CN107608803B - A Social D2D Relay Selection Method - Google Patents

Abstract

The invention provides a social D2D relay selection method. The method preprocesses input data to obtain required types of data; generates and calculates indicators, ie target decision weights, on the obtained data; The QoS conditions are transformed and integrated, and then solved through the distributed message passing mechanism to obtain the final output result. Based on the combination of the modified IFAHP method and the entropy weight method, the algorithm converts the multi-objective optimization problem into a single-objective optimization problem, and further converts the expression of the service quality condition, and then solves the D2D UE distributedly through a suitable message passing algorithm. ‑NW relay selection result. The present invention combines the subjective preference of VUE with hesitation degree and the objective information decision weight for consideration, so that the relay selection result obtains a more comprehensive performance improvement, and has higher equipment access ratio, throughput and system fairness.

Description

A Social D2D Relay Selection Method

technical field

The present invention relates to the field of mobile communications, and more particularly, to a social D2D relay selection method.

Background technique

The D2D communication technology enables direct communication between user terminals (UEs) without the need for transmission or forwarding through equipment such as base stations (eNBs), thereby reducing eNB load and expanding communication coverage.

UE-NW (UE-to-Network) relay is a new feature introduced by the 3GPP LTE standard setting group in the D2D communication topic. It has the advantage of flexible deployment and can expand network coverage without adding existing network equipment. Therefore, it can be widely used in commercial communication, public safety communication (such as earthquake, war) and other fields. As shown in FIG. 1 , a typical D2DUE-NW system includes one eNB, several relay user equipments Relay UEs (RUEs) and several Victim UEs (VUEs, ie UEs requiring D2D relay connection services). The eNB and the RUE are connected through a cellular communication link, and the RUE and the VUE are connected through a secondary link dedicated to D2D communication specified by 3GPP. VUE can be further divided into two types: In-Coverage VUE (IC VUE) within the coverage of the cellular network, and Out-of-Coverage VUE (OOC VUE) outside the coverage of the cellular network.

Although the D2D UE-NW relay communication technology has the above advantages, there is still a problem in the existing solution, that is, how to effectively perform the D2D relay selection function on the basis of satisfying the conclusion of the existing 3GPP regulations. In this problem, a key question is what standard and method should be used to perform D2D relay selection under multi-objective optimization. In the conclusion formed by the 3GPP RAN2 working group in 2015, the following requirements were put forward: "The relay UE selected by the VUE must meet the best quality of the link and the conditions specified by other high layers at the same time".

At the same time, with the development of social network technology, D2D UEs are increasingly showing the social attributes of holders. A higher social similarity between UEs often means a higher degree of trust, that is, a higher degree of connection security. In addition, high social similarity often also means that VUE is more likely to obtain interesting data content from socially similar UEs. Therefore, the social connectivity properties of D2D have gained more and more extensive research attention.

However, the existing technical methods have various design defects. A typical solution that focuses on physical link performance, in the prior art, by iteratively updating the maximum reachable broadcast rate of the D2D link in the cluster, the clustering with the minimum time-frequency resource consumption under the reliable retransmission mechanism and the corresponding clustering are obtained. head. However, this scheme ignores the social connectivity properties presented by D2D devices and has excessive computational complexity. In the prior art, D2D relay selection schemes that combine social and physical connection attributes have been proposed, but these schemes are all performed under the assumption that social and physical connection attributes have the same processing priority, which may not be in line with practical applications. Scenes. On the other hand, these methods focus on the optimization scheme under a single objective, and do not consider the situation of multi-objective optimization, which will make the improvement of the access performance of the UE-NW system relatively simple.

In addition to the above single optimization objective solution, a multi-objective D2D based on the combination of Complementary Fuzzy Analytic Hierarchy Process (CFAHP) and Mahalanobis Distance (MD) has also been proposed in the prior art. Trunk selection scheme. However, the multi-objective in this scheme does not contain social attributes. Furthermore, this scheme relies on the subjective judgment brought by CFAHP, ignoring the favorable decision-making information for relay selection that can be provided by objective data. At the same time, the implementation of the CFAHP method does not consider the psychological characteristics of hesitation in UE's subjective fuzzy judgment. On the other hand, the method fails to distinguish between different processing strategies for gain-type and loss-type indicators.

SUMMARY OF THE INVENTION

The present invention provides a social D2D relay selection method that improves the probability of acquiring target type data.

In order to achieve above-mentioned technical effect, technical scheme of the present invention is as follows:

A social D2D relay selection method, comprising the following steps:

S1: Preprocess the input data to obtain the required type of data;

S2: Generate and calculate the indicator, that is, the target decision weight, for the data in S1;

S3: Convert and integrate the single-target output obtained in S2 with the QoS condition, and then solve it through a distributed message passing mechanism to obtain the final output result.

Further, the specific process of the step S1 is:

Let the relay user equipment RUE form a set R={R ₁ ,R ₂ ,...,R _N }, where N is the number of candidate relay user equipment RUEs; the relay connection service user equipment VUE form a set V={ v ₁ ,v ₂ ,...,v _M }, M is the number of VUEs;

The indicators that this method needs to generate and calculate the decision weights are:

Link capacity between VUE v and RUE r

The social similarity S _{v,r between VUE v and RUE r} : described by the Jaccard coefficient, defined as the ratio of the common social attributes between VUE v and RUE r to the total social attributes;

The cache size β _r of the RUE r side;

The consumption C _{v,r required by VUE v to obtain the relay service of RUE r} : in the IC scenario, the cost required for the VUE to motivate the RUE to perform the relay service; in the OOC scenario, it is the power consumption of the VUE itself;

其中，容量、社交相似度以及RUE端缓存大小为增益型指标，数值越高则越优；另一方面，消耗则为损耗型指标，数值越低则越优；The above multi-targets constitute the target set of the VUE v-side

Among them, capacity, social similarity and RUE side cache size are gain-type indicators, and the higher the value, the better; on the other hand, consumption is a loss-type indicator, and the lower the value, the better;

At the same time, a binary selection variable X _v,r is defined to indicate whether VUE v selects candidate RUE r:

In addition to the above multi-objectives, the indicators that need to be calculated for decision weight generation are:

The QoS conditions required by the VUE side for each indicator;

The acceptance capability of the RUE side, that is, the maximum number of accessible VUEs K _r ;

Each VUE can only access only one RUE;

Based on the above content, the optimization model of the demand solution of this method is as follows:

max{P1,P2,P3,-P4} (2)

The model is limited by:

in:

S_v,thresh,β_v,thresh,C_v,thresh分别为VUE v端的容量QoS阈值，社交相似度QoS阈值，缓存QoS阈值，消耗QoS阈值。

S _v,thresh ,β _v,thresh ,C _v,thresh are the capacity QoS threshold, social similarity QoS threshold, cache QoS threshold, and consumption QoS threshold of VUE v, respectively.

Further, the specific process of step S2 includes generating subjective preference decision weights and generating objective decision weights;

The generation of the subjective preference decision weight includes the following steps:

1), construct an intuitive fuzzy preference relationship;

2), construct a perfect multiplicative consistent intuitive fuzzy relationship matrix;

3), generate intuitive fuzzy numerical weights;

4), generate a certain number of sorting values

即VUE v端的第i个指标所对应的主观偏好权重；5), normalize the sorting value to generate the output subjective preference weight

That is, the subjective preference weight corresponding to the i-th indicator on the VUE v end;

The objective decision weight generation includes the following steps:

1), input the index values of all candidate RUEs;

2), data preprocessing;

3) Entropy weight method to obtain objective decision weight;

4) Integrating the results obtained by the subjective preference decision weight generation and processing to generate the single-target relative approximation sequence Τv of the output VUE _v end.

Further, the specific process of the step S3 is:

Through Τ _v , the original optimization problem model is converted into a new optimization model with the smallest sum of RPD of single objective relative approximation:

Limited by: C1～C6 (6)

Since this optimization model including QoS threshold conditions cannot be directly solved by the existing distributed message passing algorithm, to further solve the optimization problem (5), the following indicators are introduced:

Further, combining the objectives of (7) and (4), the following transformation model is obtained:

limited by:

Among them, C7 is a new condition introduced to ensure that (8) and (5) are mathematically equivalent. For model (8), the objective function of (5) has been combined with the QoS threshold condition, so that (8) The solution is equivalent to solving the relay selection result that can satisfy the minimum RPD and all QoS threshold conditions at the same time, because in the actual communication system, the condition C7 is difficult to be solved in the binary selection problem under this kind of QoS constraints. Fully satisfied, in order to apply the distributed message passing mechanism to solve the model, omit C7 to solve the expected output result of (8). In order to solve (8) with the distributed message passing mechanism, redefine the t-th iteration, RUE The message from the end to the VUE end is

And the message from the VUE side to the RUE side is

表示VUE的组成集合V删除第v个VUE的剩余子集(即V/{v})中的第K_r个最小消息量，且0＜K_r≤M,K_r∈Z⁺为预定义的参数；Among them, ω is a predefined damping coefficient to ensure the convergence of the algorithm,

Denotes that the constituent set V of VUE deletes the K _rth smallest message volume in the remaining subset of the vth VUE (ie V/{v}), and 0<K _r ≤M, K _r ∈ Z ⁺ is predefined parameter;

Since the solution target type of (8) is to minimize the optimal summation, correspondingly, the present invention sets the solution target of the distributed message passing mechanism to minimize the sum of the message volume, and obtain the final iterative output:

Compared with the prior art, the beneficial effects of the technical solution of the present invention are:

The method of the invention preprocesses the input data to obtain the required type of data; performs the index, ie the target decision weight generation calculation, on the obtained data; transforms and integrates the obtained single target output and the service quality QoS condition, and then distributes the message through the distributed message. The transfer mechanism is solved to obtain the final output result. Based on the combination of Modified IFAHP (MIFAHP) and entropy weight method, the algorithm converts the multi-objective optimization problem into a single-objective optimization problem, and further converts the expression of the quality of service (QoS) condition, and then The D2D UE-NW relay selection result is solved distributedly through a suitable message passing algorithm. The present invention considers the subjective preference of the VUE with hesitation degree in combination with the objective information decision weight, so that the D2D UE-NW relay selection result can obtain a more comprehensive performance improvement. At the same time, compared with the existing solutions, the present invention has higher device access ratio, higher throughput, and higher system fairness, and thus is more suitable for the actual D2D UE-NW communication system.

Description of drawings

Figure 1 is a typical D2D UE-NW system;

Fig. 2 is the basic flow of the present invention;

Fig. 3 is based on the modified IFAHP method (Modified IFAHP, MIFAHP) flow chart;

Fig. 4 is the target weight generation algorithm flow chart of joint IFAHP method and entropy weight method;

Fig. 5 is the flow chart of solving through distributed message passing mechanism;

Figure 6 shows the access ratio vs capacity QoS threshold in the IC scenario;

Figure 7 shows the access ratio of the IC scene vs the social similarity QoS threshold;

Figure 8 shows the access ratio vs consumption QoS threshold in the IC scenario;

Figure 9 shows the access ratio of the IC scenario vs the cache size QoS threshold;

Figure 10 shows the access ratio in the IC scenario vs the full QoS threshold;

Figure 11 shows the access ratio vs capacity QoS threshold in the OOC scenario;

Figure 12 is the OOC scenario access ratio vs social similarity QoS threshold;

Figure 13 shows the access ratio vs consumption QoS threshold in the OOC scenario;

Figure 14 shows the access ratio in the OOC scenario vs the cache QoS threshold;

Figure 15 shows the access ratio in the OOC scenario vs the full QoS threshold;

Figure 16 is the IC scenario throughput vs full QoS threshold;

Figure 17 shows system fairness vs full QoS threshold in IC scenario;

Figure 18 is an OOC scenario throughput vs full QoS threshold;

Figure 19 shows the system fairness vs full QoS threshold in the OOC scenario.

Detailed ways

The accompanying drawings are for illustrative purposes only, and should not be construed as limitations on this patent;

In order to better illustrate this embodiment, some parts of the drawings are omitted, enlarged or reduced, which do not represent the size of the actual product;

It will be understood by those skilled in the art that some well-known structures and their descriptions may be omitted from the drawings.

The technical solutions of the present invention will be further described below with reference to the accompanying drawings and embodiments.

Example 1

A social D2D relay selection method, comprising the following steps:

S1: Preprocess the input data to obtain the required type of data;

S2: Generate and calculate the indicator, that is, the target decision weight, for the data in S1;

S3: Convert and integrate the single-target output obtained in S2 with the QoS condition, and then solve it through a distributed message passing mechanism to obtain the final output result.

Further, the specific process of the step S1 is:

Let the relay user equipment RUE form a set R={R ₁ ,R ₂ ,...,R _N }, where N is the number of candidate relay user equipment RUEs; the relay connection service user equipment VUE form a set V={ v ₁ ,v ₂ ,...,v _M }, M is the number of VUEs;

The indicators that this method needs to generate and calculate the decision weights are:

Link capacity between VUE v and RUE r

The social similarity S _{v,r between VUE v and RUE r} : described by the Jaccard coefficient, defined as the ratio of the common social attributes between VUE v and RUE r to the total social attributes;

The cache size β _r of the RUE r side;

The consumption C _{v,r required by VUE v to obtain the relay service of RUE r} : in the IC scenario, the cost required for the VUE to motivate the RUE to perform the relay service; in the OOC scenario, it is the power consumption of the VUE itself;

其中，容量、社交相似度以及RUE端缓存大小为增益型指标，数值越高则越优；另一方面，消耗则为损耗型指标，数值越低则越优；The above multi-targets constitute the target set of the VUE v-side

Among them, capacity, social similarity and RUE side cache size are gain-type indicators, and the higher the value, the better; on the other hand, consumption is a loss-type indicator, and the lower the value, the better;

At the same time, a binary selection variable X _v,r is defined to indicate whether VUE v selects candidate RUE r:

In addition to the above multi-objectives, the indicators that need to be calculated for decision weight generation are:

The QoS conditions required by the VUE side for each indicator;

The acceptance capability of the RUE side, that is, the maximum number of accessible VUEs K _r ;

Each VUE can only access only one RUE;

Based on the above content, the optimization model of the demand solution of this method is as follows:

max{P1,P2,P3,-P4} (2)

The model is limited by:

in:

S_v,thresh,β_v,thresh,C_v,thresh分别为VUE v端的容量QoS阈值，社交相似度QoS阈值，缓存QoS阈值，消耗QoS阈值。

S _v,thresh ,β _v,thresh ,C _v,thresh are the capacity QoS threshold, social similarity QoS threshold, cache QoS threshold, and consumption QoS threshold of VUE v, respectively.

Further, the specific process of step S2 includes generating subjective preference decision weights and generating objective decision weights;

The generation of the subjective preference decision weight includes the following steps:

1), construct an intuitive fuzzy preference relationship;

2), construct a perfect multiplicative consistent intuitive fuzzy relationship matrix;

3), generate intuitive fuzzy numerical weights;

4), generate a certain number of sorting values

即VUE v端的第i个指标所对应的主观偏好权重；5), normalize the sorting value to generate the output subjective preference weight

That is, the subjective preference weight corresponding to the i-th indicator on the VUE v end;

The objective decision weight generation includes the following steps:

1), enter the index values of all candidate RUEs;

2), data preprocessing;

3) Entropy weight method to obtain objective decision weight;

4) Integrating the results obtained by the subjective preference decision weight generation and processing to generate the single-target relative approximation sequence Τv of the output VUE _v end.

Further, the specific process of the step S3 is:

Through Τ _v , the original optimization problem model is converted into a new optimization model with the smallest sum of RPD of single objective relative approximation:

Limited by: C1～C6 (6)

Since this optimization model including QoS threshold conditions cannot be directly solved by the existing distributed message passing algorithm, to further solve the optimization problem (5), the following indicators are introduced:

Further, combining the objectives of (7) and (4), the following transformation model is obtained:

limited by:

Among them, C7 is a new condition introduced to ensure that (8) and (5) are mathematically equivalent. For model (8), the objective function of (5) has been combined with the QoS threshold condition, so that (8) The solution is equivalent to solving the relay selection result that can satisfy the minimum RPD and all QoS threshold conditions at the same time, because in the actual communication system, the condition C7 is difficult to be solved in the binary selection problem under this kind of QoS constraints. Fully satisfied, in order to apply the distributed message passing mechanism to solve the model, omit C7 to solve the expected output result of (8). In order to solve (8) with the distributed message passing mechanism, redefine the t-th iteration, RUE The message from the end to the VUE end is

And the message from the VUE side to the RUE side is

表示VUE的组成集合V删除第v个VUE的剩余子集(即V/{v})中的第K_r个最小消息量，且0＜K_r≤M,K_r∈Z⁺为预定义的参数；Among them, ω is a predefined damping coefficient to ensure the convergence of the algorithm,

Denotes that the constituent set V of VUE deletes the K _rth smallest message volume in the remaining subset of the vth VUE (ie V/{v}), and 0<K _r ≤M, K _r ∈ Z ⁺ is predefined parameter;

Since the solution target type of (8) is to minimize the optimal summation, correspondingly, the present invention sets the solution target of the distributed message passing mechanism to minimize the sum of the message volume, and obtain the final iterative output:

In order to more fully illustrate the beneficial effects of the present invention, the following describes the effectiveness and advancement of the present invention in combination with specific embodiments and related simulation results and analysis.

It is assumed that the system consists of one eNB, N=10 randomly and uniformly distributed D2D RUEs, and M=30 randomly and uniformly distributed VUEs. Among them, the value of the maximum number of VUEs K _r that can be accessed by each RUE end is uniformly and randomly generated within the range of [4, 6]. In practical applications, the specific value of K _r can be determined by the RUE according to its own situation (remaining power, security, user's willingness to share, etc.), and the value is reported to the eNB. VUE is further divided into IC VUE and OOC VUE according to whether it is covered or not. D2D links are described by large-scale losses plus small-scale fading channels. For the convenience of expression, the value of 0.5 on the horizontal axis of the coordinate represents the average value of the index QoS values of all candidate RUEs, then 0.4 represents the average value of 80%, and 0.6 represents the average value of 120%. This embodiment uses the real social network experimental data of CRAWDAD upb/hyccups (v. 2016-10-17) to simulate the social similarity index value.

The USARA scheme of the present invention will be compared with the aforementioned typical D2D UE-NW relay selection scheme, namely: maximum physical link capacity (Max Physical), maximum physical link capacity and maximum social similarity (Max Physical MaxSocial, MPMS), Hybrid Selection Scheme (HRS) and complementary fuzzy analytic hierarchy process- Mahalanobis distance method (FAHP-M). Among them, in order to achieve comparability, in the simulation test, the minimum physical distance target in the MPMS method and the HRS method is equivalently converted into the maximum capacity target.

On the input side of the system, a typical subjective fuzzy decision ordering (from high to low) is set as follows:

■IC scenario: capacity→social similarity→consumption→cache size;

■ OOC scenario: Capacity → Consumption → Cache Size → Social Similarity.

As mentioned earlier, in the IC scenario, "consumption" is defined as the cost required by the VUE to motivate the RUE to perform relay services; in the OOC scenario, "consumption" is defined as the power consumption of the VUE itself.

For the MIFAHP method and the FAHP-M method proposed by the present invention, its implementation needs to input a predefined typical intuitive fuzzy input matrix and a complementary fuzzy input matrix. An example of these matrices is given in Tables 1 and 2. In the actual system, the value of each element in the matrix should be selected according to the network conditions.

Table 1 Typical intuitive fuzzy input matrix required by MIFAHP method

(a) IC scenario

index capacity social similarity consume cache size capacity (0.5,0.5) (0.6,0.2) (0.7,0.1) (0.8,0.1) social similarity (0.2,0.6) (0.5,0.5) (0.6,0.2) (0.7,0.1) consume (0.1,0.7) (0.2,0.6) (0.5,0.5) (0.6,0.2) cache size (0.1,0.8) (0.1,0.7) (0.2,0.6) (0.5,0.5)

(b) OOC scenario

index capacity social similarity consume cache size capacity (0.5,0.5) (0.9,0.1) (0.6,0.2) (0.7,0.1) social similarity (0.1,0.9) (0.5,0.5) (0.1,0.7) (0.2,0.6) consume (0.2,0.6) (0.7,0.1) (0.5,0.5) (0.7,0.1) cache size (0.1,0.7) (0.6,0.2) (0.1,0.7) (0.5,0.5)

Table 2 Typical complementary fuzzy input matrices required by FAHP-M method

(a) IC scenario

index capacity social similarity consume cache size capacity 0.5 0.6 0.7 0.8 social similarity 0.4 0.5 0.6 0.7 consume 0.3 0.4 0.5 0.6 cache size 0.2 0.3 0.4 0.5

(b) OOC scenario

index capacity social similarity consume cache size capacity 0.5 0.9 0.6 0.7 social similarity 0.1 0.5 0.1 0.2 consume 0.4 0.9 0.5 0.7 cache size 0.3 0.8 0.3 0.5

In Figure 6-Figure 9 and Figure 11-Figure 14, in order to examine one type of indicator, its QoS threshold is changed as an independent variable, while the values of other indicators are fixed at 0.5; in Figure 9 and Figure 14, The QoS thresholds (Allthresholds) of all types of metrics are all changed as independent variables. When the QoS threshold is lowered, the access threshold of VUE is lowered, so the access ratio of all schemes will increase. When the QoS threshold is the lowest, the access ratio of each scheme reaches the maximum. It can be seen from FIG. 5 to FIG. 14 that the solution proposed by the present invention (ie, the proposed curve in each figure) can obtain a higher access ratio than all other solutions. This is because this scheme minimizes the RPD and the QoS indication quantity φ at the same time, so that the system can obtain the ideal weighted value of all the indicators as much as possible, and satisfy the QoS conditions of all the indicators as much as possible. In contrast, none of the other existing solutions can satisfy the above two purposes at the same time.

The present invention tests the system throughput and fairness. It can be seen from Fig. 16-Fig. 19 that the USARA scheme proposed by the present invention can achieve higher system throughput and fairness than the existing scheme. Among them, the advantage of throughput rate directly benefits from the high access ratio performance of this solution. In addition, the optimization process of the USARA scheme considers the QoS conditions of all types of indicators, instead of only focusing on the QoS conditions of a single or part of the types of indicators, so that the system can obtain higher fairness.

In particular, compared with FAHP-M, in addition to referring to the VUE subjective preference weight, this solution also makes a more comprehensive RUE relay selection judgment with the help of the objective decision weight provided by the index data. In addition, in the data normalization preprocessing stage, this scheme adopts different processing methods for gain-type indicators and loss-type indicators to reflect the preference trend of different types of indicators.

The same or similar reference numbers correspond to the same or similar parts;

The positional relationship described in the accompanying drawings is only for exemplary illustration, and should not be construed as a limitation on this patent;

Obviously, the above-mentioned embodiments of the present invention are only examples for clearly illustrating the present invention, and are not intended to limit the embodiments of the present invention. For those of ordinary skill in the art, changes or modifications in other different forms can also be made on the basis of the above description. There is no need and cannot be exhaustive of all implementations here. Any modification, equivalent replacement and improvement made within the spirit and principle of the present invention shall be included within the protection scope of the claims of the present invention.

Claims (1)

1. A social D2D relay selection method is characterized by comprising the following steps:

s1: preprocessing input data to obtain data of a required type;

s2: performing index, namely target decision weight generation calculation on the data in the S1;

s3: converting and integrating the single target output quantity obtained in the S2 and the QoS condition, and further solving through a distributed message transmission mechanism to obtain a final output result;

the specific process of step S1 is:

let the relay user equipment RUE form a set R ═ { R ═ R₁,R₂,...,R_NN is the number of candidate relay user equipment RUE; user equipment VUE of relay connection service forms set V ═ { V ═ V-₁,v₂,...,v_MM is the number of VUE;

the indexes of the method which need to carry out decision weight generation calculation are as follows:

link capacity between VUE v and RUE r

Social similarity between VUE v and RUE r S_v,r: the social attribute is described by a Jacard coefficient and is defined as the proportion of the common social attribute owned by the VUE v and the RUE r to the total social attribute;

buffer size beta of RUE r end_r；

VUE v acquires consumption C required for RUE r relay service_v,r: in IC scenarios, the cost required to implement relay services for VUE-excited RUEs; in an OOC scenario, the power consumption of the VUE itself is assumed;

the multiple targets form a target set of VUE v ends

The capacity, the social similarity and the cache size of the RUE end are gain indexes, and the higher the numerical value is, the better the numerical value is; on the other hand, the consumption is a loss index, and the lower the value is, the better the value is;

at the same time, a binary selection variable X is defined_v,rTo indicate whether VUE v selects a candidate RUE r:

in addition to the above multiple targets, the indexes to be calculated for generating decision weight include:

QoS conditions required by the VUE end for each index;

acceptance of the RUE end, i.e. maximum number of accessible VUs K_r；

Each VUE can only access one unique RUE;

based on the above, the optimization model of the solution required by the method is as follows:

max{P1,P2,P3,-P4} (2)

the model is limited to:

wherein:

S_v,thresh,β_v,thresh,C_v,threshcapacity QoS threshold, social similarity QoS threshold, cache QoS threshold and consumption QoS threshold of a VUE v end respectively;

the specific process of step S2 includes performing subjective preference decision weight generation and objective decision weight generation;

the subjective preference decision weight generation comprises the following steps:

1) constructing a visual fuzzy preference relationship;

2) constructing a perfect multiplicative consistency visual fuzzy relation matrix;

3) generating visual fuzzy numerical weight;

4) generating a determined number sequencing numerical value;

5) normalizing the ranking values to generate an output subjective preference weight λ'_i ^,vI.e. the subjective preference weight corresponding to the ith index of the VUE v end;

the objective decision weight generation comprises the steps of:

1) inputting all index values of all candidate RUEs;

2) data preprocessing;

3) obtaining objective decision weight by an entropy weight method;

4) processing a result obtained by integrating the subjective preference decision weight to generate a single target relative approximation degree sequence T of an output VUE v end_v；

The specific process of step S3 is:

through gamma_vConverting the original optimization problem model into a new optimization model with the minimum sum of solving single-target relative approximation RPD:

limited by: C1-C6 (6)

Since this optimization model containing QoS threshold conditions cannot be solved directly by existing distributed messaging algorithms, to further solve the optimization problem (5), the following indicators are introduced:

further, combining the objectives of (7) and (4), the following transformation model is obtained:

limited by:

wherein C7 is a new condition introduced to ensure that (8) and (5) are mathematically equivalent, the objective function for models (8), (5) has been combined with QoS threshold conditions such that the solution of (8) is semantically equivalent to solving for relay selection results that can simultaneously satisfy the minimum RPD and satisfy all QoS threshold conditions, since in practical communication systems condition C7 is difficult to be fully satisfied in a binary selection problem under this type of QoS constraint, C7 is omitted for model solution with distributed messaging mechanisms, to solve for (8) the desired output result, and in order to solve for (8) with distributed messaging mechanisms, the message from RUE end to e end is redefined in the tth iteration of VUE end to e end

And the message from the VUE end to the RUE end is

Where ω is a predefined damping coefficient that ensures algorithm convergence,

the constituent set V of VUs represents the Kth in the Vth remaining subset of VUs (i.e., V/{ V }) that is deleted_rA minimum message size, and 0<K_r≤M,K_r∈Z⁺Is a predefined parameter;

because the type of the solution objective in (8) is the minimum optimization summation, correspondingly, the invention sets the solution objective of the distributed message transfer mechanism as the sum of the minimum message quantity to obtain the final iteration output:

Images