A QoS-QoE driven VLC-RF heterogeneous network energy efficiency optimization method

Abstract

The application discloses a QoS-QoE driven VLC-RF heterogeneous network energy efficiency optimization method, comprising the following steps: acquiring the physical layer transmission rate of a VLC access point and an RF access point in a VLC-RF heterogeneous network; mapping the system physical layer transmission rate to a link layer transmission rate, and modeling the system service process according to the link layer transmission rate. The super-martingale theory is introduced to evaluate the network delay QoS performance, and a more compact delay violation probability bound is obtained; the throughput and total power of the system are calculated, and a heterogeneous network QoE model is constructed. An optimization problem is constructed with the maximum energy efficiency as the target and the delay violation probability bound and user experience guarantee as the constraint condition; the maximum energy efficiency optimization problem is solved through an improved golden jackal optimization algorithm, and the optimal energy efficiency value and the transmission power of the VLC-RF heterogeneous network system are obtained.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of communication, and particularly relates to a QoS-QoE driven VLC-RF heterogeneous network energy efficiency optimization method. BACKGROUND

[0002] Traditional Radio Frequency (RF) communication technology is faced with resource bottleneck and interference problem, and Visible Light Communication (VLC) has the advantages of high transmission rate, low interference, rich spectrum resources, etc., and can be a powerful supplement to RF communication technology. Full use of the technical advantages of VLC and RF, and fusion of the two technologies to form a VLC-RF heterogeneous network, can not only overcome the limitation of RF spectrum resources and provide high-reliable data transmission, but also provide better area coverage, and has broad application prospects.

[0003] Communication users are experiencing a change from focusing on network Quality of Service (QoS) to emphasizing Quality of Experience (QoE). At the same time, in the service process of a heterogeneous network, energy efficiency is a problem that needs to be focused on. Therefore, designing a QoS-QoE driven VLC-RF heterogeneous network energy efficiency optimization method has important significance for effectively improving the service efficiency of a communication network. SUMMARY

[0004] The application provides a QoS-QoE driven VLC-RF heterogeneous network energy efficiency optimization method to solve the problems in the prior art.

[0005] To achieve the above purpose, the application provides a QoS-QoE driven VLC-RF heterogeneous network energy efficiency optimization method, which comprises the following steps:

[0006] Obtaining the physical layer transmission rate of a VLC access point and an RF access point in a VLC-RF heterogeneous network; mapping the physical layer transmission rate to a link layer transmission rate, modeling the service process of the heterogeneous network according to the link layer transmission rate, evaluating the network delay QoS performance of the service process of the heterogeneous network through the super-martingale theory, and obtaining the delay violation probability bound;

[0007] Calculating the throughput and total power of the heterogeneous network according to the link layer transmission rate, and constructing a QoE model of a file download service of the heterogeneous network;

[0008] Constructing an energy efficiency maximization optimization problem based on the maximum energy efficiency as an objective and the QoE model of the file download service of the heterogeneous network as a constraint condition;

[0009] The energy efficiency maximization optimization problem is solved by improving the golden wolf optimization algorithm, and optimal energy efficiency and transmission power of the VLC-RF heterogeneous network system are obtained.

[0010] Optionally, the obtaining process of the physical layer transmission rate comprises:

[0011] The channel gain between the VLC access point, the RF access point and the user terminal, and the signal-to-noise ratio of the user terminal to each access point are obtained respectively; and the physical layer transmission rate is calculated according to the channel gain and the signal-to-noise ratio.

[0012] Optionally, the conversion process of the link layer transmission rate comprises:

[0013]

[0014]

[0015] wherein, R R RF respectively represent the physical layer transmission rate of the VLC system and the RF system, T represents the duration of a time slot, and L represents the length of a data packet.

[0016] Optionally, the modeling process of the heterogeneous network service process comprises:

[0017] The rate of the instantaneous service process of the heterogeneous network and the service martingale of the heterogeneous network are constructed.

[0018] wherein, the rate of the instantaneous service process is:

[0019]

[0020] wherein, s m (n) represents the rate of the instantaneous service process of the heterogeneous network system at time slot n, β represents the blocking probability of the terminal in the time slot, and M represents the number of VLC access points.

[0021] The service martingale M (s,m) (n) of the heterogeneous network is:

[0022]

[0023] wherein, s m (n) and Sm(n) respectively represent the instantaneous service and the cumulative service of the system at time slot n, (θ m represents a QoS parameter, K (s,m) (θ m ) is a supermartingale correction function of the service process.

[0024] Optionally, the delay violation probability boundary of the martingale domain is:

[0025]

[0026] where h (s,m) is the correlation function of the service process, h (u,m) is the correlation function of the arrival process, is the martingale parameter that links the arrival and service martingale parameters, h (s,m) and h (u,m) values are dependent on H m denotes the threshold value, D m is the target delay, denotes the initial value expectation of the mth link service process and the correlation function of the arrival process, u m (n) denotes the instantaneous arrival packet amount, U is the arrival rate, ε m is the delay violation probability threshold value.

[0027] Optionally, the throughput acquisition process is:

[0028]

[0029] where T denotes the duration of a time slot, L denotes the length of a packet, is the transmission rate between the VLC system and the RF system and the user terminal, β denotes the blocking probability of the terminal in a time slot, and M denotes the number of VLC access points.

[0030] Optionally, the heterogeneous network file download service QoE model is:

[0031]

[0032] where ω and υ are weight coefficients determined by the maximum and minimum values of the user throughput, C m denotes the throughput, and MOS m denotes the QoE measurement value.

[0033] Optionally, the energy efficiency maximization optimization model is:

[0034]

[0035] where EE denotes the energy efficiency, ω and υ are weight coefficients, T denotes the duration of a time slot, L denotes the length of a packet, is the transmission rate between the VLC system and the RF system and the user terminal, β denotes the blocking probability of the terminal in a time slot, and M denotes the number of VLC access points, h (s,m) is the correlation function of the service process, h (u,m) is the correlation function of the arrival process, is a special martingale parameter that links the arrival and service martingale parameters, h (s,m) and h (u,m) values are dependent on H m denotes a threshold value, D m is the target delay, denote the initial value expectation of the mth link service process and the arrival process related characteristic function, respectively, u m (n) denotes the instantaneous arrival packet amount, U is the arrival rate, and ε m is the delay violation probability threshold value, is the transmission power of the VLC AP to the mth user terminal; p RF denotes the transmission power of the RF AP to the user, and MOS m denotes the QoE metric value, and MOS min is the minimum QoE requirement of the user, and m denotes the user terminal label, and M denotes the total amount of user terminals.

[0036] Optionally, the process of solving the energy efficiency maximization optimization problem by improving the golden wolf optimization algorithm includes the following steps:

[0037] initializing a population, wherein the individual positions in the population are initialized by a Logistic-tent chaotic mapping;

[0038] calculating the fitness of the individual positions, and selecting the individual with the minimum fitness value as the population optimal male golden wolf and the individual with the suboptimal fitness as the population optimal female golden wolf;

[0039] updating the individual positions in the population by adding a positive cosine search algorithm;

[0040] updating the individual positions in the population again by an adaptive t mutation strategy;

[0041] based on the fitness of the individual positions, determining whether to adopt the individual positions in the population updated again by the adaptive t mutation strategy by a greedy rule;

[0042] iterating the initialized operation until a maximum iteration number is reached, and then obtaining the optimal energy efficiency value and the transmission power of each VLC and RF in the VLC-RF heterogeneous network system.

[0043] Compared with the prior art, the present application has the following advantages and technical effects:

[0044] The research on the energy efficiency optimization of the VLC-RF heterogeneous communication system can generate valuable results in network resource management and provide theoretical and technical support for future heterogeneous communication network service resource management. BRIEF DESCRIPTION OF DRAWINGS

[0045] The accompanying drawings, which form a part of the present application, are intended to provide further understanding of the present application and are incorporated herein for a purpose of explanations and are not for a purpose of limiting the present application. In the drawings:

[0046] Figure 1 The method flowchart of the embodiment of the present application. DETAILED DESCRIPTION

[0047] It should be noted that the embodiments and the features in the embodiments in the present application can be combined with each other without conflicts. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0048] It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a group of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described herein can be executed in an order different from that shown herein.

[0049] The present application proposes a QoS-QoE driven VLC-RF heterogeneous network energy efficiency optimization method, which is oriented to a VLC-RF heterogeneous network, derives the channel gain, signal-to-noise ratio and communication rate of a visible light communication system and a radio frequency communication system respectively, and models a service process by considering the shielding characteristics of the system. Then, the super-martingale theory is introduced to evaluate the network delay QoS performance, and a more compact delay violation probability bound is obtained. The throughput and total power of the system are calculated, and the user throughput is mapped into the quality of experience QoE level, and the MOS mapping function is used to measure the user QoE demand. The optimization problem is constructed by taking the maximum system energy efficiency as the target, the Logistic-tent chaotic mapping, the positive sine search algorithm and the adaptive t mutation strategy are combined to improve the golden eagle optimization algorithm, and the improved algorithm is used to solve the maximum system energy efficiency problem. The method proposed in the present application can maximize the energy efficiency under the premise of guaranteeing the QoS requirement and improving the user QoE level, so as to improve the utilization rate of system resources.

[0050] The present application provides a QoS-QoE driven VLC-RF heterogeneous network energy efficiency optimization method, which can realize efficient configuration of network service resources, improve energy efficiency under the premise of improving user experience, and aims to provide reliable theoretical and technical support for heterogeneous network operation.

[0051] In order to achieve the above technical purposes and achieve the above technical effects, the technical scheme of the present application comprises the following steps:

[0052] Step (1): For indoor VLC-RF heterogeneous network, a system model of M users and M VLC Access Points (APs), one RF AP is established. Each user is equipped with a photoelectric sensor receiver and corresponds to a VLC AP. The coverage of the VLC AP is limited and will be affected by physical obstacles, which will affect the signal strength and quality. The RF network has strong diffraction ability and the signal can cover the entire room.

[0053] Step (2): For the heterogeneous network communication scenario, the channel gain, signal-to-noise ratio and physical layer transmission rate between the VLC AP and the user are calculated respectively. The services of the physical layer in the VLC system and the RF system are mapped to the services of the data link layer, and the service process is modeled considering the blocking characteristics of the system. Based on the supermartingale theory, the martingale process of system analysis is constructed, and the delay violation probability boundary of the martingale domain is derived.

[0054] Step (3): The rate of the service process of the data link layer in the VLC and RF systems is solved, and then the throughput and total power of the system are calculated. The user throughput is mapped to the QoE level and the MOS mapping function is used to measure the user QoE demand, and the QoE model of the file download service in the heterogeneous network is constructed.

[0055] Step (4): Taking the system delay requirement, user QoE and system transmission power as constraint conditions, and maximizing the energy efficiency as the objective function, an optimization problem is constructed.

[0056] Step (5): The Logistic-tent chaotic mapping is used to initialize the golden jackal prey population, the improved golden jackal optimization algorithm is obtained by updating the population position using the sine cosine search algorithm and the adaptive t mutation strategy, and the improved algorithm is used to solve the optimization problem to maximize the system energy efficiency.

[0057] The above steps are described in detail as follows:

[0058] In step (1), for indoor VLC-RF heterogeneous network, a system model of M users and M VLC Access Points (APs), one RF AP is established. Each user is equipped with a photoelectric sensor receiver and corresponds to a VLC, the VLC AP will be affected by physical obstacles, which will affect the signal strength and quality, and the RF network covers the entire room.

[0059] Step (2) Based on the VLC-RF heterogeneous network model, the channel gain, signal-to-noise ratio and rate between VLC AP and RF AP and user are calculated respectively. In order to guarantee the delay QoS requirement of VLC-RF system, the physical layer service process of VLC system and RF system is converted into the service process of link layer, considering the blocking characteristics of the system, the service process of the system is established and the service martingale of the system is constructed, and the delay QoS performance is obtained by the super-martingale stopping time theorem.

[0060] Step (2.1) The channel gain between VLC AP and user, the signal-to-noise ratio of user to each access point, and then the communication rate (physical layer transmission rate) of VLC system are calculated.

[0061] Step (2.2) The channel gain between RF AP and user, the signal-to-noise ratio of user to each access point, and then the communication rate (physical layer transmission rate) of RF system are calculated.

[0062] Step (2.3) The achievable rate in "bits / s" is converted into the rate in "packets / slot". Let T [s / slot] represent the duration of the time slot, and L [bits / packet] represent the length of the data packet. For terminal m, the converted link layer transmission rate of VLC system and RF system are calculated respectively :

[0063]

[0064] wherein, R RF represent the physical layer transmission rate of VLC system and RF system respectively.

[0065] Step (2.4) In order to guarantee the delay QoS requirement of VLC-RF system, the service process of VLC-RF system network is established according to the converted transmission rate of step (2.3).

[0066] Step (2.4.1) When the direct link of VLC occurs blocking, the system will switch to RF network, and the instantaneous service process rate of the terminal can be represented as:

[0067]

[0068] wherein β represents the blocking probability of terminal in time slot, and M represents the number of VLC access points.

[0069] Step (2.4.2) The service martingale of the heterogeneous network system is constructed, and then the delay violation probability bound is obtained by the super-martingale stopping time theorem.

[0070] Supermartingale M of system service process (s,m) (n) is:

[0071]

[0072] wherein s m (n) and Sm(n) represent instantaneous and cumulative service of the system at time slot n respectively, (θ m represents QoS parameter, K (s,m) (θ m ) is a supermartingale correction function of the service process, and is obtained by the following formula:

[0073]

[0074] For the arrival process, a supermartingale M of the arrival process is constructed (u,m) (n) is:

[0075]

[0076] wherein u m (n) and U m (n) represent instantaneous and cumulative arrival packet amount. The present application adopts constant rate arrival, and the arrival rate is U, so that h (u,m) (u m (n)) = 1.

[0077] A supermartingale M relative to the queue length is constructed (L,m) (n) as follows:

[0078]

[0079] According to the supermartingale stop-time theorem, the delay violation probability bound can be derived as:

[0080]

[0081] wherein d m (n) represents delay, D m is target delay, respectively represent initial value expectations of the mth link service process and the arrival process related characteristic functions.

[0082] is a special martingale parameter that links the arrival martingale and the service martingale parameters, and is determined by the following formula:

[0083]

[0084] Threshold H m is determined by the following formula:

[0085] H m:=min{(h (u,m) (u m (n)))(h (s,m) (s m (n)):u m (n)-s m (n)>0}

[0086] To ensure the latency QoS requirements of all terminals in the system, the following inequality must be maintained:

[0087] P m {d m (n)≥D m}≤ε m

[0088] In the above formula, ε m This is the threshold for the probability of delay violation.

[0089] The martingale domain delay violation probability bound obtained in this patent is as follows:

[0090]

[0091] Step (3) In the VLC-RF heterogeneous network, calculate the network system throughput and total power based on the rates of the VLC and RF systems. Establish a MOS model for file download services in the heterogeneous network, map user throughput to QoE levels, and use the MOS mapping function to measure user QoE requirements.

[0092] Step (3.1) Calculate the system throughput C m The total power P is:

[0093]

[0094] Where T[s / slot] represents the duration of the time slot, and L[bits / packet] represents the length of the data packet. p is the transmission power from the VLC AP to user m. RF p represents the transmission power from the RF AP to the user. c It is the power loss of other components in the transmitting circuit.

[0095] Step (3.2) User experience quality describes the overall performance of the network from the user's perspective, and directly reflects the degree of user's approval of network services. Compared with QoS, QoE can better measure user satisfaction. In order to measure the transformation of user QoE level, the QoE metric value Mean Opinion Score (MOS) is used to measure the user's QoE. According to the user's satisfaction, the subjective feeling of QoE is divided into five levels: "excellent", "good", "medium", "poor" and "very poor". Mapping the throughput to the QoE level, the MOS mapping function can be used to measure the user QoE demand.

[0096] File download service is a non-real-time service commonly used in wireless network services, which realizes the fast transmission of data files through file transfer protocol. The core feature of this service is that the user's service experience is affected by the wireless network transmission rate, and the user's expectation of file download service is that the higher the rate is, the better. The QoE model of file download service is constructed to obtain the mapping relationship from user throughput to MOS function, and the file download service QoE model can be described as:

[0097]

[0098] Where ω, υ are weight coefficients, which are determined by the maximum and minimum values of user throughput. If the user receives the throughput C m = C max , then the user's satisfaction on MOS should be the maximum, that is, 5. On the other hand, a minimum throughput is defined and given a MOS value of 1.

[0099] Step (4) measures the satisfaction of QoE from the user's perspective, and derives the energy efficiency of the user based on QoE. The energy efficiency maximization optimization problem under VLC-RF network is constructed. The optimization problem is expressed as:

[0100]

[0101] Where constraint condition C1 represents the delay violation probability boundary; conditions C2-C3 represent the total power constraints of VLC and RF systems respectively; conditions C4-C5 guarantee the non-negative power requirements of VLC and RF. Condition C6 is the minimum QoE limit of the user, MOS min is the minimum QoE requirement of the user, MOS min Generally takes the value of 3.

[0102] Step (5) design improved Goldenjackal optimization (GJO) algorithm and solve the optimization problem, introduce three major improvements, by introducing Logistic-tent chaotic mapping to initialize the prey population, add positive sine search algorithm and adaptive t mutation strategy to update the population position. Through the improved Goldenjackal algorithm optimization solves the step (4) constructs the martingale energy efficiency maximization optimization problem under the VLC-RF network, its specific steps are.

[0103] Step (5.1) in the algorithm initialization population stage, initialize the Goldenjackal population parameters, use Logistic-tent chaotic mapping to initialize the population position matrix, wherein the population position matrix represents the energy efficiency value and the transmission power of the VLC-RF heterogeneous network system in the optimization process. The function expression of Logistic-tent mapping to generate chaotic particle sequence is:

[0104]

[0105] Wherein, r is a related control factor, when r ∈ (0, 1), x i ∈ (0, 1), the chaotic function is in a chaotic state. Mod represents the modulo operation. Through the modulo operation, the distribution trajectory range of the chaotic mapping is expanded.

[0106] Step (5.2) the objective function is the maximum value of the martingale energy efficiency under the VLC-RF network. According to the objective function, the fitness value of the Goldenjackal position in the population is calculated, and the individual with the minimum fitness value is selected as the optimal male Goldenjackal in the population, and the individual with the second minimum fitness value is selected as the optimal female Goldenjackal in the population.

[0107] Step (5.3) add positive sine search algorithm to update the position of the population in the process of Goldenjackal searching for prey, use the information of the Goldenjackal pair obtained in the last iteration, update the value of the candidate solution in each dimension according to the sine and cosine functions, and the cyclic pattern of the positive sine function only allows one solution to be relocated around other solutions, which ensures that the search can be carried out in the space between two solutions, which can improve the convergence speed of global convergence. The improved search formula is:

[0108]

[0109] In the formula, Indicates the position of the i-th individual in the space in the t+1 iteration; Indicates the position of the optimal Goldenjackal population obtained in the t iteration. R2 ∈ [0, 2π], r3 ∈ [0, 2], r4 ∈ [0, 1] are random numbers, r1 is a linear decreasing function, which indicates the position relationship between the next solution and the current solution and the optimal solution, and the expression is:

[0110]

[0111] Where a is a constant.

[0112] In step (5.4), during the later stages of algorithm iteration, to address the issues of the algorithm getting trapped in local optima and low convergence accuracy, an adaptive t-mutation strategy is used to improve the position update formula for the golden jackals in the population. This enhances the algorithm's local optimization ability and improves convergence accuracy. The position update formula is as follows:

[0113] X b ' est =X best +X best *trnd(t)

[0114] In the formula, X best Indicates the current optimal position for the golden jackal; X b ' est This indicates the position of the golden jackal after adaptive t-mutation perturbation.

[0115] In step (5.5), after the mutation update, in order to compare the quality of the golden jackal's position before and after the mutation, a greedy rule is introduced to determine whether to use the position updated by mutation.

[0116]

[0117] Step (5.6) determines whether the population has reached the maximum number of iterations. If so, output the optimal energy efficiency value and transmit power of the VLC-RF heterogeneous network system; otherwise, return to step (5.2) to continue the optimization.

[0118] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.

[0119] like Figure 1 As shown, step (1) constructs a heterogeneous network scenario containing M users, M VLC APs, and one RF AP. The set of users is m = {1, 2, ..., M}. Each VLC AP is placed on the ceiling, and the RF AP is fixed on the wall. Each user is equipped with a photoelectric sensor receiver and corresponds to one VLC AP. The coverage of the VLC AP is limited and its signal strength and quality are affected by physical obstacles. The RF network has strong diffraction capabilities and its signal can cover the entire room. Since the VLC system and the RF system are two different systems, the VLC channel will not interfere with the RF channel.

[0120] Step (2) calculates the channel gain, signal-to-noise ratio and physical layer rate between VLC AP and user, respectively. The service mapping of physical layer in VLC system and RF system is mapped to the service of data link layer, and the service process is modeled considering the blocking characteristics of the system. Based on the supermartingale theory, the martingale process of system analysis is constructed, and the delay violation probability boundary of the martingale region is derived.

[0121] Step (2.1) calculates the channel gain between VLC AP and user.

[0122] The method provided by the application is general, and the following is an example of user m. The system gain of the VLC system in the LOS case is considered. According to the Lambert radiation model, the channel gain of the mth user and the VLC AP is:

[0123]

[0124] In the above formula, b represents the distance from the user to the VLC access point, ψ m is the incident angle of the user optical detector, A m is the physical area of the optical detector, φ m represents the radiation angle of the LED, g(ψ m ) represents the optical filter gain of the receiver, the Lambert order is represented by ξ, and the expression is as follows: θ 12 represents the half-power angle of the VLC AP. f(ψ m ) is the gain of the light collector, that is,

[0125]

[0126] ψ FOV is the field of view (FOV) of the user optical detector, and l is the refractive index.

[0127] Step (2.2) calculates the channel gain between RF AP and user.

[0128] The path loss model of the RF network usually has the following form:

[0129]

[0130] Where S represents the path loss, b0 represents the distance, and the path loss exponent is η. The RF channel gain is G RF = 10 -S / 10 .

[0131] Step (2.3) calculates the physical layer data rate of the VLC system and the RF system respectively by using the Shannon capacity formula as follows:

[0132]

[0133] where, and p RF denote the transmission power of the VLC system and the RF system, respectively, and B RF denote the frequency bandwidth of the VLC system and the RF system, respectively, denote the noise power spectral density of the received signal by the VLC receiver and the RF receiver, respectively.

[0134] Step (2.4) converts the achievable rate in "bits / s" to the heterogeneous network transmission rate in "packets / slot". Let T [s / slot] denote the duration of a time slot, and L [bits / packet] denote the length of a data packet. The transmission rates of the VLC system and the RF system are:

[0135]

[0136]

[0137] Step (2.5) constructs the system service model. When none of the VLC links is obstructed, the VLC AP serves the users with the transmission rate ; when only one of the M VLC APs is obstructed, the RF AP serves the users with the transmission rate ; otherwise, the users are not served. Then, the instantaneous service process of the heterogeneous network can be expressed as:

[0138]

[0139] where β denotes the obstruction probability of the system.

[0140] Step (2.6) derives the system's probability bound of the delay violation in the ergodic region. Based on the supermartingale theory, the supermartingale of the system's service process is:

[0141]

[0142] where s m (n), S m (n) denote the instantaneous and cumulative service of the system at time slot n, respectively, and K (s,m) (θ m ) is the supermartingale correction function of the service process, which is obtained by:

[0143]

[0144] The system's data packet arrival process is a constant rate arrival with the arrival rate U. The supermartingale of the system's arrival process is:

[0145]

[0146] Wherein, um(n), Um(n) represent instantaneous, cumulative arrival data packet amount. The present application takes h um (u m (n))=1, K (u,m) (θ m ) for the supermartingale correction function of arrival process, K (u,m) (θ m ) the value of constant U.

[0147] The service martingale and arrival martingale are multiplied to obtain the supermartingale relative to the queue length, as follows:

[0148]

[0149] According to the supermartingale stopping time theorem, the delay violation probability bound can be derived as follows:

[0150]

[0151] Wherein, D m is the target delay, respectively represent the initial value expectation of the mth link service process, arrival process related characteristic function.

[0152] The value of θ m is set to

[0153]

[0154] H m is determined by the following formula:

[0155] H m :=min{(h (u,m) (u m (n)))(h (s,m) (s m (n))):u m (n)-s m (n)>0}

[0156] In order to guarantee the delay QoS requirements of each terminal of the system, the following inequality needs to be guaranteed:

[0157] P m {d m (n)≥D m}≤ε m

[0158] In the above formula, ε m is the delay violation probability threshold.

[0159] Therefore, the martingale region delay violation probability bound can be expressed as:

[0160]

[0161] Step (3) Service rate of VLC and RF systems, then calculate the throughput and total power of the system. Map user throughput to QoE levels and use MOS mapping function to measure user QoE requirements, and build a MOS model for file download service in heterogeneous networks.

[0162] Step (3.1) Calculate the throughput and total power of the system:

[0163]

[0164] Where T [s / slot] represents the duration of a time slot, and L [bits / packet] represents the length of a data packet. is the transmission power of the VLC AP to user m; p RF is the transmission power of the RF AP to user m; p c is the loss power of other components in the transmitting end circuit.

[0165] Step (3.2) Build a file download service QoE model. File download service is a commonly used non-real-time service in wireless network service, which realizes fast transmission of data files through file transfer protocol. The core feature of this service is that the user's service experience is affected by the transmission rate of the wireless network, and the user's expectation of the file download service is that the higher the rate, the better. User experience quality describes the overall performance of the network from the user's perspective, and directly reflects the user's recognition of the service. In order to measure the transformation of user QoE levels, the Mean Opinion Score (MOS) is used as a performance indicator. According to user satisfaction, the subjective experience of QoE is divided into five levels: excellent, good, medium, poor and very poor.

[0166] The specific quantification method of MOS value is shown in Table 1:

[0167] Table 1

[0168]

[0169] By mapping the throughput to the QoE level, the MOS mapping function can be used to measure the user QoE requirements.

[0170]

[0171] Where ω, υ are determined by the maximum and minimum values of user throughput.

[0172] Step (4) constructs an optimization problem aiming at maximizing the system energy efficiency. Unlike the traditional definition of energy efficiency, the present application measures the satisfaction of QoE from the user's perspective, and defines the energy efficiency as the ratio of the total MOS value of all users in the system to the total power consumed in the system, i.e. maximizing the average opinion score per unit power consumption. The optimization problem of maximizing the energy efficiency of the martingale under the VLC-RF network is constructed. The optimization problem is expressed as:

[0173]

[0174] Where constraint condition C1 represents the delay violation probability boundary; conditions C2-C3 represent the total power constraints of the VLC and RF systems respectively; conditions C4-C5 guarantee the non-negative power requirements of the VLC and RF; condition C6 is the minimum QoE limit for users, MOS min is the minimum QoE requirement for users, MOS min Generally takes a value of 3.

[0175] Step (5) improves the golden jackal algorithm by introducing a Logistic-tent chaotic mapping to initialize the prey population, adding a positive sine search algorithm and an adaptive t mutation strategy to update the population position, and uses the improved golden jackal algorithm to optimize and solve the optimization problem of maximizing the energy efficiency of the martingale under the VLC-RF network constructed in step (4). The specific steps are as follows:

[0176] Step (5.1) initializes the golden jackal population parameters in the algorithm initialization population stage, including the number of male and female golden jackals in the population and the dimension of the population individuals, sets the maximum number of population iterations T, and initializes the population individual position using the Logistic-tent chaotic mapping. The function expression of the Logistic-tent mapping to generate chaotic particle sequences is as follows:

[0177]

[0178] Where r is a related control factor, when r ∈ (0, 1), the individual position x i ∈ (0, 1), the chaotic function is in a chaotic state. The mod represents the modulo operation. Through the modulo operation, the distribution trajectory range of the chaotic mapping is expanded.

[0179] Step (5.2) the objective function is the maximum energy efficiency of the martingale under the VLC-RF network. According to the objective function, the fitness value of the golden jackal position in the population is calculated, and the individual with the minimum fitness value is selected as the optimal male golden jackal in the population, and the individual with the second minimum fitness value is selected as the optimal female golden jackal in the population.

[0180] Step (5.3) adds the positive sine-cosine search algorithm to update the position of the population in the process of searching for prey by the golden jackal. The golden jackal pair information obtained in the last iteration is used to update the value of the candidate solution in each dimension according to the sine and cosine functions, and the cyclic pattern of the positive sine-cosine function only allows one solution to be relocated around other solutions, ensuring that searching in the space between two solutions can improve the convergence speed of global convergence. The improved search formula is:

[0181]

[0182] wherein, represents the position of the ith individual in space in the t+1th iteration; represents the position of the optimal golden jackal population obtained in the tth iteration. r2∈[0,2π], r3∈[0,2], r4∈[0,1] are random numbers, and r1 is a linear decreasing function, which represents the position relationship between the next solution and the current solution and the optimal solution. The specific expression is:

[0183]

[0184] wherein a is a constant.

[0185] Step (5.4) In the later stage of algorithm iteration, in order to solve the problem of the algorithm falling into local optimal value and low convergence precision, the adaptive t mutation strategy is used to improve the position update formula of the golden jackal in the population, enhance the local optimization ability of the algorithm, and improve the convergence precision. The position update formula is:

[0186] X b ' est =X best +X best *trnd(t)

[0187] wherein, X best represents the current optimal position of the golden jackal; X b ' est represents the golden jackal position after adaptive t mutation disturbance, and trnd(t) is the T distribution about the iteration number t.

[0188] After mutation update, in order to compare the good and bad of the golden jackal position before and after mutation, whether to adopt the position of mutation update is determined according to the greedy rule, and f(·) is the fitness function.

[0189]

[0190] Step (5.5) judges whether the population reaches the maximum iteration number. If not, return to step (5.2) to continue optimization, and the current iteration number t=t+1 at the same time. Otherwise, the algorithm ends, and the optimal solution energy efficiency value and transmission power of the system are output.

[0191] By using the improved algorithm to solve the optimization problem, the user transmits data according to the solved power, energy efficiency maximization can be realized under the premise of guaranteeing QoS requirements and improving user QoE level, and system resource utilization is improved.

[0192] The above is only a preferred specific embodiment of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A QoS-QoE driven VLC-RF heterogeneous network energy efficiency optimization method, characterized in that, The method comprises the following steps: obtaining the physical layer transmission rate of the VLC access point and the RF access point in the VLC-RF heterogeneous network; mapping the physical layer transmission rate to the link layer transmission rate, modeling the heterogeneous network service process according to the link layer transmission rate, evaluating the network delay QoS performance of the heterogeneous network service process through the super-martingale theory, and obtaining the delay violation probability bound; calculating the throughput and the total power of the heterogeneous network according to the link layer transmission rate, and constructing a heterogeneous network file download service QoE model; constructing an energy efficiency maximization optimization problem based on the maximum energy efficiency as the target, the delay violation probability bound and the heterogeneous network file download service QoE model for maintaining the user experience guarantee as the constraint condition; solving the energy efficiency maximization optimization problem through the improved golden cat optimization algorithm to obtain the optimal energy efficiency value and the transmission power of the VLC-RF heterogeneous network system; the conversion process of the link layer transmission rate comprises: wherein, are the link layer transmission rates for the VLC system and the RF system, respectively, and R RF denote the physical layer transmission rates for the VLC system and the RF system, respectively, T denotes the duration of a time slot, and L denotes the length of a data packet. the process of modeling the heterogeneous network service process comprises: constructing the rate of the instantaneous service process of the heterogeneous network and the service martingale of the heterogeneous network; wherein the rate of the instantaneous service process is: where s m (n) denotes the rate of the instantaneous service process of the heterogeneous network system at time slot n, β denotes the blocking probability of the terminal in the time slot, and M denotes the number of VLC access points. Service martingale M of heterogeneous network (s,m) (n) is: where s m (n), S m (n) represent the instantaneous, cumulative service of the system at time slot n, respectively, θ m denotes the QoS parameter, K (s,m) (θ m ) is the supermartingale correction function of the service process; the heterogeneous network file download service QoE model is: where ω,υ are weight coefficients determined by the maximum and minimum values of user throughput, C m denotes throughput, MOS m denotes QoE metric value; the energy efficiency maximization optimization problem is: C4: 0 < p RF Where EE represents energy efficiency, ω and υ are weighting coefficients, T represents the duration of the time slot, and L represents the length of the data packet. = represents the transmission rate between the VLC system and the RF system and the user terminal, β represents the probability of the terminal being blocked in the time slot, M represents the number of VLC access points, and D m For target latency, Let u represent the initial expected values ​​of the characteristic functions related to the service process and arrival process of the m-th link, respectively. m (n) represents the instantaneous number of data packets arriving, h (s,m) These are the relevant functions of the service process, h (u,m) It is a function related to the arrival process. h is a martingale parameter that links the arrival and service martingale parameters. (s,m) and h (u,m) Values ​​all depend on H m The threshold is represented by ε, U is the arrival rate, and ε is the threshold value. m For the delay violation probability threshold, p represents the transmission power from the VLC AP to the m-th user terminal. RF Indicates the transmission power from the RF AP to the user, MOS m MOS represents the QoE metric. min The minimum QoE requirement for users is given by m, which represents the user terminal number, and M represents the total number of user terminals.

2. The method of claim 1, wherein, the process of obtaining the physical layer transmission rate comprises: respectively obtaining the channel gain between the VLC access point, the RF access point and the user terminal, and the signal-to-noise ratio of the user terminal to each access point; and calculating the physical layer transmission rate according to the channel gain and the signal-to-noise ratio.

3. The method of claim 1, wherein, the delay violation probability bound is: where h (s,m) is the correlation function of the service process, h (u,m) is the correlation function of the arrival process, is the martingale parameter that links the arrival and service martingale parameters, h (s,m) and h (u,m) values are dependent on H m denotes the threshold, D m is the target delay, denotes the initial expected value of the mth link service process, arrival process correlation function, u m (n) denotes the instantaneous arrival packet amount, U is the arrival rate, ε m is the delay violation probability threshold.

4. The method of claim 1, wherein, the process of obtaining the throughput is: where T represents the duration of a time slot, L represents the length of a data packet, for the transmission rate between the VLC system and the RF system and the user terminals, β represents the blocking probability of the terminals in a time slot, and M represents the number of VLC access points.

5. The method of claim 1, wherein, the process of solving the energy efficiency maximization optimization problem through the improved golden cat optimization algorithm comprises: initializing the population, wherein the individual positions in the population are initialized through the Logistic-tent chaotic mapping; calculating the fitness of the individual positions, and selecting the individual with the minimum fitness value as the optimal male golden cat and the individual with the suboptimal fitness as the optimal female golden cat; updating the individual positions in the population by adding the positive sine search algorithm; updating the individual positions in the population again through the adaptive t mutation strategy; based on the fitness of the individual positions, determining whether to adopt the individual positions in the population updated again through the adaptive t mutation strategy through the greedy rule; iterating the initialized operation until the maximum iteration number is reached, and then obtaining the optimal energy efficiency value and the transmission power of each VLC and RF in the VLC-RF heterogeneous network system.

Images