A dynamic resource optimization allocation method for VLC / mmWave / RF heterogeneous networks

Abstract

The present invention discloses a method for dynamic resource optimization allocation of VLC / mmWave / RF heterogeneous networks, which relates to the field of resource allocation in heterogeneous networks, and includes the following steps: S1: establishing a mathematical model based on indoor hybrid VLC, millimeter channel mmWave and radio frequency RF heterogeneous network scenarios; S2: establishing a joint optimization problem based on the mathematical model established in step S1; S3: optimizing the problem transformation, using the Lyapunov optimization method to transform the long-term optimization problem into a series of single-time slot deterministic optimization problems; S4: using the penalty dual decomposition algorithm to solve the single-time slot deterministic optimization problem, and realizing the optimal allocation of resources for each time slot. The present invention adopts the above-mentioned method for dynamic resource optimization allocation of VLC / mmWave / RF heterogeneous networks to solve the problem of limited transmission rate under the constraints of wireless spectrum resources.

Description

Technical Field

[0001] The present invention relates to the field of resource allocation in heterogeneous networks, and in particular to a dynamic resource optimization allocation method for VLC / mmWave / RF heterogeneous networks. Background Art

[0002] With the rapid development of the Internet of Things and the Industrial Internet, the number of mobile devices connected to the Internet has increased dramatically, and the demand for wireless network traffic has risen rapidly. With more and more Internet-enabled electronic devices and content-rich multimedia applications, wireless networks are unable to cope with the exponential growth of data traffic. The rapid growth of mobile traffic has put forward higher requirements for current 5G and other networks, such as shorter latency, higher communication rates, and better network coverage. Traditional radio frequency (RF) systems mainly focus on frequency bands below 10 GHz, and limited spectrum resources can no longer meet people's explosive data demand for wireless spectrum.

[0003] 5G wireless communications will achieve extremely high capacity, high data rates, high spectral efficiency, high energy efficiency, and stringent user experience. The spectrum allocated to 2G / 3G / 4G systems is mainly concentrated below 3 GHz, and most of it is currently occupied. The higher frequency bands targeted by 5G communications are essential to meet the growing performance requirements. Millimeter wave (mmWave) spectrum and visible light (VLC) spectrum have become promising candidate technologies for 5G networks due to their abundant available spectrum at very high frequencies, which can provide very high capacity and gigabits per second of data transmission, especially in indoor environments. High data rate services indoors and in hotspots will lead to a surge in 5G network traffic. In order to meet the future demand for mobile data traffic, wireless networks are becoming hybrid networks by combining more than two radio access technologies (RATs).

[0004] Therefore, it is necessary to provide a dynamic resource optimization allocation method for VLC / mmWave / RF heterogeneous networks to solve the above problems. Summary of the invention

[0005] The purpose of the present invention is to provide a dynamic resource optimization allocation method for a VLC / mmWave / RF heterogeneous network to solve the problems of limited spectrum resources and low transmission rate.

[0006] To achieve the above object, the present invention provides a method for dynamic resource optimization allocation of a VLC / mmWave / RF heterogeneous network, comprising the following steps:

[0007] S1: Establish a mathematical model based on indoor hybrid VLC, mmWave and RF heterogeneous network scenarios;

[0008] S2: Establish a joint optimization problem according to the mathematical model established in step S1;

[0009] S21: Establishing a transmission rate expression between the RF access point RF-AP and the user;

[0010] S22: Establishing a transmission rate expression between the VLC access point VLC-AP and the user;

[0011] S23: establishing a transmission rate expression between the mmWave access point mmWave-AP and the user;

[0012] S24: Establishing an optimization objective function expression;

[0013] S25: Establishing optimization problem constraint expressions;

[0014] S26: Under the constraints in S25, the entire optimization problem is established with the goal of maximizing the expression of the objective function in S24;

[0015] S3: Optimization problem transformation, using Lyapunov optimization method, transforming the long-term optimization problem into a series of single-slot deterministic optimization problems;

[0016] S4: Use the penalty dual decomposition algorithm to solve the single-time slot deterministic optimization problem and achieve optimal resource allocation for each time slot.

[0017] Preferably, in step S1, the system model includes LED light emitting diodes and millimeter wave beam modules, RF-AP and user equipment UE, VLC-AP is equipped with LED lamp-based lighting equipment to meet lighting needs and communication services, millimeter wave beam modules are high-speed APs that cover a specific area, RF-APs provide low data rates, and APs are all connected to a system center controller. The controller collects user feedback, performs scheduling and association, and resource allocation tasks, and establishes a mathematical model based on the total transmission rate of the heterogeneous network under the actual constraints of AP power budget, backhaul link capacity, and system stability.

[0018] Preferably, in step S21, the research The working status of the system in a time slot is Represents a set of time slots, and the set of APs is represented by express, Indicates the number of RF-APs. Indicates the number of VLC-APs, represents the number of mmWave-APs, and the set of user equipment is composed of Indicates; let the position of all UEs be in time slot The inside remains unchanged, Select variables for the network, where Indicates UE In time slot Connect to RF-AP, otherwise ; Indicates UE In time slot Access VLC-AP; Indicates UE In time slot Accessing mmWave-AP; If the UE does not have multi-homing capability and the UE A given time slot can only be served by one AP, making the following constraints:

[0019] ;

[0020] The path loss of radio frequency communication in the RF channel model is expressed in dB as follows:

[0021] ;

[0022] in For RF-AP to The distance between users, is the carrier frequency in GHz, is the RF link loss model constant, represents the wall penetration loss in the non-line-of-sight NLoS case, set For thin walls or obstacles, where is the number of obstacles between the RF access point and the user;

[0023] In RF networks, channel gain is expressed as: ;

[0024] RF uses orthogonal frequency division multiplexing (OFDM) communication. The radio resources of RF include resource blocks (RBs) and power. is the RBs set of RF-AP, the number is , Select variables for RBs, Indicates that the RF-AP is in the time slot For the k User UE Allocate i Resource Blocks RB To communicate, vice versa , indicating that RF-AP is in the time slot For UE Allocate RB No communication took place;

[0025] There is only one RF AP in the hybrid network, and there is no interference between the UE and the VLC channel. Receives data from RF-AP The downlink received signal-to-noise ratio (SNR) of RBs is:

[0026] ;

[0027] in For RF-AP to UE In RB The transmission power on It is a time slot RB of RF-AP and UE The channel gain between is the noise power of RF-AP;

[0028] According to Shannon's formula, in the time slot RB of RF-AP With UE The downlink transmission rate between is:

[0029] ;

[0030] In time slot RF-AP and UE The transmission rate between is:

[0031] ;

[0032] An RB is allocated to at most one UE in a time slot. , with the following restrictions:

[0033] ;

[0034] The transmit power of the RF-AP must meet the following requirements:

[0035] ;

[0036] in The maximum transmit power of the RF-AP.

[0037] Preferably, in step S22, the LoS channel gain of the VLC link is represented by a Lambertian transmission model, and the LoS channel gain of the VLC link is represented as:

[0038] ;

[0039] in, represents the effective detection area of ​​PD, and represents the distance between the LED and the user and the incident angle, is the radiation angle from the LED to the user, represents the Lambert index, is the half-power angle of the LED, and are the gain of the filter and the gain of the light concentrator at the PD, is the field of view FOV of the receiver;

[0040] In time slot In the VLC-AP and UE The SNR between them is:

[0041] ;

[0042] in is the photoelectric conversion coefficient received by the UE, Indicates the VLC-AP to the UE The transmission power, It is a time slot When VLC-AP and UE The channel gain between In each time slot, it is time-varying and independent and identically distributed. is the noise power in the VLC channel;

[0043] Time slot During this period, each VLC-AP and UE The information transmission rate between is expressed as:

[0044] ;

[0045] like is the maximum transmit power of the VLC-AP, subject to the following constraints:

[0046] .

[0047] Preferably, in step S23, the path loss effect in the millimeter wave channel of the millimeter wave beam module is processed by a statistical model based on actual measurements, and the model is summarized as follows:

[0048] ;

[0049] in represents a constant, determined by the least squares fit of the propagation model to the floating intercept and slope at measurement distances up to 200 m, It is UE Distance to mmWave-AP, is the lognormal shading The variance of

[0050] When the millimeter wave carrier frequency is 73GHz, for LoS propagation, ;

[0051] UE The signal-to-noise ratio function of the link between and mmWave-AP is:

[0052] ;

[0053] in is the random channel gain, Each time slot is time-varying and independent and identically distributed, is the channel bandwidth of mmWave, is the noise figure of mmWave, is the noise power spectral density, Use transceiver antenna length and propagation environment density It is calculated that Indicates from mmWave-AP to UE The transmission power;

[0054] exist Time slot mmWave-AP and UE The information transmission rate between is:

[0055] ;

[0056] The maximum instantaneous transmit power of the mmWave AP is subject to the following constraints:

[0057] ;

[0058] Preferably, in step S24, the optimization target is the transmission rate of the entire VLC / RF heterogeneous network system. for:

[0059] ;

[0060] In the formula, It is a time slot During this period, each VLC-AP and UE The information transmission rate between yes Time slot mmWave-AP and UE The information transmission rate between It is a time slot RF-AP and UE The transmission rate between.

[0061] Preferably, in step S25, the central controller maintains the status of the UE buffer queue, and the AP is the UE Maintain a separate buffer queue and divide the time slots In Defined as AP For UE The queue length maintained in the time slot During this period, a certain amount of data arrives and is stored in the buffer at the beginning of the time slot, defining UE In time slot The amount of data received is , the data volume is independent and identically distributed in different time slots, UE The average arrival rate , The distribution satisfies ,and is a bounded variable, The discrete-time queue dynamics of the arrival and departure processes are expressed as:

[0062] ;

[0063] A model for data queue stability is established. The system queue stability requirements must meet the following requirements:

[0064] .

[0065] Preferably, in step S25, the backhaul link establishes a connection between the access point and the core network to achieve access to all service or content providers, and performs peak capacity limitation to ensure that the data transmission rate of the access point does not exceed or fall below the capacity of the maximum backhaul link, and the constraints are as follows:

[0066] ;

[0067] in , and Represent the capacities of RF, VLC and mmWave maximum backhaul links respectively.

[0068] Preferably, in step S26, in the time slot , the transmission rate of the entire VLC / RF heterogeneous network system is:

[0069] ;

[0070] The optimization goal is to maximize the time-averaged total rate while ensuring system stability. The above function is expressed as a time-averaged expression. The average transmission rate maximization problem is transformed into a stochastic optimization problem, and the following mathematical model is established:

[0071] ;

[0072] is the optimal variable set of P1, in which C1 ensures that , a UE can only be served by one AP; C2 ensures that each RB can be allocated to at most one UE in a time slot; C3-C5 set a total transmit power limit for each AP; C6 requires that the queue is limited in the average time; C7 limits the system power and limits the expected transmit power of each AP to ensure that the system does not overflow when working; C8-C10 represent the maximum available capacity of the backhaul link respectively; C11 ensures the non-negativity of the transmit power; C12 represents a binary variable.

[0073] Preferably, in step S3, the following steps are specifically included:

[0074] The Lyapunov optimization method is used to transform C7 into the evolution process of the virtual queue. The relationship between the virtual power queue and the time average power limit constraint is as follows:

[0075] ;

[0076] in , virtual queue Used to constrain queues, if there are virtual queues Stable strategy,automatically satisfies the average power constraint C7 and ensures the non-negativity of the queue;

[0077] and Respectively represent time slots The corresponding vector of queue states at , is the matrix of all virtual and real queues, and the quadratic Lyapunov function is defined as:

[0078] ;

[0079] Lyapunov function It is a scalar indicator that quantifies the congestion state of the queue. The Lyapunov function is in the time slot The single-slot drift at is given by:

[0080] ;

[0081] In the formula, B is a constant, ;

[0082] The Lyapunov drift of a time slot is defined as:

[0083] ;

[0084] Drift minus slot superior The expectation of , the drift-penalty function term is as follows:

[0085] ;

[0086] in is a non-negative constant parameter that controls the drift and rewards The trade-off between The larger the value, the higher the priority assigned to maximize the system transmission rate at the expense of queue length. The smaller it is, the higher the queue length is given priority;

[0087] Using the Lyapunov optimization method, the basic goal of P1 is to minimize the upper limit of the Lyapunov drift plus penalty for each time slot. The established Lyapunov drift-penalty function is calculated by the following formula:

[0088] ;

[0089] The Lyapunov optimization method minimizes the queue length by minimizing the drift and penalty terms, using Indicates the upper limit:

[0090] ;

[0091] According to Lyapunov optimization theory, the average value of P1 over a period of time is transformed into the problem of minimizing the upper bound of the Lyapunov drift plus penalty in each time slot. The simplified optimization problem P2 is:

[0092] .

[0093] Preferably, after Lyapunov optimization transformation, there are discrete variables in the problem , introduce auxiliary variables , transform constraint C12 into the following equality constraint:

[0094] ;

[0095] Then problem P2 can be equivalently transformed into P3:

[0096] ;

[0097] In step S4, the transformed C12 equality constraint is used as the dual of the AL term and penalized into the objective function. The problem is transformed into an augmented Lagrangian problem through the penalty dual decomposition method and expressed as:

[0098] ;

[0099] ;

[0100] Use AL multiplier and penalty factor To enhance the objective function, the PDD-based algorithm has a double-loop structure. In the inner loop, a fixed AL multiplier and penalty factor are used. The AL problem is solved by the BCD method; in the outer loop, the equality constraint is violated and updated. and ; If the constraint violations are below a certain threshold, update the dual variable , accelerates convergence; otherwise, the penalty coefficient will be updated to reduce violations; when When , problem P4 is equivalent to problem P3;

[0101] Use the BCD method to solve problem P4 and decompose it into four more manageable sub-problems: , , and , and solve them in turn in an alternating manner;

[0102] Optimize by fixing the remaining optimization variables , the optimization problem P5 is expressed as follows:

[0103] ;

[0104] P5 is a convex optimization problem with inequality constraints, which is solved with low complexity using the convex optimization toolbox CVX and CPLEX;

[0105] Optimize by fixing other variables , we use the convex programming toolbox CVX to solve problem P6, which can be expressed as:

[0106] ;

[0107] Fix the remaining optimization variables block and optimize , the CPLEX solver is used to solve the power allocation optimization problem. Problem P7 is stated as follows:

[0108] ;

[0109] By fixing the remaining variables and optimizing , the optimization problem is stated as follows:

[0110] ;

[0111] Its closed-form solution is:

[0112] ;

[0113] also, The closed-form solution of is as follows:

[0114] ;

[0115] ;

[0116] ;

[0117] In each iteration of the BCD method, the above update steps are performed sequentially according to the block structure of the optimization variables.

[0118] Therefore, the present invention adopts the above-mentioned dynamic resource optimization allocation method for VLC / mmWave / RF heterogeneous networks, and establishes a joint optimization problem of mixed integer nonlinear programming MINLP with multivariable coupling according to the mathematical model, so as to realize the optimization problem of maximizing the total transmission rate of the system under the actual constraints such as AP power budget, backhaul link capacity and system stability. A dual-loop algorithm based on PDD was developed on the basis of the AL method and the BCD algorithm. The dual-loop algorithm is a fully distributed algorithm that can iteratively converge to a stable point. While achieving queue stability, it is also proved that a trade-off between transmission rate and delay can be achieved.

[0119] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0120] Figure 1 It is a system model diagram of the indoor hybrid VLC / mmWave / RF heterogeneous network of the present invention;

[0121] Figure 2 It is a flow chart of a dynamic resource optimization allocation method for a VLC / mmWave / RF heterogeneous network of the present invention. DETAILED DESCRIPTION

[0122] The technical solution of the present invention is further described below through the accompanying drawings and embodiments.

[0123] Unless otherwise defined, technical or scientific terms used in the present invention shall have the common meanings understood by one having ordinary skills in the field to which the present invention belongs.

[0124] The words "include" or "comprises" and the like used in the present invention mean that the elements before the word include the elements listed after the word, and do not exclude the possibility of also including other elements. The orientation or position relationship indicated by the terms "inside", "outside", "upper", "lower", etc. is based on the orientation or position relationship shown in the drawings, which is only for the convenience of describing the present invention and simplifying the description, and does not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it cannot be understood as a limitation of the present invention. When the absolute position of the described object changes, the relative position relationship may also change accordingly. In the present invention, unless otherwise clearly specified and limited, the terms "attachment" and the like should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral body; it can be directly connected, or indirectly connected through an intermediate medium, and it can be the internal connection of two elements or the interaction relationship between two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0125] like Figure 2 As shown, the present invention provides a method for dynamic resource optimization allocation of a VLC / mmWave / RF heterogeneous network, comprising the following steps:

[0126] S1: Establish a mathematical model according to the scenario of a heterogeneous network of indoor hybrid visible light communication VLC, millimeter channel mmWave and radio frequency RF system; in step S1, if Figure 1 As shown in the figure, the system model includes LED light-emitting diodes and millimeter wave beam modules, RF-AP and user equipment UE. The VLC-AP is equipped with LED-based lighting equipment to meet lighting needs and communication services. The millimeter wave beam module is a high-speed AP that covers a specific area. The RF-AP provides a low data rate. The APs are all connected to the system center controller. The controller collects user feedback, performs scheduling and association, and allocates resources. A mathematical model is established based on the total transmission rate of the heterogeneous network under the actual constraints of AP power budget, backhaul link capacity and system stability.

[0127] S2: Establish a joint optimization problem according to the mathematical model established in step S1;

[0128] S21: Establishing the transmission rate expression between the RF access point RF-AP and the user; in step S21, studying The working status of the system in each time slot. T Indicates the number of time slots, using Represents a set of time slots, and the set of APs is represented by express, Indicates the number of RF-APs. Indicates the number of VLC-APs, represents the number of mmWave-APs, and the set of user equipment is composed of Indicates; let the position of all UEs be in time slot The inside remains unchanged, Select variables for the network, where Indicates UE In time slot Connect to RF-AP, otherwise ; Indicates UE In time slot Access VLC-AP; Indicates UE In time slot Accessing mmWave-AP; If the UE does not have multi-homing capability and the UE A given time slot can only be served by one AP, making the following constraints:

[0129] ;

[0130] The path loss of radio frequency communication in the RF channel model is expressed in dB as follows:

[0131] ;

[0132] in For RF-AP to The distance between users, is the carrier frequency in GHz, is the RF link loss model constant, represents the wall penetration loss in the non-line-of-sight NLoS case, set For thin walls or obstacles, where is the number of obstacles between the RF access point and the user;

[0133] In RF networks, channel gain is expressed as: ;

[0134] RF uses orthogonal frequency division multiplexing (OFDM) communication. The radio resources of RF include resource blocks (RBs) and power. is the RBs set of RF-AP, the number is , Select variables for RBs, Indicates that the RF-AP is in the time slot For the k User UE Allocate i Resource Blocks RB To communicate, vice versa , indicating that RF-AP is in the time slot For UE Allocate RB No communication took place;

[0135] There is only one RF AP in the hybrid network, and there is no interference between the UE and the VLC channel. Receives data from RF-AP The downlink received signal-to-noise ratio (SNR) of RBs is:

[0136] ;

[0137] in For RF-AP to UE In RB The transmission power on It is a time slot RB of RF-AP and UE The channel gain between is the noise power of RF-AP;

[0138] According to Shannon's formula, in the time slot RB of RF-AP With UE The downlink transmission rate between is:

[0139] ;

[0140] In time slot RF-AP and UE The transmission rate between is:

[0141] ;

[0142] An RB is allocated to at most one UE in a time slot. , with the following restrictions:

[0143] ;

[0144] The transmit power of the RF-AP must meet the following requirements:

[0145] ;

[0146] in The maximum transmit power of the RF-AP.

[0147] S22: Establish a transmission rate expression between the VLC access point VLC-AP and the user; In step S22, the LoS channel gain of the VLC link is expressed using the Lambertian transmission model, and the LoS channel gain of the VLC link is expressed as:

[0148] ;

[0149] in, represents the effective detection area of ​​PD, and represents the distance between the LED and the user and the incident angle, is the radiation angle from the LED to the user, represents the Lambert index, is the half-power angle of the LED, and are the gain of the filter and the gain of the light concentrator at the PD, is the field of view FOV of the receiver;

[0150] In time slot In the VLC-AP and UE The SNR between them is:

[0151] ;

[0152] in is the photoelectric conversion coefficient received by the UE, Indicates the VLC-AP to the UE The transmission power, It is a time slot When VLC-AP and UE The channel gain between In each time slot, it is time-varying and independent and identically distributed. is the noise power in the VLC channel;

[0153] Time slot During this period, each VLC-AP and UE The information transmission rate between is expressed as:

[0154] ;

[0155] like is the maximum transmit power of the VLC-AP, subject to the following constraints:

[0156] ;

[0157] S23: Establishing a transmission rate expression between the mmWave access point mmWave-AP and the user; In step S23, the path loss effect in the mmWave channel of the mmWave beam module is processed by a statistical model based on actual measurements, and the model is summarized as follows:

[0158] ;

[0159] in represents a constant, determined by the least squares fit of the propagation model to the floating intercept and slope at measurement distances up to 200 m, It is UE Distance to mmWave-AP, is the lognormal shading The variance of

[0160] When the millimeter wave carrier frequency is 73GHz, for line-of-sight LoS propagation, ;

[0161] UE The signal-to-noise ratio function of the link between and mmWave-AP is:

[0162] ;

[0163] in is the random channel gain, Each time slot is time-varying and independent and identically distributed, is the channel bandwidth of mmWave, is the noise figure of mmWave, is the noise power spectral density, Use transceiver antenna length and propagation environment density It is calculated that Indicates from mmWave-AP to UE The transmission power;

[0164] exist Time slot mmWave-AP and UE The information transmission rate between is:

[0165] ;

[0166] The maximum instantaneous transmit power of the mmWave AP is subject to the following constraints:

[0167] ;

[0168] S24: Establishing an optimization objective function expression; In step S24, the optimization objective is the transmission rate of the entire VLC / RF heterogeneous network system. for:

[0169] ;

[0170] In the formula, It is a time slot During this period, each VLC-AP and UE The information transmission rate between yes Time slot mmWave-AP and UE The information transmission rate between It is a time slot RF-AP and UE The transmission rate between.

[0171] S25: Establishing the constraint expression of the optimization problem; the central controller maintains the UE buffer queue, and the AP is the UE Maintain a separate buffer queue and divide the time slots In Defined as AP For UE The queue length maintained in the time slot During this period, a certain amount of data arrives and is stored in the buffer at the beginning of the time slot, defining UE In time slot The amount of data received is , the data volume is independent and identically distributed in different time slots, UE The average arrival rate , The distribution satisfies ,and is a bounded variable, The discrete-time queue dynamics of the arrival and departure processes are expressed as:

[0172] ;

[0173] A model for data queue stability is established. The system queue stability requirements must meet the following requirements:

[0174] ;

[0175] The backhaul link establishes a connection between the access point and the core network, enabling access to all service or content providers. Peak capacity limitation is implemented to ensure that the data transmission rate of the access point does not exceed or fall below the capacity of the maximum backhaul link. The constraints are as follows:

[0176] ;

[0177] in , and Represent the capacities of RF, VLC and mmWave maximum backhaul links respectively.

[0178] S26: Under the constraints in S25, the entire optimization problem is established with the goal of maximizing the expression of the objective function in S24; in step S26, in the time slot , the transmission rate of the entire VLC / RF heterogeneous network system is:

[0179] ;

[0180] The optimization goal is to maximize the time-averaged total rate while ensuring system stability. The above function is expressed as a time-averaged expression. The average transmission rate maximization problem is transformed into a stochastic optimization problem, and the following mathematical model is established:

[0181] ;

[0182] is the optimal variable set of P1, in which C1 ensures that , a UE can only be served by one AP; C2 ensures that each RB can be allocated to at most one UE in a time slot; C3-C5 set a total transmit power limit for each AP; C6 requires that the queue is limited in the average time; C7 limits the system power and limits the expected transmit power of each AP to ensure that the system does not overflow when working; C8-C10 represent the maximum available capacity of the backhaul link respectively; C11 ensures the non-negativity of the transmit power; C12 represents a binary variable.

[0183] S3: Optimization problem transformation: using the Lyapunov optimization method, the long-term optimization problem is transformed into a series of single-time-slot deterministic optimization problems; in step S3, the following steps are specifically included:

[0184] The Lyapunov optimization method is used to transform C7 into the evolution process of the virtual queue. The relationship between the virtual power queue and the time average power limit constraint is as follows:

[0185] ;

[0186] in , virtual queue Used to constrain queues, if there are virtual queues Stable strategy,automatically satisfies the average power constraint C7 and ensures the non-negativity of the queue;

[0187] and Respectively represent time slots The corresponding vector of queue states at , is the matrix of all virtual and real queues, and the quadratic Lyapunov function is defined as:

[0188] ;

[0189] Lyapunov function It is a scalar indicator that quantifies the congestion state of the queue. The Lyapunov function is in the time slot The single-slot drift at is given by:

[0190] ;

[0191] In the formula, B is a constant, ;

[0192] The Lyapunov drift of a time slot is defined as:

[0193] ;

[0194] Drift minus slot superior The expectation of , the drift-penalty function term is as follows:

[0195] ;

[0196] in is a non-negative constant parameter that controls the drift and rewards The trade-off between The larger the value, the higher the priority assigned to maximize the system transmission rate at the expense of queue length. The smaller it is, the higher the queue length is given priority;

[0197] Using the Lyapunov optimization method, the basic goal of P1 is to minimize the upper limit of the Lyapunov drift plus penalty for each time slot. The established Lyapunov drift-penalty function is calculated by the following formula:

[0198] ;

[0199] The Lyapunov optimization method minimizes the queue length by minimizing the drift and penalty terms, using Indicates the upper limit:

[0200] ;

[0201] According to Lyapunov optimization theory, the average value of P1 over a period of time is transformed into the problem of minimizing the upper bound of the Lyapunov drift plus penalty in each time slot. The simplified optimization problem P2 is:

[0202] .

[0203] S4: Use the penalty dual decomposition algorithm to solve the single-time slot deterministic optimization problem and achieve optimal resource allocation for each time slot. After the Lyapunov optimization transformation, there are discrete variables in the problem. , introduce auxiliary variables , transform constraint C12 into the following equality constraint:

[0204] ;

[0205] Then problem P2 can be equivalently transformed into P3:

[0206] ;

[0207] In step S4, the transformed C12 equality constraint is used as the dual of the AL term and penalized into the objective function. The problem is transformed into an augmented Lagrangian problem through the penalty dual decomposition method and expressed as:

[0208] ;

[0209] Use AL multiplier and penalty factor To enhance the objective function, the PDD-based algorithm has a double-loop structure. In the inner loop, a fixed AL multiplier and penalty factor are used. The AL problem is solved by the BCD method; in the outer loop, the equality constraint is violated and updated. and ; If the constraint violations are below a certain threshold, update the dual variable , accelerates convergence; otherwise, the penalty coefficient will be updated to reduce violations; when When , problem P4 is equivalent to problem P3;

[0210] Use the BCD method to solve problem P4 and decompose it into four more manageable sub-problems: , , and , and solve them in turn in an alternating manner;

[0211] Optimize by fixing the remaining optimization variables , the optimization problem P5 is expressed as follows:

[0212] ;

[0213] P5 is a convex optimization problem with inequality constraints, which is solved with low complexity using the convex optimization toolbox CVX and CPLEX;

[0214] Optimize by fixing other variables , we use the convex programming toolbox CVX to solve problem P6, which can be expressed as:

[0215] ;

[0216] Fix the remaining optimization variables block and optimize , the CPLEX solver is used to solve the power allocation optimization problem. Problem P7 is stated as follows:

[0217] ;

[0218] By fixing the remaining variables and optimizing , the optimization problem is stated as follows:

[0219] ;

[0220] Its closed-form solution is:

[0221] ;

[0222] also, The closed-form solution of is as follows:

[0223] ;

[0224] ;

[0225] ;

[0226] In each iteration of the BCD method, the above update steps are performed sequentially according to the block structure of the optimization variables.

[0227] Introducing constraint violation indicators to measure the violation of the equality constraints and indicate the termination of the algorithm.

[0228] ;

[0229] After the inner loop of the proposed PDD-based algorithm, the penalty parameter is updated according to the constraint violation condition , AL multiplier variable , specifically, the AL multiplier is updated by the following expression:

[0230] ;

[0231] The superscript Represents the number of external iterations. By updating the multiplier and penalty parameters in the outer loop and using the continuous iteration of the BCD algorithm in the inner loop, it can be ensured that the result of the converted problem converges to the KKT point of the original problem.

[0232] Therefore, the present invention adopts the above-mentioned dynamic resource optimization allocation method for VLC / mmWave / RF heterogeneous networks, and establishes a joint optimization problem of mixed integer nonlinear programming MINLP with multivariable coupling according to the mathematical model, so as to realize the optimization problem of maximizing the total transmission rate of the system under the actual constraints such as AP power budget, backhaul link capacity and system stability. A dual-loop algorithm based on PDD was developed on the basis of the AL method and the BCD algorithm. The dual-loop algorithm is a fully distributed algorithm that can iteratively converge to a stable point. While achieving queue stability, the problem of limited transmission rate under the constraints of wireless spectrum resources is solved.

[0233] 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 they can still modify or replace the technical solution of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solution to deviate from the spirit and scope of the technical solution of the present invention.

Claims

1. A method for dynamic resource optimization allocation in a VLC / mmWave / RF heterogeneous network, characterized in that: The following steps are involved: S1: Establish a mathematical model based on the indoor hybrid VLC, mmWave and RF heterogeneous network scenario; S2: Establish a joint optimization problem according to the mathematical model established in step S1; S21: Establishing a transmission rate expression between the RF access point AP and the user; In step S21, the The working status of the system in a time slot is Represents a set of time slots, and the set of APs is represented by Indicates that the set of user devices is composed of Indicates; let the position of all UEs be in time slot The inside remains unchanged, Select variables for the network, where Indicates UE In time slot Connect to RF-AP, otherwise ; Indicates UE In time slot Access VLC-AP; Indicates UE In time slot Accessing mmWave-AP; if the UE does not have multi-homing capability and UE-k can only be served by one AP in a given time slot, the following constraints are formulated: The path loss of radio frequency communication in the RF channel model is expressed in dB as follows: in For RF-AP to The distance between users, is the carrier frequency in GHz, is the RF link loss model constant, represents the wall penetration loss in the NLoS case, setting For thin walls or obstacles, where is the number of obstacles between the RF access point and the user; In RF networks, channel gain is expressed as: ; RF uses OFDM communication mode. The radio resources of RF include resource blocks RBs and power. is the RBs set of RF-AP, the number is , Select variables for RBs, Indicates that the RF-AP is in the time slot For UE Allocate RB To communicate, vice versa , indicating that RF-AP is in the time slot For UE Allocate RB No communication took place; There is only one RF AP in the hybrid network, and there is no interference between the UE and the VLC channel. Receives data from RF-AP The downlink received signal-to-noise ratio (SNR) of RB is: in For RF-AP to UE In RB The transmission power on It is a time slot RB of RF-AP and UE The channel gain between is the noise power of RF-AP; According to Shannon's formula, in the time slot RB of RF-AP With UE The downlink transmission rate between is: In time slot RF-AP and UE The transmission rate between for: An RB is allocated to at most one UE in a time slot. , with the following restrictions: The transmit power of the RF-AP must meet the following requirements: in is the maximum transmit power of the RF-AP; S22: Establishing a transmission rate expression between the VLC-AP and the user; In step S22, the LoS channel gain of the VLC link is expressed using a Lambertian transmission model. The LoS channel gain of the VLC link is expressed as: in, represents the effective detection area of ​​PD, and represents the distance between the LED and the user and the incident angle, is the radiation angle from the LED to the user, represents the Lambert index, is the half-power angle of the LED, and are the gain of the filter and the gain of the light concentrator at the PD, is the field of view FOV of the receiver; In time slot In the VLC-AP and UE The SNR between them is: in is the photoelectric conversion coefficient received by the UE, Indicates the VLC-AP to the UE The transmission power, It is a time slot When VLC-AP and UE The channel gain between In each time slot, it is time-varying and independent and identically distributed. is the noise power in the VLC channel; Time slot During this period, each VLC-AP and UE The information transmission rate between It is expressed as: like is the maximum transmit power of the VLC-AP, subject to the following constraints: ; S23: Establishing a transmission rate expression between mmWave-AP and user; In step S23, the path loss effect in the millimeter wave channel of the millimeter wave beam module is processed by a statistical model based on actual measurements, and the model is summarized as: in represents a constant, determined by the least squares fit of the propagation model to the floating intercept and slope at measurement distances up to 200 m, It is UE The distance to the mmWave AP, is the lognormal shading The variance of When the millimeter wave carrier frequency is 73GHz, for LoS propagation, ; UE The signal-to-noise ratio function of the link between the mmWave AP for: in is the random channel gain, Each time slot is time-varying and independent and identically distributed, is the channel bandwidth of mmWave, is the noise figure of mmWave, is the noise power spectral density, Use transceiver antenna length and propagation environment density It is calculated that Indicates from mmWave AP to UE The transmission power; exist Time slot mmWave-AP and UE The information transmission rate between for: The maximum instantaneous transmit power of mmWave-AP is subject to the following constraints: ; S24: Establish optimization objective function expression; In step S24, the optimization target is the transmission rate of the entire VLC / RF heterogeneous network system. for: In the formula, It is a time slot During this period, each VLC-AP and UE The information transmission rate between yes Time slot mmWave-AP and UE The information transmission rate between It is a time slot RF-AP and UE The transmission rate between S25: Establishing optimization problem constraint expressions; In step S25, the central controller maintains the status of the UE buffer queue, and the AP is the UE Maintain a separate buffer queue and divide the time slots In Defined as AP For UE The queue length maintained in the time slot During this period, a certain amount of data arrives and is stored in the buffer at the beginning of the time slot, defining UE In time slot The amount of data received is , the data volume is independent and identically distributed in different time slots, UE The average arrival rate , The distribution satisfies ,and is a bounded variable, The discrete-time queue dynamics of the arrival and departure processes are expressed as: A model for data queue stability is established. The system queue stability requirements must meet the following requirements: ； In addition, the backhaul link establishes a connection between the access point and the core network to enable access to all service or content providers, and implements peak capacity limitation to ensure that the data transmission rate of the access point does not exceed or fall below the capacity of the maximum backhaul link. The constraints are as follows: in , and They represent the capacity of RF, VLC and mmWave maximum backhaul links respectively; S26: Under the constraints in S25, the entire optimization problem is established with the goal of maximizing the expression of the objective function in S24; In step S26, the optimization goal is to maximize the time-averaged sum rate while ensuring system stability. The above function is expressed as a time-averaged expression. The average transmission rate maximization problem is transformed into a stochastic optimization problem, and the following mathematical model is established: is the optimal variable set of P1, in which C1 ensures that , a UE can only be served by one AP; C2 ensures that each RB can be allocated to at most one UE in a time slot; C3-C5 set a total transmit power limit for each AP; C6 requires that the queue is limited in the average time; C7 limits the system power and limits the expected transmit power of each AP to ensure that the system does not overflow when working; C8-C10 represent the maximum available capacity of the backhaul link respectively; C11 ensures the non-negativity of the transmit power; C12 represents a binary variable; S3: Optimization problem transformation: using the Lyapunov optimization method, the long-term optimization problem is transformed into a series of single-time-slot deterministic optimization problems; in step S3, the following steps are specifically included: The Lyapunov optimization method is used to transform C7 into the evolution process of the virtual queue. The relationship between the virtual power queue and the time average power limit constraint is as follows: in , virtual queue Used to constrain queues, if there are virtual queues Stable strategy,automatically satisfies the average power constraint C7 and ensures the non-negativity of the queue; and Respectively represent time slots The corresponding vector of queue states at , is the matrix of all virtual and real queues, and the quadratic Lyapunov function is defined as: Lyapunov function It is a scalar indicator that quantifies the congestion state of the queue. The Lyapunov function is in the time slot The single-slot drift at is given by: In the formula, B is a constant, ; The Lyapunov drift of a time slot is defined as: Drift minus slot superior The expectation of , the drift-penalty function term is as follows: in is a non-negative constant parameter that controls the drift and rewards The trade-off between The larger the value, the higher the priority assigned to maximize the system transmission rate at the expense of queue length. The smaller it is, the higher the queue length is given priority; Using the Lyapunov optimization method, the basic goal of P1 is to minimize the upper limit of the Lyapunov drift plus penalty for each time slot. The established Lyapunov drift-penalty function is calculated by the following formula: The Lyapunov optimization method minimizes the queue length by minimizing the drift and penalty terms, using Indicates the upper limit: According to Lyapunov optimization theory, the average value of P1 over a period of time is transformed into the problem of minimizing the upper bound of the Lyapunov drift plus penalty in each time slot. The simplified optimization problem P2 is: ； After Lyapunov optimization transformation, there are discrete variables in the problem , introduce auxiliary variables , transform constraint C12 into the following equality constraint: Then problem P2 can be equivalently transformed into P3: In step S4, the transformed C12 equality constraint is used as the dual of the AL term and penalized into the objective function. The problem is transformed into an augmented Lagrangian problem through the penalty dual decomposition method and expressed as: Use AL multiplier and penalty factor To enhance the objective function, the PDD-based algorithm has a double-loop structure. In the inner loop, a fixed AL multiplier and penalty factor are used. The AL problem is solved by the BCD method; in the outer loop, the equality constraint is violated and updated. and ; If the constraint violations are below a certain threshold, update the dual variable , accelerates convergence; otherwise, the penalty coefficient will be updated to reduce violations; when When , problem P4 is equivalent to problem P3; Use the BCD method to solve problem P4 and decompose it into four more manageable sub-problems: , , and , and solve them in turn in an alternating manner; Optimize by fixing the remaining optimization variables , the optimization problem P5 is expressed as follows: P5 is a convex optimization problem with inequality constraints, which is solved with low complexity using the convex optimization toolbox CVX and CPLEX; Optimize by fixing other variables , we use the convex programming toolbox CVX to solve problem P6, which can be expressed as: Fix the remaining optimization variables block and optimize , the CPLEX solver is used to solve the power allocation optimization problem. Problem P7 is stated as follows: By fixing the remaining variables and optimizing , the optimization problem is stated as follows: Its closed-form solution is: also, The closed-form solution of is as follows: In each iteration of the BCD method, the above update steps are performed sequentially according to the block structure of the optimization variables; S4: Use the penalty dual decomposition algorithm to solve the single-time slot deterministic optimization problem and achieve optimal resource allocation for each time slot.

2. According to claim 1, a method for dynamic resource optimization allocation of a VLC / mmWave / RF heterogeneous network is characterized in that: In step S1, the system model includes LED photodiodes and millimeter wave beam modules, RF-APs and user equipment. The VLC-AP is equipped with LED lamp-based lighting equipment to meet lighting needs and communication services. The millimeter wave beam module is a high-speed AP that covers a specific area. The RF-AP provides a low data rate. The APs are all connected to the system center controller. The controller collects user feedback, performs scheduling and association, and allocates resources. The mathematical model is established based on the total transmission rate of the heterogeneous network under the actual constraints of AP power budget, backhaul link capacity, and system stability.

Images