A dual mode traffic flow allocation method

Abstract

The application relates to a dual-mode service flow distribution method, belonging to the field of user information collection systems, and comprising the following steps: S1: different utility functions are designed according to the characteristics of different types of services, and the utility values of each network attribute to different services are calculated; S2: the weight of each network attribute is calculated; S3: a TOPSIS algorithm based on relative entropy is used to calculate the relative closeness of the candidate network; S4: the comprehensive score of the candidate network is calculated by combining the relative closeness and the service preference value, and the network with the highest score is selected for service flow distribution.

Description

Technical Field

[0001] This invention belongs to the field of user information collection systems and relates to a dual-mode service flow allocation method. Background Technology

[0002] In the electricity information collection system, the local communication system uses wired and wireless communication technologies. Wired communication technologies include narrowband power line carrier communication technology and broadband power line carrier communication technology. Wireless communication technologies include 230M wireless communication technology, narrowband low-power wireless communication technology and broadband low-power wireless communication technology. Among them, broadband power line carrier communication technology and broadband low-power wireless communication technology can provide higher transmission bandwidth and faster transmission rates to meet the ever-growing needs of the electricity user information collection system, and have become the mainstream technologies for development and application.

[0003] Broadband Power Line Communication (BPLC) technology, commonly known as BPLC access networks, has been widely used for local communication in low-voltage distribution networks to enable automatic meter reading. Furthermore, BPLC is particularly suitable for communication in IoT and smart home systems, showing great promise for future development. BPLC communication technology can directly utilize power lines without the need for rewiring, resulting in simple and quick network setup, low cost, and a wide range of applications. Figure 1 The image shows a typical network topology provided by the State Grid Corporation of China in its technical specifications for high-speed carrier communication interconnection of low-voltage power lines.

[0004] Micro-power wireless communication is divided into narrowband micro-power wireless communication and broadband micro-power wireless (BMPW) communication. While narrowband micro-power wireless communication technology is mature and commercially available, its channel bandwidth is less than 100kHz, resulting in a low effective communication rate and limited practical applications, currently limiting its use in the State Grid. Correspondingly, to address the shortcomings of narrowband micro-power communication technology, broadband micro-power communication technology has emerged. Currently, several solutions exist for broadband micro-power communication technology under development. One representative solution employs a hybrid high-order modulation technique based on chirp and PSK technologies at the physical layer, and uses a communication protocol similar to high-speed carrier communication technology in the protocol stack, achieving a point-to-point communication rate of up to 3.84Mbps.

[0005] For electricity information collection systems, broadband power line carrier communication and broadband low-power wireless communication have similar network architectures, namely: a tree-like network with a central coordinator (CCO) as the center and proxy coordinators (PCOs) as relay agents, connecting all stations (STAs) in multiple levels, with a maximum network hierarchy of 16 levels.

[0006] The power distribution network has a complex topology, with a large number and variety of power distribution equipment covering a wide area. Furthermore, its application scenarios are complex, business requirements are diverse, and transmission reliability requirements are high. Neither HPLC nor BMP communication alone can fully meet the communication needs of the smart grid. A dual-mode communication system integrating HPLC and BMPW can leverage the complementary advantages of both, eliminate communication blind spots, expand communication coverage, and meet the urgent needs of the smart grid for high-performance, high-reliability communication. This is a key area of ​​research. Service flow allocation is a crucial aspect of researching dual-mode communication systems.

[0007] The current dual-mode system model consists of two independent networks: HPLC and BMPW, forming two separate tree-like networks. Each network has its own characteristics. HPLC offers higher bandwidth and lower latency, but its power line load characteristics significantly impact communication quality and reliability. BMPW, on the other hand, offers greater flexibility in network configuration; as the number of nodes increases, the number of service transmission paths also increases, resulting in higher reliability. This design categorizes services into control services, data acquisition services, event reporting services, and equipment upgrade services. Network attributes such as latency, packet loss rate, bandwidth, and hop count are selected, and service flow allocation under the dual-mode model is performed based on the characteristics of different services. A schematic diagram of the system model is shown below. Figure 2 As shown.

[0008] Current dual-mode service flow allocation methods cannot meet the service quality requirements of various services, ignore the impact of service characteristics and service preferences on the service flow allocation process, have low rationality and accuracy of service flow allocation, and require services to switch between different networks many times. Summary of the Invention

[0009] In view of this, the purpose of this invention is to provide a dual-mode traffic flow allocation method that combines multi-attribute decision-making algorithms and utility theory.

[0010] To achieve the above objectives, the present invention provides the following technical solution:

[0011] A dual-mode service flow allocation method includes the following steps:

[0012] S1: Design different utility functions for different types of services, and calculate the utility value of each network attribute for different services;

[0013] S2: Calculate the weight of each network attribute;

[0014] S3: Use the TOPSIS algorithm based on relative entropy to calculate the relative proximity of candidate networks;

[0015] S4: Combine relative proximity and business preference values ​​to calculate the comprehensive score of the candidate networks, and select the network with the highest score for business flow allocation.

[0016] Furthermore, in step S1, the services are categorized into four types: control, data acquisition, event reporting, and device upgrade; network attributes include latency, packet loss rate, bandwidth, and hop count; among which:

[0017] S11: The delay utility function for all four business types is represented by an sigmoid function:

[0018]

[0019] Where x represents the average network latency, u(x) represents the latency utility value, and a and b represent constants.

[0020] S12: The packet loss rate utility function for the four business types is represented by a linear function:

[0021] u(x) = gx + h

[0022] Where x represents the average packet loss rate of the network, u(x) represents the packet loss rate utility value, and g and h represent constants.

[0023] S13: For the bandwidth utility function of control services and event reporting services, an exponential function is used:

[0024] u(x) = 1 - e -cx or u(x)=e (x)-c

[0025] Where x represents the average network bandwidth, u(x) represents the bandwidth utility value, and c represents a constant.

[0026] For bandwidth utility functions of data acquisition and equipment upgrade services, an sigmoid function is used.

[0027] S14: The hop count utility function for the four business types is represented by a linear function.

[0028] Furthermore, the calculation of the weight of each network attribute in step S2 includes using the Fuzzy-Analytic Hierarchy Process (FAHP) and the CriteriaImportance Through Intercrieria Correlation (CRITIC) algorithm to calculate the subjective and objective weights of the network attributes respectively. Based on modeling by minimizing the deviation between the subjective and objective attribute weights, the comprehensive subjective and objective weights of the network attributes are obtained, and the FAHP algorithm is used to calculate the business's preference for different candidate networks.

[0029] Furthermore, the calculation of subjective weights of network attributes using the fuzzy hierarchical analysis (FAHP) method specifically includes the following steps:

[0030] Step 1: Divide the analysis objects into three layers from bottom to top: the solution layer, the indicator layer, and the target layer. The solution layer includes all candidate networks used for service flow allocation, the indicator layer includes various network attributes, and the target layer refers to the optimal network for service flow allocation.

[0031] Step 2: Compare the importance of each attribute in the indicator layer pairwise according to the business type; use r ij Represents element x i Relative to element x j The degree of importance, and r ij It is also a component of the fuzzy consistency matrix, and the consistency of the matrix is ​​determined by the following formula:

[0032]

[0033] Where n represents the number of network attributes considered, r ii Represents element x i Relative to element x i The degree of importance, r ji Represents element x j Relative to element x i The degree of importance, r ik Represents element x i Relative to element x k The degree of importance, r jk Represents element x j Relative to element x k The degree of importance;

[0034] Step 3: Calculate the subjective weights of each network attribute:

[0035]

[0036] Thus, we obtain the subjective attribute weight vector α = (α1, α2, ..., α) for the n networks.n ).

[0037] Furthermore, the CRITIC algorithm for calculating the correlation of indicators in step S2 calculates the objective weights of network attributes, specifically including the following steps:

[0038] Step 1: Construct the parameter matrix. Assuming there are m candidate networks and n network attributes, construct the decision matrix X = (x ij ) m×n Composition, x ij This represents the j-th attribute value of network i;

[0039]

[0040] Step 2: Standardize parameters using different standardization methods for positive and negative indicators;

[0041] For positive indicators:

[0042]

[0043] For negative indicators:

[0044]

[0045] Step 3: Calculate the standard deviation of the j-th attribute:

[0046]

[0047] Step 4: Calculate the information content of the j-th attribute:

[0048]

[0049] Step 5: Calculate the objective weights of each attribute:

[0050]

[0051] We obtain the objective attribute weight vector β = (β1, β2, ..., β) for n networks. n ).

[0052] Furthermore, the comprehensive weight calculation steps described in step S2 are as follows:

[0053] Assuming there are m candidate networks and n network attributes, construct the decision matrix Y = (y ij ) m×n Composition, y ij This represents the j-th attribute value of network i;

[0054] Let t and τ be the subjective weight coefficient and objective weight coefficient in the combined weights, respectively, where t+τ≥0 and t+τ=1. Then the combined weights are:

[0055] w j =tα j +τβ j (j = 1, 2, ..., n)

[0056] The subjective attribute weight of the j-th attribute of the i-th network is represented as tα. j y ij The objective attribute weight is τβ j y ij Then the distance between their subjective and objective attribute weights is expressed as:

[0057]

[0058] In the formula, Z i This represents the distance between the subjective and objective attribute weights of the i-th network.

[0059] A weight combination optimization model is established with the objective function of minimizing the distance between the subjective and objective attribute weights:

[0060]

[0061] In the formula, minZ represents minimizing the distance between the subjective and objective attribute weights;

[0062] Optimization model solution process:

[0063] Create the Lagrange function:

[0064]

[0065] In the formula, λ represents the Lagrange multiplier, and we get

[0066]

[0067]

[0068]

[0069] Combining the above three equations, we obtain the weighting coefficients t and τ as follows:

[0070]

[0071] Furthermore, step S2, which involves using the FAHP algorithm to calculate the service's preference for different candidate networks, specifically includes the following steps:

[0072] Step 1: Calculate the importance ranking of the indicator layer to the target layer. Based on the calculated weights of the indicator layer with respect to the target layer, obtain the weight relationships between different services and network attributes.

[0073] Step 2: Calculate the importance ranking of the scheme layer to the indicator layer, compare the importance of each candidate network with respect to network attributes, obtain the fuzzy consistency matrix and corresponding weights, and then obtain the weight relationship between different network attributes and candidate networks;

[0074] Step 3: Calculate the importance ranking of the scheme layer with respect to the target layer;

[0075] Multiply the weight relationship matrix between different services and network attributes with the relationship matrix between network attributes and candidate networks to calculate the ranking of the scheme layer with respect to the target layer, that is, the preference value of the service for different candidate networks.

[0076] Furthermore, step S3 specifically includes the following steps:

[0077] S31: Assume w j The weight value of the j-th attribute is used to construct the utility matrix Y = (y ij ) m×n Each column in the vector is associated with its corresponding weight vector w = [w1, w2, ..., w...]. n ] T Multiplying the weights one by one, we get the weighted utility matrix Z = (z ij ) m×n :

[0078]

[0079] S32: Determine the positive ideal network With negative ideal network

[0080]

[0081]

[0082] S33: Calculate the relative entropy distance between each candidate network and the positive and negative ideal networks. The relative entropy distance between a candidate network and the positive ideal network is... The relative entropy distance to the negative ideal network is

[0083]

[0084]

[0085] S34: Calculate relative proximity:

[0086]

[0087] Furthermore, step S4 specifically includes:

[0088] The business preference value T is calculated. iThe comprehensive score R of the candidate network is a combination of business preference value and relative proximity. i :

[0089]

[0090] In the formula, To adjust the factors, the ratio of relative closeness to business preference values ​​is adjusted according to business needs; R i As the criterion for determining the best network for service flow allocation, candidate networks are ranked according to their comprehensive scores, and the network with the largest comprehensive score is selected as the best network for service flow allocation.

[0091] The beneficial effects of this invention are as follows:

[0092] First, the utility function is designed according to the characteristics of different services, taking into account the characteristics and preferences of different services. This allows the services to be allocated to the most suitable communication network, meeting the QoS requirements of various services and improving the overall service quality of the services.

[0093] Second, the attribute weighting method has been optimized. To address the shortcomings of traditional subjective and objective weighting methods, the two methods are combined. Then, the subjective and objective comprehensive weights are calculated by minimizing the distance between the subjective and objective attribute weights, making the weight settings more reasonable.

[0094] Third: Improve the traditional network ranking method. To address the problems of ranking anomalies and low accuracy in the traditional TOPSIS algorithm, relative entropy is used to improve the accuracy of network selection, reduce the number of times services switch between different networks, and reduce the impact of the ping-pong effect.

[0095] Other advantages, objectives, and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination, or may be learned from practice of the invention. The objectives and other advantages of the invention can be realized and obtained through the following description. Attached Figure Description

[0096] To make the objectives, technical solutions, and advantages of the present invention clearer, the preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings, wherein:

[0097] Figure 1 This is a topology diagram of a broadband carrier communication network.

[0098] Figure 2 This is a schematic diagram of the HPLC-BMPW dual-mode system model;

[0099] Figure 3 This is a flowchart of the dual-mode service flow allocation method described in this invention;

[0100] Figure 4 This is a FAHP hierarchical structure diagram. Detailed Implementation

[0101] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0102] The accompanying drawings are for illustrative purposes only and are schematic diagrams, not actual pictures. They should not be construed as limiting the invention. To better illustrate the embodiments of the invention, some parts in the drawings may be omitted, enlarged, or reduced, and do not represent the actual product dimensions. It is understandable to those skilled in the art that some well-known structures and their descriptions may be omitted in the drawings.

[0103] In the accompanying drawings of the embodiments of the present invention, the same or similar reference numerals correspond to the same or similar components. In the description of the present invention, it should be understood that if terms such as "upper," "lower," "left," "right," "front," and "rear" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, they are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, the terms used to describe positional relationships in the drawings are only for illustrative purposes and should not be construed as limiting the present invention. For those skilled in the art, the specific meaning of the above terms can be understood according to the specific circumstances.

[0104] Based on practical engineering needs, this invention proposes a dual-mode traffic flow allocation method that combines multi-attribute decision-making algorithms and utility theory. A schematic diagram of the traffic flow allocation method is shown below. Figure 3 As shown.

[0105] First, the services are divided into four types: control, data acquisition, event reporting, and equipment upgrade. Different utility functions are designed for the characteristics of different service types, and the utility value of each network attribute for different services is calculated, fully considering the characteristics and QoS requirements of different services. For different service characteristics, appropriate utility functions are designed to quantify the service's satisfaction with network attributes. Commonly used utility function types include linear functions, exponential functions, logarithmic functions, and sigmoid functions, as shown in formulas (1) to (5).

[0106] S-shaped function:

[0107]

[0108] Exponential function:

[0109] u(x) = 1 - e -cx or u(x)=e (x)-c (2)

[0110] Logarithmic function:

[0111] u(x)=d+eln(x+f) (3)

[0112] Linear functions:

[0113] u(x)=gx+h (4)

[0114] Linear piecewise functions:

[0115]

[0116] Delay utility function: First, control-related services are real-time services with strict latency requirements. Failure to complete the service within the specified time limit will lead to system failure and crash. Second, event reporting and equipment upgrade services have relatively relaxed latency requirements, do not require strict time synchronization, and have a wider range of latency variations. Finally, data acquisition services have the least relaxed latency requirements among the four types of services. Based on the above analysis, the delay utility function adopts an S-shaped function, and the delay utility function settings for various services are shown in Table 1.

[0117] Table 1

[0118]

[0119] Packet loss rate utility function: Packet loss rate refers to the proportion of data packets lost during transmission out of the total number of data packets. Control services, data acquisition services, event reporting services, and equipment upgrade services all have a certain tolerance for packet loss rate, but satisfaction decreases continuously as the packet loss rate increases. Since these four services have different tolerances for packet loss rate, it is assumed that event reporting services reach their lowest satisfaction level when the packet loss rate exceeds 1%, control and data acquisition services reach their lowest satisfaction level when the packet loss rate exceeds 5%, and equipment upgrade services reach their lowest satisfaction level when the packet loss rate exceeds 10%. Therefore, this invention uses a linear function to represent the utility function of packet loss rate. The packet loss rate utility function settings for various services are shown in Table 2.

[0120] Table 2

[0121]

[0122] Bandwidth utility function: Equipment upgrade services require significant bandwidth due to the large data volume, while data acquisition services have relatively smaller data volumes and less stringent bandwidth requirements. Both types of services need to meet a minimum bandwidth threshold, with the threshold for data acquisition services being lower than that for equipment upgrade services. Therefore, an sigmoid function is suitable as the utility function for these two types of services. Control and event reporting services do not require a minimum bandwidth threshold, and their utility value is positively correlated with bandwidth. This invention uses an exponential function as the utility function for these two types of services. Assume that the bandwidth requirement for equipment upgrade services is 1 Mbps and the bandwidth requirement for data acquisition services is 0.5 Mbps. The bandwidth utility function settings for various services are shown in Table 3.

[0123] Table 3

[0124]

[0125] Hop Count Utility Function: The hop count requirements are consistent for four types of services: control services, data acquisition services, event reporting services, and equipment upgrade services. Service satisfaction decreases as the hop count increases. The project stipulates that the maximum number of hops per node cannot exceed 15. Therefore, this invention sets the utility value to its lowest point when the hop count exceeds 15. A linear function is used as the hop count utility function, and the hop count utility function settings for various services are shown in Table 4.

[0126] Table 4

[0127]

[0128] Then, this invention improves upon traditional attribute weighting methods by proposing a comprehensive weighting method that combines subjective and objective weights. Subjective weighting methods determine network attribute weights based on various business needs and the subjective experience of decision-makers, but they neglect the actual attribute characteristics of the network. Objective weighting methods, while accurately reflecting the characteristics of network attributes, fail to accurately reflect the decision-makers' preferences for attributes. This invention first employs Fuzzy-Analytic Hierarchy Process (FAHP) and Criteria Importance Through Intercrieria Correlation (CRITIC) algorithms to calculate the subjective and objective weights of network attributes, respectively. Based on minimizing the deviation between subjective and objective attribute weights, a comprehensive subjective and objective weight of the network attribute is obtained. Finally, the FAHP algorithm is used to calculate the business's preference for different candidate networks.

[0129] FAHP Calculation of Subjective Weights: The FAHP algorithm is an improvement over the AHP algorithm, especially in terms of consistency. FAHP is based on a consistent pairwise comparison matrix, and the consistency of the matrix is ​​guaranteed during matrix construction. This section uses the FAHP algorithm based on a fuzzy consistent matrix to calculate the subjective weights of network attributes. The specific steps are as follows:

[0130] Step 1: To clarify the relationships between various network attributes, the analysis object needs to be divided into three layers from bottom to top: the solution layer, the indicator layer, and the target layer. The solution layer includes all candidate networks used for service flow allocation; the indicator layer includes various network attributes; and the target layer refers to the optimal network for service flow allocation. The hierarchical structure diagram is as follows: Figure 4 As shown.

[0131] Step 2: Since different communication services have different requirements for network attributes, it is necessary to compare the importance of each attribute in the indicator layer pairwise according to the service type. ij Represents element x i Relative to element x j The degree of importance, and r ij It is also a component of the fuzzy consistent matrix, and the importance scale is shown in Table 5. Based on the property that the difference between corresponding elements of any two rows of the fuzzy consistent matrix is ​​a constant, and that the difference between any row and any other row of the matrix is ​​also a constant, the consistency of the matrix can be judged by formula (6).

[0132]

[0133] Step 3: Calculate the subjective weights of each network attribute using formula (7).

[0134]

[0135] Thus, we obtain the subjective attribute weight vector α = (α1, α2, ..., α) for the n networks. n ).

[0136] Table 5

[0137]

[0138] Based on the characteristics and QoS requirements of the four service types, the present invention sets fuzzy consistency matrices for the four services as shown in Tables 6 to 9, and calculates the attribute weights.

[0139] Table 6

[0140]

[0141] Table 7

[0142]

[0143]

[0144] Table 8

[0145]

[0146] Table 9

[0147]

[0148] CRITIC algorithm calculates objective weights: Since the FAHP algorithm inherently relies heavily on subjective experience and judgment, it cannot obtain objective and accurate weights for network attributes. Therefore, the CRITIC algorithm is introduced to calculate objective weights for network attributes without considering subjective opinions, thus avoiding overly subjective weight settings. The main steps are as follows:

[0149] Step 1: Construct the parameter matrix

[0150] Assuming there are m candidate networks and n network attributes, construct the decision matrix X = (x ij ) m×n Composition, x ij This represents the j-th attribute value of network i.

[0151]

[0152] Step 2: Parameter Standardization

[0153] Since the dimensions of each network attribute parameter are different, they need to be standardized to facilitate subsequent calculations. Different standardization methods are adopted for positive and negative indices.

[0154] For positive indicators,

[0155]

[0156] For negative indicators,

[0157]

[0158] Step 3: Calculate the standard deviation of the j-th attribute

[0159]

[0160] Step 4: Calculate the information content of the j-th attribute

[0161] For ease of subsequent calculations, the standardized parameters will still be represented by x. ij express.

[0162]

[0163] Step 5: Calculate the objective weights of each attribute.

[0164]

[0165] We obtain the objective attribute weight vector β = (β1, β2, ..., β) for n networks. n ).

[0166] Calculate the overall weight: The overall weight is calculated by minimizing the distance between the subjective and objective attribute weights. The main steps are as follows:

[0167] Assuming there are m candidate networks and n network attributes, construct the decision matrix Y = (y ij ) m×n Composition, y ij This represents the j-th attribute value of network i.

[0168] Let t and τ be the subjective weight coefficient and objective weight coefficient in the combined weights, respectively, where t+τ≥0 and t+τ=1. Then the combined weights are:

[0169] w j =tα j +τβ j (j=1,2,…,n) (13)

[0170] The subjective attribute weight of the j-th attribute of the i-th network is represented as tα. j y ij The objective attribute weight is τβ j y ij Then the distance between their subjective and objective attribute weights can be expressed as:

[0171]

[0172] In the formula, Z i Z represents the distance between the subjective and objective attribute weights of the i-th network. Clearly, Z... i The smaller the distance, the more consistent the subjective and objective attribute weights of evaluation object i become. A weight combination optimization model is established with minimizing the distance between the subjective and objective attribute weights as the objective function:

[0173]

[0174] In the formula, minZ represents minimizing the distance between the subjective and objective attribute weights. The following is the solution process for the optimization model.

[0175] Create the Lagrange function:

[0176]

[0177] In the formula, λ represents the Lagrange multiplier, and we get

[0178]

[0179]

[0180]

[0181] Combining the above three equations, we obtain the weighting coefficients t and τ as follows:

[0182]

[0183] This invention considers the degree of preference of different services for candidate networks during the service flow allocation process and uses FAHP to calculate service preferences. The main steps are as follows:

[0184] Step 1: Calculate the importance ranking of the indicator layer to the target layer.

[0185] Based on the calculated weights of the indicator layer with respect to the target layer, the weight relationships between different services and network attributes can be obtained as shown in Table 10.

[0186] Table 10

[0187]

[0188] Step 2: Rank the importance of the scheme layer to the indicator layer.

[0189] The importance of each candidate network with respect to network attributes is compared to obtain a fuzzy consistency matrix and corresponding weights. Then, the weight relationship between different network attributes and candidate networks is obtained. The results are shown in Tables 11 and 15.

[0190] Table 11

[0191]

[0192] Table 12

[0193]

[0194] Table 13

[0195]

[0196] Table 14

[0197]

[0198] Table 15

[0199]

[0200] To calculate the importance ranking of the scheme layer with respect to the target layer, multiply the matrices in Table 10 and Table 15 from step 1 to calculate the ranking of the scheme layer with respect to the target layer. That is, the business's preference values ​​for different candidate networks are shown in Table 16.

[0201] Table 16

[0202]

[0203] The aforementioned schemes calculate the utility function, attribute weights, and business preference values ​​respectively, in preparation for the implementation of the dual-mode business flow allocation method.

[0204] The specific implementation steps of this invention are as follows:

[0205] 1. Assuming there are m candidate networks and n network attributes, construct the decision matrix X = (x ij ) m×n Composition, x ij This represents the j-th attribute value of network i.

[0206] 2. Determine the utility functions for different attributes of different business types.

[0207] 3. Construct the utility matrix

[0208] Substitute the attribute values ​​from the decision matrix in step 1 into the utility function determined in step 2, and obtain the utility matrix Y = (y ij ) m×n .

[0209] y ij =y j (x ij ) (twenty one)

[0210] In the formula, y j (x) is the utility function corresponding to attribute j.

[0211] 4. Calculate the weighted utility matrix

[0212] Assume w j The value of the j-th attribute is the weight value of the utility matrix Y = (y_j) obtained in step 3. ij ) m×n Each column in the vector is associated with its corresponding weight vector w = [w1, w2, ..., w...]. n ] T The weights in the matrix are multiplied one by one, and the weighted utility matrix Z = (z) is obtained by formula (22). ij ) m×n .

[0213]

[0214] 5. Determine the positive ideal network With negative ideal network

[0215]

[0216]

[0217] 6. Calculate the relative entropy distance between each candidate network and the positive and negative ideal networks.

[0218] The TOPSIS algorithm accurately reflects the differences between various candidate networks and is a classic ranking algorithm. By calculating the Euclidean distance between each candidate network and the positive and negative ideal networks, the candidate network that is closest to the positive ideal network and furthest from the negative ideal network is the optimal network. However, the TOPSIS algorithm is not without its drawbacks. Because it uses Euclidean distance, ranking anomalies can occur when candidate solutions are equidistant from both the positive and negative ideal networks. To improve this problem, this invention uses relative entropy to calculate the distance between candidate networks.

[0219] The relative entropy distance between the candidate network and the positive ideal network is The relative entropy distance to the negative ideal network is

[0220]

[0221]

[0222] 7. Calculate the relative closeness

[0223]

[0224] 8. Rank the candidate networks by combining relative relevance and business preference values.

[0225] The business preference value T is calculated using formula (28). i The comprehensive score R of the candidate network is a combination of business preference value and relative proximity. i .

[0226]

[0227] In the formula, To adjust the factors, the ratio of relative closeness to business preference values ​​is adjusted according to business needs. In this paper... Take 0.5, R i As the criterion for determining the best network for service flow allocation, candidate networks are ranked according to their comprehensive scores, and the network with the largest comprehensive score is selected as the best network for service flow allocation.

[0228] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A dual-mode service flow allocation method, characterized in that: Includes the following steps: S1: Design different utility functions for different types of services, and calculate the utility value of each network attribute for different services; S2: Calculate the weight of each network attribute; Step S2 involves calculating the weight of each network attribute by using the fuzzy hierarchical analysis method (FAHP) and the CRITIC algorithm for the correlation of indicators to calculate the subjective and objective weights of the network attribute respectively, modeling based on minimizing the deviation of the subjective and objective attribute weights, obtaining the comprehensive subjective and objective weights of the network attribute, and using the FAHP algorithm to calculate the service's preference for different candidate networks. S3: Calculate the relative proximity of candidate networks using the TOPSIS algorithm improved based on relative entropy; Step S3 specifically includes the following steps: S31: Assuming there is a total A candidate network, A network attribute, It is the weight value of the j-th attribute. Represents network The Each attribute value will affect the utility matrix. Each column in the data and its corresponding weight vector Multiply the weights one by one to obtain the weighted utility matrix. : in ; S32: Determine the positive ideal network With negative ideal network S33: Calculate the relative entropy distance between each candidate network and the positive and negative ideal networks. The relative entropy distance between a candidate network and the positive ideal network is... The relative entropy distance to the negative ideal network is ; S34: Calculate relative proximity: in ; S4: Combine relative proximity and business preference values ​​to calculate the comprehensive score of the candidate networks, and select the network with the highest score for business flow allocation.

2. The dual-mode service flow allocation method according to claim 1, characterized in that: In step S1, services are categorized into four types: control, data acquisition, event reporting, and device upgrade. Network attributes include latency, packet loss rate, bandwidth, and hop count. S11: The delay utility function for all four business types is represented by an sigmoid function: in, Indicates the average network latency. Indicates the time delay utility value. , Represents a constant; S12: The packet loss rate utility function for the four business types is represented by a linear function: in, This represents the average packet loss rate of the network. This represents the utility value of packet loss rate. , Represents a constant; S13: For the bandwidth utility function of control services and event reporting services, an exponential function is used: in, Indicates the average network bandwidth. Indicates bandwidth utility value, Represents a constant; For bandwidth utility functions of data acquisition and equipment upgrade services, an sigmoid function is used. S14: The hop count utility function for the four business types is represented by a linear function.

3. The dual-mode service flow allocation method according to claim 1, characterized in that: The calculation of subjective weights for network attributes using the Fuzzy Hierarchical Analysis (FAHP) method specifically includes the following steps: Step 1: Divide the analysis objects into three layers from bottom to top: the solution layer, the indicator layer, and the target layer. The solution layer includes all candidate networks used for service flow allocation, the indicator layer includes various network attributes, and the target layer refers to the optimal network for service flow allocation. Step 2: Compare the importance of each attribute in the indicator layer pairwise according to the business type; use Represents element Relative to element The degree of importance, at the same time It is also a component of the fuzzy consistency matrix, and the consistency of the matrix is ​​determined by the following formula: in, , n This indicates the number of network attributes considered. Represents element Relative to element The degree of importance, Represents element Relative to element The degree of importance, Represents element Relative to element The degree of importance, Represents element Relative to element The degree of importance; Step 3: Calculate the subjective weights of each network attribute: Therefore, we obtain The subjective attribute weight vector of each network .

4. The dual-mode service flow allocation method according to claim 1, characterized in that: Step S2 describes the CRITIC algorithm for calculating the objective weights of network attributes based on the correlation of indicators. This algorithm specifically includes the following steps: Step 1: Construct the parameter matrix, assuming there are a total of A candidate network, Construct a decision matrix based on network attributes. composition, Represents network The Each attribute value; Step 2: Standardize parameters using different standardization methods for positive and negative indicators; For positive indicators: For negative indicators: Step 3: Calculate the standard deviation of the j-th attribute: Step 4: Calculate the information content of the j-th attribute: Step 5: Calculate the objective weights of each attribute: Obtain the objective attribute weight vectors of n networks .

5. The dual-mode service flow allocation method according to claim 1, characterized in that: The comprehensive weight calculation steps described in step S2 are as follows: Assuming there is a total A candidate network, Construct a decision matrix based on network attributes. composition, Represents network The Each attribute value; Define the subjective weight coefficient and the objective weight coefficient in the combined weight as follows: and ,in Then the combined weights are: in For the first The network's first The subjective attribute weights of each attribute are expressed as follows: The objective attribute weight is Then the distance between their subjective and objective attribute weights is expressed as: In the formula, , Indicates the first Distance between the weights of the network's subjective and objective attributes; A weight combination optimization model is established with the objective function of minimizing the distance between the subjective and objective attribute weights: In the formula, minZ represents minimizing the distance between the subjective and objective attribute weights; Optimization model solution process: Create the Lagrange function: In the formula, Representing the Lagrange multipliers, we obtain By combining the three equations above, we obtain the weighting coefficients. and They are respectively: 。 6. The dual-mode service flow allocation method according to claim 1, characterized in that: Step S2, which involves using the FAHP algorithm to calculate the service's preference for different candidate networks, specifically includes the following steps: Step 1: Calculate the importance ranking of the indicator layer to the target layer. Based on the calculated weights of the indicator layer with respect to the target layer, obtain the weight relationships between different services and network attributes. Step 2: Calculate the importance ranking of the scheme layer to the indicator layer, compare the importance of each candidate network with respect to network attributes, obtain the fuzzy consistency matrix and corresponding weights, and then obtain the weight relationship between different network attributes and candidate networks; Step 3: Calculate the importance ranking of the scheme layer with respect to the target layer; Multiply the weight relationship matrix between different services and network attributes with the relationship matrix between network attributes and candidate networks to calculate the ranking of the scheme layer with respect to the target layer, that is, the preference value of the service for different candidate networks.

7. The dual-mode service flow allocation method according to claim 1, characterized in that: Step S4 specifically includes: Calculate the business preference value The comprehensive score of the candidate network is calculated by combining the business preference value and the relative proximity. : In the formula, , To adjust the factors, the ratio of relative proximity to business preference value is adjusted according to business needs; As the criterion for determining the best network for service flow allocation, candidate networks are ranked according to their comprehensive scores, and the network with the largest comprehensive score is selected as the best network for service flow allocation.

Images