Access point selection and resource allocation for indoor VLC-RF to meet multi-service QoE requirements

Abstract

The invention relates to access point selection and resource allocation for indoor VLC-RF (Virtual Link Control-Radio Frequency) to meet multi-service QoE (Quality of Experience) requirements, and belongs to the field of optical wireless communication. The method comprises the following steps: aiming at access point (AP) attributes of an indoor VLC-RF heterogeneous network and demand characteristics of a user request service, firstly, adopting an analytic hierarchy process and an entropy weight method, and combining a multi-attribute decision-making method to obtain a comprehensive weight value of a hierarchical weight value and an entropy weight value of the user service to a candidate AP attribute expected value; then, obtaining AP attribute grey correlation values of energy efficiency, link stability, transmission delay and service price attributes of the candidate APs by adopting a grey correlation analysis method, and evaluating QoE values obtained by accessing the candidate APs by a user; and finally, constructing a QoE value weighted bigraph of a user request service connection candidate AP, and proposing a maximum weight matching algorithm of an AP selection user meeting the multi-service request of the user under the condition of meeting system fairness. According to the method provided by the invention, the QoE and fairness of the multi-service request of the user in the indoor VLC-RF heterogeneous network can be improved.

Description

Technical Field

[0001] The invention belongs to the technical field of optical wireless communication and relates to access point selection and resource allocation for indoor VLC-RF that meets multi-service QoE requirements. Background Art

[0002] In 6G (6th Generation Mobile Communication Technology), visible light communication (VLC) technology plays a vital role. With its broad potential communication bandwidth, extremely high data transmission speed and superior transmission quality, VLC is considered to be the core technology for future short-distance mobile communication and indoor high-speed data transmission, and provides a key solution for achieving the goals of 6G communication systems, including ultra-high speed, ultra-low latency and ultra-large bandwidth data communication connections. The VLC system is an important communication technology recommended by the 6G communication system. It uses light emitting diode (LED) technology to emit light waves for illumination and light signals for communication. The system can provide users with faster high-speed data transmission speeds. The VLC system is not only to increase communication speed, but also aims to create a unified network with full spectrum, comprehensive coverage, energy conservation and environmental protection, and wide application to support the goal of the intelligent connection of all things. This vision means that all devices and objects can be connected to each other through the network to achieve comprehensive information sharing and intelligent interaction.

[0003] In order to promote the common progress of various communication technologies in 6G, the VLC-RF heterogeneous network combining radio frequency (RF) communication and VLC technology has become a seamless coverage converged communication mode. This fusion is expected to promote the collaborative work between different communication technologies, promote the sharing of spectrum resources and the seamless coverage of the network. With the integration of RF and VLC, IoT applications will be able to develop at a deeper level, laying the foundation for the future smart society. In terms of meeting indoor communication needs, the VLC-RF heterogeneous network based on VLC and RF technology has become a direction that has attracted much attention. Through a large number of VLC access points (APs) deployed indoors, high-speed short-distance data communication optical path connections are provided for users with unobstructed indoor light, and interference management and RF communication connections for users with light obstruction are provided for the dense deployment of cells and irregular user distribution. Therefore, the VLC-RF heterogeneous network is expected to become one of the important technologies for achieving efficient indoor communication, contribute to the comprehensive development of 6G communication systems, and help realize a more intelligent, efficient, and connected communication era. In this intelligent process, visible light communication, the integration of RF and VLC, VLC-RF heterogeneous networks and other technologies will play a key role and lead the continuous innovation and development of communication technology. However, indoor occasions have high user density, interference management difficulties caused by dense deployment of VLC cells, low VLC radiation optical power, and limited modulation bandwidth and resources available for communication. At the same time, indoor user distribution is often irregular, resulting in unbalanced load of VLC APs, limiting the communication capacity and resource utilization of VLC-RF. On the other hand, people have higher requirements for communication service quality and QoE (Quality of Experience) in 6G communication. Different services have different communication performance indicators that reflect QoE, resulting in different QoE requirements for different service requests generated by users and different expectations for QoE service performance provided by the network, further increasing the complexity of VLC-RF interference management and resource allocation. Therefore, designing and using effective AP access mechanisms and resource allocation schemes for indoor VLC-RF heterogeneous networks is very important to improve indoor users' high QoE when requesting different services. Summary of the invention

[0004] In view of this, the purpose of the present invention is to provide an indoor VLC-RF access point selection and resource allocation method that meets the QoE requirements of multiple services, while optimizing the user's request for service access AP and resource allocation on the basis of meeting the fairness of the system, and improving the user's QoE.

[0005] In order to achieve the above object, the present invention provides the following technical solutions:

[0006] An adaptive access resource allocation method for meeting multi-service QoE requirements in an indoor VLC-RF heterogeneous network, characterized in that the method comprises the following steps:

[0007] S1: Input user set M, service request set S, service bandwidth, and service attribute expected value a si The expected value, minimum value and maximum value of network energy efficiency, link stability, delay and price attributes; the set Y of VLC (Visible Light Communication) access points (AP) uniformly distributed on the indoor roof in a rectangular shape, and the transmission power of the VLC access point is P V and available resource values, one RF (Radio Frequency) access point located at the center of the indoor roof has a transmission power of P R and available resource values, the AP set of all VLC APs and RF APs is N, and the element number of RF AP in the set N is 0, the attribute set of AP is K, and the attribute g of the service i that AP can provide is ni , link stability, delay and price attribute parameters, their minimum and maximum values, and the maximum and minimum values ​​of network energy efficiency; based on the Lambertian radiation model of VLC, calculate the channel gain value from each user to each VLC access point; based on the Rayleigh fading channel model, calculate the channel gain from each user to the RF access point;

[0008] Among them, the formula for calculating the LoS (Line of Sight) channel gain from VLC AP n to the receiver PD (Photo Detector) of user m by the Lambertian radiation model is:

[0009]

[0010] In the formula, A PD Indicates the light receiving area; D m,n is the straight-line distance between user m and AP n; is the VLC AP emission angle; f represents the Lambertian radiation order, and is the VLC AP half-power angle; ψ is the incident angle of the optical receiver; Ψ c is the field of view of the optical receiver; T(ψ) and g(ψ) are the optical filter gain and optical concentrator gain of the PD, respectively; if 0≤ψ≤ψ c , then the concentrator g(ψ)=c 2 / sin 2 (ψ c ), where the refractive index c = 1.5; if ψ>ψ c , then g(ψ)=0;

[0011] According to the Rayleigh fading model, the channel gain between the RF AP and user m is calculated as follows:

[0012]

[0013] Where L (dB) represents the path loss of the RF channel in dB. The calculation method is:

[0014]

[0015] Where, d mn is the length of the communication link from RF AP n to user m; f c is the center carrier frequency in GHz; X σ is a zero-mean Gaussian random variable with a standard deviation of 3 dB, representing shadow fading; d BP It is the reference distance between the user and the RF AP, which is generally taken as d BP =10m; free space path loss L(d mn )=20log 10 (f c d mn )-147.5;

[0016] S2: According to the expected value of the user's requested service for AP attributes, the hierarchical analysis method and entropy weight method are used to execute the comprehensive weight value evaluation algorithm of the requested service based on multi-attribute decision-making, and the comprehensive weight value vector W of the expected value of the user's requested service s for the candidate AP attributes is obtained. s =(w e ,w l ,w d ,w p ); Among them, the expected values ​​of the business for AP attributes include: network energy efficiency a se , Link stabilitya sl , delay a sd and price a sp ;

[0017] S3: According to Shannon's formula, the energy efficiency of RF AP and user m is calculated in turn. Value, energy efficiency of VLC AP n and user m value, m∈X, n∈Y; perform the QoE (Quality of Experience) evaluation method of the grey correlation analysis of the user requested service and the candidate AP attributes, obtain the AP attribute grey correlation value between the candidate AP attribute parameters and the requested service, and evaluate the QoE value of the requested service to the candidate AP; wherein the attribute parameters of the candidate AP include: energy efficiency g ne , Link stability gnl , delay g nd and price parameter g np ;

[0018] S4: Construct a weighted bipartite graph of user multi-service requests and candidate APs, and adopt a maximum weight matching AP selection algorithm that meets multi-service QoE requirements. On the basis of satisfying system fairness, find an AP association that meets the user's service requirements and provides the maximum QoE value.

[0019] Further, the specific method of S2 is:

[0020] S201: Based on the expected value of the network attribute i of the service s requested by the user m, the importance of the expected value is quantified by using the hierarchical analysis method with reference to the importance scale table, and a comparison matrix D = (d ij ) K×K , where the expected value a of the network attribute i for the user's requested service s is si Includes: Energy efficiency se , Link stabilitya sl , delay a sd and price a sp , s∈S, K represents the number of attributes in the AP attribute set K, and the element d in the matrix D ij Indicates the expected value of the business a si Relative to the expected value a sj The relative importance evaluation value of ii =1,d ij =1 / d ji , d ij None of the values ​​are 0; Table 1 shows the quantitative values ​​of the 9-level importance scale;

[0021] S202: Calculate the hierarchical weight coefficient of the attribute expected value i of the requested service s of user m and save it in the hierarchical weight coefficient vector W′ s =(w′ e ,w′ l ,w′ d ,w′ p ), s∈S;

[0022] The calculation formula of the hierarchical weight coefficient of the attribute expected value i of the requested service s of user m is:

[0023]

[0024] Table 1: 9-level importance scale

[0025]

[0026] S203: Calculate the consistency ratio CR of the elements of the matrix D = CI / RI value, where RI is the average random consistency index, with a value of 0.89; CI is the consistency index, and its calculation formula is:

[0027]

[0028] In the formula, λ max is the maximum eigenvalue of matrix D;

[0029] S204: Determine whether the consistency ratio CR is greater than 0.1. If so, it means that there is a contradiction in the element values ​​of the matrix D. Go to step S201 and reset the relative importance evaluation value d. ij ; Otherwise, go to step S205;

[0030] S205: Construct the normalized attribute expected value matrix A of all services of the user = [a′ si ] S×K , where S is the service set, S is the number of services, and a′ si is the normalized expected attribute value, and is the highest and lowest expected value of service s for attribute i, where the network energy efficiency a in the expected value se and link stabilitya sl is a positive attribute, and the delay a sd and price a sp is a negative attribute. For a positive attribute, its normalized attribute value is expressed as For negative attributes, the normalized attribute value is

[0031] S206: Calculate the normalized attribute expected value matrix A of the service s of user m = [a′ si ] S×K The weight of all elements in ; where the weight of the expected value of the i-th attribute of the service s of user m is calculated as follows:

[0032]

[0033] S207: Calculate the information entropy of the expected value of each network attribute of the service s of user m, s∈S;

[0034] The calculation formula of the information entropy of the expected value of the i-th attribute of the network by the service s of user m is:

[0035]

[0036] S208: Calculate the information utility value and entropy weight coefficient value of the expected value of the network attribute of the service s of user m, and save it in the entropy weight coefficient vector W″ of the expected value of the attribute of the service s of user ms =(w″ e ,w″ l ,w″ d ,w″ p ), s∈S;

[0037] The calculation formula of the information utility value of user m's service s to network attribute i is:

[0038] d i =1-Q si (8)

[0039] The calculation formula of the entropy weight coefficient of the expected value of the service s of user m to the network attribute i is:

[0040]

[0041] S209: Calculate the hierarchical weight value vector W′ that minimizes the expected value of network attributes of all services s =(w′ e ,w′ l ,w′ d ,w′ p ) and the entropy weight vector W″ s =(w″ e ,w″ l ,w″ d ,w″ p ) is the sum of the squares of all element differences between them, s∈S, and the distribution coefficients of the subjective weight value and the objective entropy weight value are obtained;

[0042] Among them, the hierarchical weight value that minimizes the expected value of all network attributes of business s and entropy weight The optimization objective of the sum of squares of all element differences between is:

[0043]

[0044] In the formula, α and β are respectively

[0045] Solve the minF optimization objective formula to obtain the distribution coefficients of the hierarchical weight values ​​and entropy weights;

[0046] S210: After determining the values ​​of α and β, evaluate the hierarchical weight coefficient vector and the comprehensive weight coefficient vector of the entropy weight of the user m's service s for the candidate AP preference, and save it in the comprehensive weight value vector W s =(w e ,w l ,w d ,w p ), s∈S.

[0047] Among them, the evaluation formula of the comprehensive weight value of the hierarchical weight value and entropy weight value preferred by user m's service s for candidate AP attribute i is:

[0048] w si =αw i ′+βw i ″ (11)

[0049] Further, the specific method of S3 is:

[0050] S301: Calculate the signal-to-noise ratio value of user m served by the RF AP, and the calculation formula is:

[0051]

[0052] In the above formula, N is the RF AP electrical transmission power; R B is the RF noise power spectrum density, which is generally -174dBm / Hz; R is the communication bandwidth of RF;

[0053] S302: Calculate the maximum achievable rate value of user m served by the RF AP using the Shannon formula, where the calculation formula is:

[0054]

[0055] S303: Calculating the energy efficiency of RF AP serving user m The value is calculated as follows:

[0056]

[0057] Where P R Indicates the fixed power of the RF AP. is the power consumed by the RF AP in communicating with user m; Θ0 represents the set of users served by the RF AP;

[0058] S304: Calculate the signal-to-interference-and-noise ratio value of user m served by VLC AP n, and the calculation formula is:

[0059]

[0060] In the formula, is the optical transmission power of VLC AP n; γ is the photoelectric conversion coefficient; is the VLC communication bandwidth of user m; N V is the noise power spectrum density of VLC, which is generally 1×10 -21 A 2 / Hz;

[0061] S305: Calculate the maximum achievable rate of VLC AP n serving user m, and the calculation formula is:

[0062]

[0063] S306: Calculating the energy efficiency of VLC AP n serving user m The value is calculated as follows:

[0064]

[0065] Where P V is the fixed power at which the VLC AP operates; is the power consumed by the communication between VLC AP n and user m;

[0066] S307: Constructing the AP attribute parameter matrix G of candidate VLC APs and RF APs = (g ni ) N×K , where element g ni Defines the parameter value of attribute i that AP n can provide. AP attributes include: AP energy efficiency g ne , Link stability g nl , delay g nd and price parameter g np ,in,

[0067] S308: Calculate the normalized attribute value of AP and construct the normalized attribute decision matrix H of AP = (h ni ) N×K ;

[0068] Among them, the energy efficiency g of AP n with positive attributes is ne and link stability g nl The normalized property of i∈{e,l},n∈N, The general formula for calculating the value is:

[0069]

[0070] In the formula, and They represent the upper and lower limits that the AP can provide for the attribute i of service s respectively; is the intermediate threshold value that the AP can provide to service s for attribute i. Generally speaking,

[0071] Among them, for the delay attribute g of AP n with negative attribute nd The normalized property of The calculation formula is:

[0072]

[0073] In the formula, and They represent the minimum and maximum values ​​required by service s for AP delay attribute d respectively; is the middle threshold value of the service s's requirement for the AP delay attribute. Generally speaking,

[0074] Where, for AP n, the network price g nc The normalized attribute of nc , n∈N, the calculation formula is:

[0075]

[0076] in, and They represent the price of the network providing user services g nc The lower and upper limits of

[0077] S309: Calculate the ideal reference value of the AP attribute expected by the user service s;

[0078] The calculation formula of the ideal reference value of AP attribute i expected by user service s is:

[0079]

[0080] S310: Calculate the AP attribute grey correlation value between the attribute of candidate AP n and the ideal reference value of the AP attribute; evaluate the QoE value obtained when the service s requested by user m accesses the candidate AP n according to the comprehensive weight value of the service preference AP and the grey correlation value of the AP attribute;

[0081] The calculation formula of the AP attribute grey correlation value between the attribute i of candidate AP n and the ideal reference attribute i of AP is:

[0082]

[0083] The evaluation formula for the QoE value obtained when the service s requested by user m accesses candidate AP n is:

[0084]

[0085] In the formula, Q mn The value is in [0,1], Q mn The higher the value of, the better the QoE experience obtained by user m when accessing AP n. mn When it is 1, candidate AP n is the ideal AP that the user wants to access.

[0086] Further, the specific method of S4 is:

[0087] S401: Determine the matching value ω according to the QoE value mn , with ω mn The weight of the edge between the node m represented by the user and the node n represented by the AP is used to construct a weighted undirected bipartite graph Z of the user access candidate APs. m-n , and mark the required resources of user m requesting service s and the idle available resource values ​​of AP n node next to node m and node n respectively;

[0088] Among them, the matching value of user m accessing candidate AP n is ω mn The calculation formula is:

[0089]

[0090] Where M is the total number of users; Δ m is the expected value of network delay attribute for user m requesting service s, that is: Δ m =a sd ;

[0091] S402: Calculate the fairness index of the system;

[0092] Among them, the calculation formula of the system's fairness index is:

[0093]

[0094] S403: weighted undirected bipartite graph Z of user access candidate APs m-n In the graph theory, the maximum weight matching method is adopted to select the node n representing the candidate AP with the largest edge weight matching value for each node m representing the user, m∈M, n∈N. If the idle available resource value of the candidate AP connected by the maximum edge weight value is greater than the resource value of the service demand requested by user m, then in the matching result graph Z′ m-n In the equation, the weight Δ m Connect user m and candidate AP n, and m-n Delete user node m, and update the remaining network bandwidth resources of AP n node selected by user m as follows: in, is the idle available resource value of AP n, is the requested resource value of user m, and the process goes to step S407; otherwise, if the remaining resources of other nodes connected by non-maximum edge weights are greater than or equal to the resources requested by user m, user m is saved in the set List u In Z m-n The edge weight ω between user m and candidate AP n is mnSet to 0, and go to step S404; otherwise, block the request connection of user m;

[0095] S404: If List u Not empty, List u The users whose maximum weight association failed are sorted in ascending order according to their maximum matching value in List u If yes, go to step S405; otherwise, go to step S406;

[0096] S405: Select List u For the first user, go to step S403;

[0097] S406: Calculate the fairness index of the system. If I F ∈(0.7,1]?n∈N, go to step 407; otherwise, in Z′ m-n In the example, we select the edge weight Δ m The user m with the largest value is deleted from Z′ m-n The corresponding edge in , go to step S403;

[0098] S407: Output matching result graph Z′ m-n The matching result of the user request service selection AP with edge connection.

[0099] The beneficial effects of the present invention are as follows: the present invention relates to an adaptive access resource allocation method that meets the QoE requirements of multiple services in an indoor VLC-RF heterogeneous network. Firstly, a comprehensive weight value evaluation algorithm for request services of multi-attribute decision-making is proposed by combining an analytic hierarchy process (AHP) and an entropy weight method (EWM). Secondly, in order to meet the personalized requirements of user request service characteristics and obtain the QoE value of a certain service requested by the user from different candidate APs, a QoE evaluation method of a grey relational analysis (GRA) of user request services and candidate AP attributes is proposed. The comprehensive weight value that can reflect the subjective evaluation and objective evaluation of the service expectation of the service on the AP service capability is combined with the grey relational value of the AP attribute supporting the service performance to obtain the QoE evaluation value of the user request service. Finally, a maximum weight matching AP selection algorithm that meets the QoE requirements of multiple services is adopted. From a weighted undirected bipartite graph between the constructed user request service and the candidate AP, an optimal matching method is adopted to find the candidate AP with the largest weight value for the request service of each user to associate or access on the basis of meeting the fairness of the system. Therefore, the method can meet the QoE of users' requests for different services and find the best AP access for different service requests, thereby improving the satisfaction of users' requested services. BRIEF DESCRIPTION OF THE DRAWINGS

[0100] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be described in detail below with reference to the following drawings, in which:

[0101] Figure 1 Schematic diagram of indoor VLC-RF heterogeneous network model;

[0102] Figure 2 Schematic diagram of the weighted bipartite graph of user access candidate APs;

[0103] Figure 3 Flowchart of the comprehensive weight value evaluation algorithm for request services based on multi-attribute decision making;

[0104] Figure 4 QoE evaluation method based on grey correlation analysis of user requested services and candidate AP attributes;

[0105] Figure 5 Access point selection and resource allocation flow chart of VLC-RF to meet multi-service QoE requirements; DETAILED DESCRIPTION

[0106] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents described in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways 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 only illustrate the basic concept of the present invention in a schematic manner, and the following embodiments and the features in the embodiments can be combined with each other without conflict.

[0107] Among them, the drawings are only used for illustrative explanations, and they only represent schematic diagrams rather than actual pictures, and should not be understood as limitations on the present invention. In order to better illustrate the embodiments of the present invention, some parts in the drawings may be omitted, enlarged or reduced, and do not represent the size of actual products. For those skilled in the art, it is understandable that some well-known structures and their descriptions in the drawings may be omitted.

[0108] The embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0109] Attached Figure 1It is a 10m×10m indoor VLC-RF heterogeneous network model, including an RF AP placed in the center of the indoor roof and 16 VLC APs evenly distributed in a rectangular shape. The VLC APs are 2m apart and connected to each other through power lines, and connected to a centralized controller (Controller) through power lines, and finally connected to the Internet. Multiple users are randomly distributed indoors, and each user can request different types of services, such as: traditional streaming media services, background data services, emerging VR (Virtual Reality) & AR (Augmented Reality) and other 5G (5th Generation Mobile Communication Technology) services, office automation services, etc. The controller can save the status information of various user channels in the heterogeneous network, the service parameters requested by the user, the resource status and service parameters of the VLC AP and RF AP, and provide algorithm operation and driving capabilities, as well as the computing power required by users and APs.

[0110] Attached Figure 2 A weighted bipartite graph diagram of a user accessing a candidate AP. In order to describe the matching relationship between the resource status of the candidate AP and the service requirements requested by the user, this application establishes a bipartite graph Z m-n Describes the matching relationship between user requests and the service capabilities provided by AP. m-n In the example, the node set V = M∪N, M is the user set, M = {u1,…,u i ,…,u M}, where M represents the total number of users, u i represents the user who requests the i-th communication service; N is the set of network AP nodes, including Y VLCAPs and 1 RF AP, where RF AP is the 0th in the sequence number, N = {y0,…,y j ,…,y N}, where N represents the number of VLC AP nodes, y j represents the jth VLC AP; E represents the user u requesting the service i and network node y j The edge weight matching relationship, E = {e i,j |i=1,2,…M;j=0,1,2,…,N}, W represents the weight function of edge E, ω ij Represents edge e i,j The weight of .

[0111] Attached Figure 3 The flowchart of the comprehensive weight value evaluation algorithm for request services based on multi-attribute decision-making is as follows:

[0112] Input: Input user set M, business request set S, input business resource requirement value, input business attribute expected value a si The expected value, minimum value and maximum value of network energy efficiency, link stability, delay and price attributes; VLC AP set Y, total AP set N of RF APs and VLC APs;

[0113] Output: The comprehensive weight vector W of the hierarchical weight value and entropy weight value of the business request s s =(w e ,w l ,w d ,w p ), s∈S;

[0114] Step 1: Based on the expected value a of the network attribute i requested by user m si , refer to the 9-level importance scale table, and construct a comparison matrix D between the importance scale values ​​of the business expected network attributes and network attributes. ij ) K×K , where the expected value a of the network attribute i for the user's requested service s is si Includes: Energy efficiency se , Link stabilitya sl , delay a sd and price a sp , element d ij Indicates the expected value of the business a si Relative to the expected value a sj The relative importance evaluation value of ii =1,d ij =1 / d ji , d ij None of the values ​​are 0;

[0115] Step 2: According to D = (d ij ) K×K , according to formula (4), calculate the hierarchical weight coefficient of the attribute expected value i of the requested service s of user m and save it in the hierarchical weight coefficient vector W′ s =(w′ e ,w′ l ,w′ d ,w′ p ), s∈S;

[0116] Step 3: Calculate the consistency ratio CR of the elements of the matrix D = CI / RI value, where RI is the average random consistency index, with a value of 0.89; CI is the consistency index, calculated by formula (5);

[0117] Step 4: If the consistency ratio CR is greater than 0.1, it means that the relative importance evaluation of the element values ​​in the matrix D is inconsistent. Go to step 1 and reset the relative importance evaluation value d. ij ; Otherwise, go to step 5;

[0118] Step 5: Construct the normalized attribute expected value matrix A of all services of the user = [a′ si ] S×K , where a′ si is the normalized expected attribute value, and is the highest and lowest expected value of service s for attribute i, where the network energy efficiency a in the expected value se and link stabilitya sl is a positive attribute, and the delay a sd and price a sp is a negative attribute. For a positive attribute, its normalized attribute value is expressed as For negative attributes, the normalized attribute value is

[0119] Step 6: According to formula (6), calculate the normalized attribute expected value matrix A of user m's service s = [a′ si ] S×K The specific gravity of all elements in;

[0120] Step 7: Calculate the information entropy of the expected value of each network attribute of user m's service s by formula (7);

[0121] Step 8: Calculate the information utility value and entropy weight coefficient value of user m's service s on the expected value of network attributes by using formula (8) and formula (9), and save it in the entropy weight coefficient vector W″ of the expected value of the attribute of user m's service s s =(w″ e ,w″ l ,w″ d ,w″ p ), s∈S;

[0122] Step 9: Based on formula (10), construct the hierarchical weight value vector W′ that minimizes the expected value of network attributes of all services s =(w′ e ,w′ l ,w′ d ,w′ p ) and the entropy weight vector W″ s =(w″ e ,w″ l ,w″ d ,w″ p) and adopt common convex optimization methods, such as Lagrange multiplier method, to obtain the distribution coefficients of subjective weight value α and objective entropy weight value β.

[0123] Step 10: After determining the values ​​of α and β, the hierarchical weight coefficient vector and the comprehensive weight coefficient vector of the entropy weight of the user m's service s for the candidate AP preference are evaluated by formula (11) and saved in the comprehensive weight value vector W s =(w e ,w l ,w d ,w p ), s∈S;

[0124] Step 13: Output the comprehensive weight value vector W of the request service s s =(w e ,w l ,w d ,w p ), s∈S.

[0125] Attached Figure 4 The QoE evaluation method based on grey correlation analysis of user request services and candidate AP attributes has the following main steps:

[0126] Input: Input user set M, service request set S, input service resource requirement value, VLC AP set Y, total AP set N of RFAP and VLC AP, VLC AP transmission power P V and available resource value, the RF AP transmit power is P R and available resource values, the attribute set of AP is K, and the attribute g of service i that AP can provide ni The link stability, delay and price attribute parameters, their minimum and maximum values, the maximum and minimum values ​​of network energy efficiency; the comprehensive weight value vector W of the user request service s s =(w e ,w l ,w d ,w p ), s∈S;

[0127] Output: The QoE value Q obtained by the service s requested by user m accessing candidate AP n mn , m∈M, n∈N.

[0128] Step 1: According to the user distribution location and AP layout, the channel gain value from user m to VLC AP n is calculated by formula (1): m∈M,n∈Y; According to formula (2), the channel gain from user m to RF AP is calculated m∈M;

[0129] Step 2: Calculate the signal-to-noise ratio and maximum achievable rate of user m served by the RF AP using formulas (12) and (13);

[0130] Step 3: Calculate the energy efficiency of RF AP serving user m using formula (14): value, m∈M;

[0131] Step 4: Calculate the signal-to-interference-and-noise ratio and the maximum achievable rate of user m served by VLC AP n using formulas (15) and (16), m∈M, n∈Y;

[0132] Step 5: Calculate the energy efficiency of VLC AP n serving user m using formula (17): value, m∈M, n∈Y;

[0133] Step 6: According to the gni value, construct the AP attribute parameter matrix G of the candidate VLC AP and RF AP = (g ni ) N×K ,in,

[0134] Step 7: Calculate the normalized attribute value of AP by using formula (18), formula (19) and formula (20), and construct the normalized attribute decision matrix H of AP = (h ni ) N×K ;

[0135] Step 8: Determine the ideal reference value of the AP attribute expected by user service s using formula (21): s∈S, i∈K;

[0136] Step 9: Calculate the AP attribute grey correlation value ζ between the attribute of candidate AP n and the ideal reference value of AP attribute by formula (22): ni , n∈N, i∈K;

[0137] Step 10: According to formula (23), the QoE value Q obtained by the service s requested by user m accessing the candidate AP n is evaluated mn , m∈M, n∈N;

[0138] Step 11: Output the QoE value Q obtained when the service s requested by user m accesses the candidate AP n mn , m∈M, n∈N.

[0139] Attached Figure 5 The access point selection and resource allocation flow chart for VLC-RF to meet the multi-service QoE requirements is as follows:

[0140] Input: Input user set M, business request set S, input business resource requirement value m∈M, the expected value of the attribute of the input business a si The expected value, minimum value and maximum value of network energy efficiency, link stability, delay and price attributes are Δ m is the delay requirement value of the service requested by user m; the VLC AP set is Y, the total AP set of RF APs and VLC APs is N, and the VLC AP transmission power is P V 、RF AP transmission power is P R and available resource values n∈N; Let the attribute set of AP be K, and the attribute g of service i that AP can provide ni The link stability, delay and price attribute parameters, their minimum and maximum values, and the maximum and minimum values ​​of network energy efficiency;

[0141] Output: Output matching result graph Z′ m-n There are edge-connected users requesting services to select the matching results of APs, m∈M, n∈N.

[0142] Step 1: According to the expected value of the attribute of the input business s, execute the attached Figure 3 The comprehensive weight value evaluation algorithm of the request service based on multi-attribute decision-making shown in the figure obtains the comprehensive weight value vector W of the hierarchical weight value and entropy weight value of the service request s. s =(w e ,w l ,w d ,w p ), s∈S;

[0143] Step 2: Execute the attached Figure 4 The QoE evaluation method based on the grey correlation analysis of the user requested service and the candidate AP attributes is shown in FIG. 1 to obtain the QoE value Q of user m requesting service s to access candidate AP n. mn , m∈M, n∈N;

[0144] Step 3: Based on the QoE value Q mn , according to formula (24), determine the matching value ω mn ;

[0145] Step 4: Take ω mn The weight of the edge between the node m represented by the user and the node n represented by the AP is used to construct a weighted undirected bipartite graph Z of the user access candidate APs. m-n , and mark the required resources of user m requesting service s and the idle available resources of AP n node next to node m and node n respectively; according to formula (25), calculate the fairness index I of the system F value;

[0146] Step 5: In the bipartite graph Z m-nIn the example, the maximum weight matching method of graph theory is used to select the node n representing the candidate AP with the largest edge weight matching value for each user node m, m∈M, n∈N. If the idle available resource value of the candidate AP connected by the maximum edge weight value is greater than the resource value of the service demand requested by user m, then in the matching result graph Z′ m-n In the equation, the weight Δ m Connect user m and candidate AP n, and m-n Delete user node m, and update the remaining network bandwidth resources of AP n node selected by user m as follows: in, is the idle available resource value of AP n, is the requested resource value of user m, go to step 9; otherwise, if there are other nodes connected by non-maximum edge weight whose remaining resources are greater than or equal to the resources requested by user m, then user m is saved in the set List u In Z m-n The edge weight ω between user m and candidate AP n is mn Set to 0 and go to step 6; otherwise, block the request connection of user m and go to step 6;

[0147] Step 6: If List u Not empty, List u The users whose maximum weight association failed are sorted in ascending order according to their maximum matching value in List u If yes, go to step 7; otherwise, go to step 8;

[0148] Step 7: Select List u For the first user, go to step 5;

[0149] Step 8: Calculate the fairness index of the system using formula (25). If I F ∈(0.7,1]?n∈N, go to step 9; otherwise, in Z′ m-n In the example, we select the edge weight Δ m The user m with the largest value is deleted from Z m-n and in Z′ m-n For the corresponding edge in , go to step 5;

[0150] Step 9: Output the matching result graph Z′ m-n The matching result of the service request of user m with edge connection in AP n is selected, m∈M, n∈N.

[0151] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution, which should be included in the scope of the claims of the present invention.

Claims

1. An indoor VLC-RF access point selection and resource allocation that meets multi-service QoE requirements: The method comprises the following steps: S1: Input user set M, service request set S, service resource requirement value, service attribute expected value a si The expected value, minimum value and maximum value of network energy efficiency, link stability, delay and price attributes; the set Y of VLC (Visible Light Communication) access points (AP) uniformly distributed on the indoor roof in a rectangular shape, and the transmission power of the VLC access point is P V and available resource values, one RF (Radio Frequency) access point located at the center of the indoor rooftop has a transmission power of P R and available resource values, the AP set of all VLC APs and RF APs is N, the AP attribute set is K, and the attribute g of the service i that the AP can provide ni The link stability, delay and price attribute parameters, their minimum and maximum values, and the maximum and minimum values ​​of network energy efficiency; based on the Lambertian radiation model of VLC, the channel gain value from each user to each VLC access point is calculated; based on the Rayleigh fading channel model, the channel gain from each user to the RF access point is calculated; Among them, the formula for calculating the LoS (Line of Sight) channel gain from VLC AP n to the receiver PD (Photo Detector) of user m by the Lambertian radiation model is: In the formula, A PD Indicates the light receiving area; D m,n is the straight-line distance between user m and AP n; is the VLC AP emission angle; f represents the Lambertian radiation order, and is the VLC AP half-power angle; ψ is the incident angle of the optical receiver; Ψ c is the field of view of the optical receiver; T(ψ) and g(ψ) are the optical filter gain and optical concentrator gain of the PD, respectively; if 0≤ψ≤ψ c , then the concentrator g(ψ)=c 2 / sin 2 (ψ c ), where the refractive index c = 1.5; if ψ>ψ c , then g(ψ)=0; According to the Rayleigh fading model, the channel gain between the RF AP and user m is calculated as follows: Where, L (dB) represents the path loss of the RF channel, in dB; S2: Using the method described in claim 2, according to the expected value of the user's requested service for the AP attribute, using the hierarchical analysis method and the entropy weight method, executing the comprehensive weight value evaluation algorithm of the requested service based on multi-attribute decision-making, and obtaining the comprehensive weight value vector W of the user's requested service s for the expected value of the candidate AP attribute s =(w e ,w l ,w d ,w p ); Among them, the expected values ​​of the business for AP attributes include: network energy efficiency a se , Link stabilitya sl , delay a sd and price a sp ; S3: Using the method described in claim 3, according to the Shannon formula, the energy efficiency of the RF AP and the user m is calculated in turn. Value, VLCAP n and energy efficiency of user m value, m∈X, n∈Y; perform the QoE (Quality of Experience) evaluation method of the grey correlation analysis of the user requested service and the candidate AP attributes, obtain the AP attribute grey correlation value between the candidate AP attribute parameters and the requested service, and evaluate the QoE value of the requested service to the candidate AP; wherein the attribute parameters of the candidate AP include: energy efficiency g ne , Link stability g nl , delay g nd and price parameter g np . S4: Adopt the method described in claim 4 to construct a weighted bipartite graph of user multi-service requests and candidate APs, adopt a maximum weight matching AP selection algorithm that meets multi-service QoE requirements, and find an AP association that provides the maximum QoE value for the user to meet its service requirements on the basis of satisfying system fairness.

2. The indoor VLC-RF access point selection and resource allocation that meets multi-service QoE requirements as claimed in claim 1, characterized in that: The specific method of S2 is: S201: Based on the expected value of the network attribute i of the service s requested by the user m, the importance of the expected value is quantified by using the hierarchical analysis method with reference to the importance scale table, and a comparison matrix D = (d ij ) K×K , where the expected value a of the network attribute i for the user's requested service s is si Includes: Energy efficiency se , Link stabilitya sl , delay a sd and price a sp , K represents the number of attributes in the AP attribute set K, and the element d in the matrix D ij Indicates the expected value of the business a si Relative to the expected value a sj The relative importance evaluation value of ii =1,d ij =1 / d ji ; S202: Calculate the hierarchical weight coefficient of the attribute expected value i of the requested service s of user m and save it in the hierarchical weight coefficient vector W s ′=(w e ′,w l ′,w′ d ,w′ p )middle; The calculation formula of the hierarchical weight coefficient of the attribute expected value i of the requested service s of user m is: S203: Calculate the consistency ratio CR of the elements of the matrix D = CI / RI value, where RI is the average random consistency index, with a value of 0.89; CI is the consistency index, and its calculation formula is: In the formula, λ max is the maximum eigenvalue of matrix D; S204: Determine whether the consistency ratio CR is greater than 0.

1. If so, it means that there is a contradiction in the element values ​​of the matrix D. Go to step S201 and reset the relative importance evaluation value d. ij ; Otherwise, go to step S205; S205: Construct the normalized attribute expected value matrix A of all services of the user = [a s ' i ] S×K , where S is the service set, S is the number of services, a s ' i is the normalized expected attribute value, and is the highest and lowest expected value of service s for attribute i, where the network energy efficiency a in the expected value se and link stabilitya sl is a positive attribute, and the delay a sd and price a sp is a negative attribute. For a positive attribute, its normalized attribute value is expressed as For negative attributes, the normalized attribute value is S206: Calculate the normalized attribute expected value matrix A of the service s of user m = [a s ' i ] S×K The specific gravity of all elements in; The calculation formula for the weight of the expected value of the i-th attribute of the service s of user m is: S207: Calculate the information entropy of the expected values ​​of each network attribute for the service s of user m; The calculation formula of the information entropy of the expected value of the i-th attribute of the network by the service s of user m is: S208: Calculate the information utility value and entropy weight coefficient value of the expected value of the network attribute of the service s of user m, and save it in the entropy weight coefficient vector W of the expected value of the attribute of the service s of user m s ″=(w e ″,w l ″,w′ d ′,w′ p ')middle; The calculation formula of the information utility value of user m's service s to network attribute i is: d i =1-Q si The calculation formula of the entropy weight coefficient of the expected value of the service s of user m to the network attribute i is: S209: Calculate the hierarchical weight value vector W that minimizes the expected value of network attributes of all services s ′=(w e ′,w l ′,w′ d ,w′ p ) and the entropy weight vector W s ″=(w e ″,w l ″,w′ d ′,w′ p ′) and the optimization target is obtained by summing the squares of all element differences between them, and obtaining the distribution coefficients of subjective weight values ​​and objective entropy weight values; Among them, the hierarchical weight value that minimizes the expected value of all network attributes of business s and entropy weight The optimization objective of the sum of squares of all element differences between is: Where α and β are the distribution coefficients of the hierarchical weight value and entropy weight respectively obtained by solving the minF optimization objective formula by the convex optimization method; S210: After determining the values ​​of α and β, evaluate the hierarchical weight coefficient vector and the comprehensive weight coefficient vector of the entropy weight of the user m's service s for the candidate AP preference, and save it in the comprehensive weight value vector W s =(w e ,w l ,w d ,w p )middle; Among them, the evaluation formula of the comprehensive weight value of the hierarchical weight value and entropy weight value preferred by user m's service s for candidate AP attribute i is: w si =αw i ′+βw i ″。 3. The indoor VLC-RF access point selection and resource allocation that meets multi-service QoE requirements as claimed in claim 1, characterized in that: The S3 specific method is: S301: Calculate the signal-to-noise ratio value of user m served by the RF AP, and the calculation formula is: In the above formula, N is the RF AP electrical transmission power; R B is the RF noise power spectrum density, which is generally -174dBm / Hz; R is the communication bandwidth of RF; S302: Calculate the maximum achievable rate value of user m served by the RF AP using the Shannon formula, where the calculation formula is: S303: Calculating the energy efficiency of RF AP serving user m The value is calculated as follows: Where P R Indicates the fixed power of the RF AP. is the power consumed by the RF AP in communicating with user m; Θ0 represents the set of users served by the RFAP; S304: Calculate the signal-to-interference-and-noise ratio value of user m served by VLC AP n, and the calculation formula is: In the formula, is the optical transmission power of VLC AP n; γ is the photoelectric conversion coefficient; is the VLC communication bandwidth of user m; N V is the noise power spectrum density of VLC, which is generally 1×10 -21 A 2 / Hz; S305: Calculate the maximum achievable rate of VLC AP n serving user m, and the calculation formula is: S306: Calculating the energy efficiency of VLC AP n serving user m The value is calculated as follows: Where P V is the fixed power at which the VLC AP operates; is the power consumed by the communication between VLC AP n and user m; S307: Constructing the AP attribute parameter matrix G of candidate VLC APs and RF APs = (g ni ) N×K , where element g ni Defines the parameter value of attribute i that APn can provide. AP attribute parameters include: AP energy efficiency g ne , Link stability g nl , delay g nd and price parameter g np ,in, S308: Calculate the normalized attribute value of AP and construct the normalized attribute decision matrix H of AP = (h ni ) N×K ; Among them, the energy efficiency g of AP n with positive attributes is ne and link stability g nl The normalized property of n∈N, The general formula for calculating the value is: In the formula, and They represent the upper and lower limits of the requirements of service s for AP attribute i respectively; is the intermediate threshold value of s's requirement for AP attribute i. Generally speaking, Among them, for the delay attribute g of AP n with negative attribute nd The normalized property of The calculation formula is: In the formula, and They represent the minimum and maximum values ​​required by service s for AP delay attribute d respectively; is the middle threshold value of the service s's requirement for the AP delay attribute. Generally speaking, Where, for AP n, the network price g nc The normalized attribute of nc , n∈N, the calculation formula is: in, and They represent the price of the network providing user services g nc The lower and upper limits of S309: Calculate the ideal reference value of the AP attribute expected by the user service s; The calculation formula of the ideal reference value of AP attribute i expected by user service s is: S310: Calculate the AP attribute grey correlation value between the attribute of candidate AP n and the ideal reference value of the AP attribute; evaluate the QoE value obtained when the service s requested by user m accesses the candidate AP n according to the comprehensive weight value of the service preference AP and the grey correlation value of the AP attribute; The calculation formula of the AP attribute grey correlation value between the attribute i of candidate AP n and the ideal reference attribute i of AP is: The evaluation formula for the QoE value obtained when the service s requested by user m accesses candidate AP n is: In the formula, Q mn The value is in [0,1], Q mn The higher the value of, the better the QoE experience obtained by user m when accessing AP n. mn When it is 1, candidate AP n is the ideal AP that the user wants to access.

4. The indoor VLC-RF access point selection and resource allocation that meets multi-service QoE requirements as claimed in claim 1, characterized in that: The specific method of S4 is: S401: Determine the matching value ω according to the QoE value mn , with ω mn The weight of the edge between the node m represented by the user and the node n represented by the AP is used to construct a weighted undirected bipartite graph Z of the user access candidate APs. m-n , and mark the required resources of user m requesting service s and the idle available resource values ​​of AP n node next to node m and node n respectively; Among them, the matching value of user m accessing candidate AP n is ω mn The calculation formula is: Where M is the total number of users; Δ m is the expected value of network delay attribute for user m requesting service s, that is: Δ m =a sd ; S402: Calculate the fairness index of the system; Among them, the calculation formula of the system's fairness index is: S403: weighted undirected bipartite graph Z of user access candidate APs m-n In the example, the maximum weight matching method of graph theory is used to select the node n representing the candidate AP with the largest edge weight matching value for each user node m, m∈M, n∈N. If the idle available resource value of the candidate AP connected by the maximum edge weight value is greater than the resource value of the service demand requested by user m, then in the matching result graph Z′ m-n In the equation, the weight Δ m Connect user m and candidate AP n, and m-n Delete user node m, and update the remaining network bandwidth resources of AP n node selected by user m as follows: in, is the idle available resource value of AP n, is the requested resource value of user m, and the process goes to step S407; otherwise, if the remaining resources of other nodes connected by non-maximum edge weights are greater than or equal to the resources requested by user m, user m is saved in the set List u In Z m-n The edge weight ω between user m and candidate AP n is mn Set to 0, and go to step S404; otherwise, block the request connection of user m; S404: If List u Not empty, List u The users whose maximum weight association failed are sorted in ascending order according to their maximum matching value in List u If yes, go to step S405; otherwise, go to step S406; S405: Select List u For the first user, go to step S403; S406: Calculate the fairness index of the system. If I F ∈(0.7,1]?n∈N, go to step 407; otherwise, in Z′ m-n In the example, we select the edge weight Δ m The user m with the largest value is deleted from Z′ m-n The corresponding edge in , go to step S403; S407: Output matching result graph Z′ m-n The matching result of the user request service selection AP with edge connection.

Images