the needs of different network paths, and perfectly integrates with SDN technology to realize the interaction between the network and applications. A programmable network that can drive business.

SR technology designs two implementations for the data plane, one is SR-MPLS that multiplexes the MPLS data plane, and the other is SRv6; SRv6 uses the IPv6 data plane and is extended based on the IPv6 routing extension header, and this part of the expansion is not damaged Standard IPv6 header, and only SRv6 nodes need to perform additional processing for the extension header, which has no impact on other common IPv6 nodes, which makes SRv6 seamlessly compatible with the existing IPv6 network, and makes the forwarding level reach the extreme of pure IPv6 Simple forwarding; SR-MPLS uses 4-byte labels to identify path information. MPLS labels can only identify three pieces of information: label value, TTL, and label stack bottom, and have no extended information capability. Unlike the Segment of SR MPLS, the Segment ID of SRv6 is called SRv6SID, which is a 128-bit value and is divided into three parts:

Locator (location identifier): An identifier assigned to a network node in the network, which can be used to route and forward data packets. Locator has two important properties, routable and aggregated. In the SRv6 SID, the Locator is a variable-length part, which is used to adapt to networks of different sizes.

Function: An ID value assigned by the device to the local forwarding command, which can be used to express the forwarding action that the device needs to perform, which is equivalent to the operation code of the computer command. In SRv6 network programming, different forwarding behaviors are expressed by different function IDs. To a certain extent, the function ID is similar to the MPLS label, and is used to identify the VPN forwarding instance and so on.

Args (variable): The parameters required by the forwarding instruction during execution, which may contain streams, services or any other related variable information.

In short, SRv6 has both routing and MPLS forwarding attributes. It has TE traffic engineering capabilities, scalability capabilities, and is compatible with IPv6. It is also convenient for future fixed-mobile convergence and unifies IP forwarding technologies.

Further, as shown in Figure 5, step S121 k nearest neighbor algorithm comprises the following steps:

S1211. Prepare data, and arrange the data according to dimensions;

S1212. Calculate the distance from the test sample point to each other sample point;

S1213. Sort each distance, and then select K points with the smallest distance;

S1214. Compare the categories to which the K points belong, and classify the test sample points into the category with the highest proportion among the K points according to the principle that the minority obeys the majority.

The working principle of the k-nearest neighbor algorithm is: there is a sample data set and data labels, and the corresponding relationship between samples and labels is known; data without labels is input, and each feature of the new data is compared with the corresponding features of the data in the sample set; Extract the classification labels of the most similar data in the sample set, and only select the top k most similar data, generally k is less than 20.

Further, the k-nearest neighbor algorithm model formula is:

In the formula: x ₁ x ₂ ... x _n is the n-dimensional data of sample X; y ₁ y ₂ ... y _n is the n-dimensional data of sample Y; X is time, Y is the network delay in milliseconds (ms); d(x, y) is the distance between samples X and Y, and in the present invention is an associated node whose 24-hour delay of the current network node is <= 50 milliseconds. .

The schematic diagram of the k-nearest neighbor algorithm model is shown in Figure 6.

The calculation of the failure rate is illustrated with the following example:

After the current SID node obtains the data calculation of 10 network nodes associated with the current SID node through log analysis, the alarm status is determined by the alarm status of the 10 nodes (0=no alarm, 1=alarm). In this way, the fault prediction of the current sample is completed.

Model calculation results: If the current network node has an alarm status of associated nodes with a delay of milliseconds <= 50 within 24 hours, if 3 out of 10 have alarmed, the network quality failure rate of the current node is 30%.

Further, the early warning model is constructed using a Markov chain, and the formula is:

X(k+1)=X(k)×P

In the formula: X(k) represents the state vector of the trend analysis and prediction object at the time t=k, P represents the one-step transition probability matrix, and X(k+1) represents the state of the trend analysis and prediction object at the time t=k+1 vector.

The calculation of the warning probability is illustrated by the following example:

The historical probability of forwarding by the current node [0.3, 0.7];

Probability of current node transitioning from alarm to normal [0.6, 0.4];

Probability of the current node transitioning from normal to alarm [0.3, 0.7];

Calculated by the model formula:

This priority forwarding probability = 0.3x0.6+0.3x0.7 = 0.39;

This non-priority forwarding probability=0.3x0.4+0.7x0.7=0.61.

From the above embodiments, the overall structure of the present invention can be obtained as shown in FIG. 7 , which creatively highlights the role of artificial intelligence in SRv6 network message forwarding. First, the starting node of the Srv6 network extracts all the SIDs of the Segment List, and connects them in series from bottom to top with ### in the middle to generate an initial address identifier. Secondly, the failure rate of all next-hop SR nodes that initiate node associations is obtained through the k-nearest neighbor algorithm model. Then use the early warning model to predict the alarm probability of all next-hop SR nodes. If the optimal address identifies each hop of SR nodes is better than the initial address identifies each hop of SR nodes. Then, update the initial address identifier of Argument in the SRH part of the IPv6 extension header. Compared with the existing Srv6 message forwarding mechanism, this method adds AI fault prediction and intelligent path selection, provides a new solution for complex service message forwarding in a super-large network, and reduces the resource consumption of message forwarding. At the same time, the accuracy of transmission data and flexible configuration are improved.

The above embodiments are only exemplary embodiments of the present invention, and are not intended to limit the present invention, and the protection scope of the present invention is defined by the claims. Those skilled in the art can make various modifications or equivalent replacements to the present invention within the spirit and protection scope of the present invention, and such modifications or equivalent replacements should also be deemed to fall within the protection scope of the present invention.

Claims (6)

1. a kind of SRv6 combines the message delivery analysis method of k nearest neighbor algorithm, it is characterized in that comprising the following steps: S11, Srv6 network node initial allocation address identifier generation module; S12. Construct a k-nearest neighbor algorithm model, an early warning model and generate an intelligent allocation identification module; Described step S12 comprises the following steps: S121. Obtain the failure rate of all next-hop SR nodes associated with the initiating node through the k-nearest neighbor algorithm model, and then predict the alarm probability of all the next-hop SR nodes through the early warning model; S122. Determine the optimal next-hop SR node by calculating the failure rate and alarm probability, until the optimal SR node for each hop to the destination SR node is obtained and connected in series to form an optimal address identifier; S123. If the optimal address identifies each hop of the SR node is better than the initial address identifies each hop of the SR node, update the Argument initial address identification of the SRH part of the IPv6 extension header. 2. SRv6 according to claim 1 combines the message delivery analysis method of k nearest neighbor algorithm, it is characterized in that: described step S11 comprises the following steps: S111. Obtain all path Segment Lists from the current SR node to the target SR node; S112, extract all the SIDs of the Segment List, connect them in series from bottom to top, and use ### as an interval in the middle to generate an initial address identifier; S113. Before the current SR node transmits to the next-hop SR node, put the initial address identifier into the Argument of the SRH part of the IPv6 extension header. 3. SRv6 according to claim 1 combines the message delivery analysis method of k nearest neighbor algorithm, it is characterized in that: described step S121 k nearest neighbor algorithm comprises the following steps: S1211. Prepare data, and arrange the data according to dimensions; S1212. Calculate the distance from the test sample point to each other sample point; S1213. Sort each distance, and then select K points with the smallest distance; S1214. Compare the categories to which the K points belong, and classify the test sample points into the category with the highest proportion among the K points according to the principle that the minority obeys the majority. 4. SRv6 according to claim 3 combines the message delivery analysis method of k nearest neighbor algorithm, it is characterized in that: described k nearest neighbor algorithm model formula is:

In the formula: x ₁ x ₂ ... x _n is the n-dimensional data of sample X; y ₁ y ₂ ... y _n is the n-dimensional data of sample Y; X is time, Y is the network delay in milliseconds (ms); d(x, y) is the distance between samples X and Y, and in the present invention is an associated node whose 24-hour delay of the current network node is <= 50 milliseconds. 5. SRv6 according to claim 4 combines the message transmission analysis method of k nearest neighbor algorithm, it is characterized in that: described early warning model adopts Markov chain to construct, and formula is: X(k+1)=X(k)×P In the formula: X(k) represents the state vector of the trend analysis and prediction object at the time t=k, P represents the one-step transition probability matrix, and X(k+1) represents the state of the trend analysis and prediction object at the time t=k+1 vector. 6. The message delivery analysis method of SRv6 combined with k-nearest neighbor algorithm according to claim 1, characterized in that: the calculation method of the failure rate and alarm probability in the step S122 is weighted average.

Images