CN107484245B - The resource allocation methods of D2D communication are supported in a kind of heterogeneous network - Google Patents

Abstract

The present invention provides a resource allocation method supporting D2D communication in a heterogeneous network, comprising: a _D2D resource _allocator receiving a resource application request of a current D2D communication pair; The game method allocates resources for the current D2D communication pair, and constitutes the D2D resource allocation vector R _d ; the CUE resource allocator receives resource application requests from all CUEs within a preset time period, and according to the resource allocation vector R _d , the resource price vector F _m , use a centralized resource allocation method to allocate resources for each CUE, and combine the resources allocated by all CUEs to form a resource allocation vector R _n ; after the convergence judge determines that the QoE value of any CUE meets the usage requirements, it will send to The resources allocated by the CUE are output, and the resource allocation confirmation information is sent to the D2D resource allocator of the sending device. The above method can effectively adapt to the characteristics of resource management in the heterogeneous network, and effectively improve the QoE level of each user.

Description

A resource allocation method supporting D2D communication in a heterogeneous network

technical field

The present invention relates to communication technology, in particular to a resource allocation method supporting D2D communication in a heterogeneous network.

Background technique

In recent years, in order to meet the ever-increasing number of users and the ever-increasing demand for broadband services, cellular networks have begun to introduce terminal-to-terminal direct communication technology, which is different from traditional communication methods that need to be transferred through base station nodes, and the distance between nodes is relatively short. When , using an end-to-end (Device to Device, D2D) direct communication method can directly reuse the spectrum resources of the base station, thereby effectively improving the resource utilization rate and the user communication rate. At the same time, in terms of network structure, cellular networks are gradually evolving from traditional single-layer homogeneous macrocells to heterogeneous networks (Heterogeneous Cellular Networks, HCN). Compared with a single-cellular network, a heterogeneous network can improve spectrum utilization efficiency at a lower cost, and has become one of the most effective means to cope with huge business data traffic demands at present and in the future. However, when the end-to-end direct communication technology is used in the heterogeneous network, in addition to considering the base station association, spectrum allocation and power control issues of the traditional cellular user equipment (CUE), it is also necessary to allocate Reasonable communication resource blocks and communication power make it meet the transmission requirements of D2D without affecting the normal communication of all CUEs.

Another issue that needs attention is that with the popularization of Internet services, especially video services, network planners have gradually shifted from focusing only on network-side QoS guarantees to user-side QoE guarantees. Different from the evaluation of network performance by QoS, QoE directly reflects the user's evaluation of the service level running on the network, so it can more intuitively express the performance of the current network, and has become the focus of attention in the next generation network.

Contents of the invention

In order to solve the problems in the prior art, the present invention provides a resource allocation method supporting D2D communication in a heterogeneous network.

In a first aspect, the present invention provides a resource allocation method supporting D2D communication in a heterogeneous network, including:

Step S1, the D2D resource allocator receives the resource application request of the current D2D communication pair sent by the D2D resource requester;

The D2D resource allocator and the D2D resource applicant are located in the sending device of the current D2D communication pair;

Step S2, the D2D resource allocator uses a distributed game method to allocate resources for the current D2D communication pair according to the resource allocation vector R _n and the resource price vector F _m , and determines the D2D resource allocation vector R _d corresponding to all D2D communication pairs ;

The resource allocation vector R _n is a CUE-based resource allocation vector output by the CUE resource allocator in the base station received by the sending device in advance;

The resource price vector F _m is the resource price vector output by the resource price allocator in the base station received by the sending device in advance;

Step S3, the CUE resource allocator located in the base station receives the resource application requests of all CUEs within the preset time period sent by the CUE resource requester in the base station;

Step S4, the CUE resource allocator adopts a centralized resource allocation method to allocate resources for each CUE according to the resource allocation vector R _d and the resource price vector F _m , and forms the resources allocated by all CUEs into a resource allocation vector R _n ;

The resource allocation vector R _{d is the resource allocation vector R d determined} by the D2D resource allocator sent by the base station to receive the D2D communication from the sending device;

Step S5, the convergence judger located in the base station determines that the QoE value of any CUE after resource allocation meets the usage requirements, then outputs the resources allocated in step S4, and sends confirmation information of resource allocation to the D2D resource allocator of the sending device.

Optionally, the method also includes:

Step S6, the convergence determiner determines that the QoE value of any CUE does not meet the usage requirements after resource allocation, then the resource price allocator adopts the resource price update strategy to update the resource price vector, and outputs the updated resource price vector;

Based on the updated resource price vector, the process from step S2 to step S5 is repeated until it is determined in step S5 that the QoE value of any CUE after resource allocation meets the usage requirement.

Optionally, step S2 includes:

S21. The sending device to which the D2D resource allocation belongs receives the resource price vector Fm and the resource allocation vector _Rn sent by the base station, and sets the running times to 0;

S22. The D2D resource allocator randomly selects an allocation strategy to allocate resources for the D2D communication pair in the D2D resource applicant of the sending device;

S23. The D2D resource allocator of the sending device acquires resource allocation information of all D2D communication pairs;

S24. The D2D resource allocator forms a D2D resource allocation vector R _d ^new according to the resource allocation information of all D2D communication pairs, and obtains a benefit function U _d ^new of each D2D communication pair;

S25. The D2D resource allocator randomly selects one D2D in the D2D communication pair with a probability of 1/D, selects another allocation strategy to allocate resources for the selected D2D, generates a resource allocation vector R _d ^rand , and obtains a benefit function U _d ^rand ;

S26. Based on the SBR mechanism, the D2D resource allocator selects a corresponding allocation strategy among the resource allocation vector R _d ^rand and the resource allocation vector R _d ^new as the final resource allocation strategy;

S28. The transmission to which the D2D communication pair belongs is that the device broadcasts the current resource allocation strategy to all neighbor nodes;

S29, running times +1, and judge whether the maximum running times is reached, if yes, return to S30; otherwise, return to S22;

S30. The D2D resource allocator adopts the last selected resource allocation strategy to allocate resources for each D2D communication pair; and forms resource allocation vector R _d with resources allocated by all D2D communication pairs.

Optionally, Formula 1 is used in sub-step S25 and sub-step S24 to obtain respective benefit functions;

Formula one:

limitation factor,

Where η _d is the communication power allocated by the D2D communication pair d, is the maximum communication power that can be used by D2D communication pair d, L is the maximum power level after discretization, and the restriction is used to ensure that each D2D communication pair can only allocate one channel, Indicates whether D2D communication pair d uses channel m, otherwise it is 0, and f _m is the price of channel m;

MOS ₃ (r _d ) is the MOS value obtained for each D2D communication pair.

Optionally, substep S26 includes:

S261. Calculate the strategy judgment probability p by using Formula 2;

Where χ is the balance factor (χ>0), and For adopting the allocation strategy in sub-step S22 And the allocation strategy in sub-step S25 The benefit function obtained by calculating according to formula 1;

S261. If P<=rand(0,1), select the allocation strategy in sub-step S22 as the final resource allocation strategy;

Otherwise, select the allocation policy in sub-step S25 as the final resource allocation policy.

Optionally, for each CUE, the optimization goal is to obtain higher resource income while meeting the QoE usage requirements of each CUE;

Optimization model:

Restrictions:

Constraint one;

Constraint two;

Constraint three;

Constraint four;

Among them, constraint condition 1 is used to ensure that the obtained QoE value of each CUE meets the respective use requirements; constraint condition 2 is used to ensure that each CUE can be associated with at most one channel of one base station; is the associated variable of the nth CUE, if the nth CUE is associated with the channel m of the base station s, then otherwise Constraint three is used to ensure that each channel of each base station is allocated to only one CUE at most, where is the set of all CUEs of association mechanism s; l _s,m in constraint condition 4 is the power allocated by base station s on channel m.

Optionally, step S4 includes:

S41. Randomly initialize the population and the number of iterations of the population G, and set the initial operation algebra of the population g=1;

S42. Restoring the initialized individual so that it satisfies the constraints;

S43. Calculate the population according to the optimization model The fitness value of each individual in ;

S44. When g≤G, let and start from i=1 to R/2;

S45, according to the roulette method from the population Select two individuals A and B in , and the channel correlation vectors in individual A and individual B and Use the improved binary crossover method and get a new channel correlation vector and

S46. For the power distribution vectors in individual A and individual B and Use the traditional binary crossover method and get a new power allocation vector and

S47. Will and Combined to generate offspring individual A, the and Combine to generate offspring individual B;

Mutate the progeny individuals A and B with q _m probability;

Repair the mutated individual to make it meet the constraints;

Calculate the population according to the optimization model The fitness value of each individual in ;

S48. Comparing the fitness values of the individuals in the population R and R' in turn, replacing the low fitness value individuals in the population R with the corresponding high fitness value individuals in R'.

Optionally, before step S1,

The CUE resource allocator in the base station randomly selects channel resources that meet the constraint conditions for each CUE, and randomly allocates communication power; and the resource price allocator in the base station is based on the channel resource information used by the current CUE, the remaining channel resource information and/or The channel resource information used by the D2D communication pair randomly initializes a price for each channel.

Optionally, the sending device of the D2D communication pair is a requester of the D2D communication pair;

The channel resources used by each D2D communication pair within the preset range can be known by all the D2D communication pairs within the range through a sharing mechanism.

In the second aspect, the present invention also provides a resource allocation system supporting D2D communication in a heterogeneous network, the system includes: a base station, sending devices of multiple D2D communication pairs using D2D communication within the coverage of the base station, multiple CUE;

Wherein, the sending device interacts with the base station, and the base station determines whether the QoE value of each CUE among the CUEs interacting with the base station meets the usage requirement according to the channel resource information occupied by the sending device for D2D communication;

If not, the base station adjusts the channel resource price, and the sending device adapts to adjust the channel resource allocation information of the D2D communication pair that is applying for channel resources, and the base station adapts to adjust the channel resource allocation information of the CUE that is applying for channel resources, until the communication with the base station The QoE value of each CUE among the interacted CUEs meets usage requirements.

The beneficial effects that the present invention has are as follows:

The method of the present invention is a semi-distributed resource allocation method that combines the genetic algorithm and the game model. The genetic algorithm is mainly based on a centralized method to allocate appropriate communication resources for each CUE, and the game model is based on a distributed method to allocate appropriate communication resources for each D2D communication. To allocate communication resources. The above semi-distributed resource management framework can effectively adapt to the characteristics of resource management in heterogeneous networks, and effectively improve the QoE level of each user.

Description of drawings

FIG. 1A is a schematic flow diagram of a method according to an embodiment of the present invention;

FIG. 1B is a schematic structural diagram of a resource allocation device of the present invention;

Fig. 2 is a schematic diagram of the individual encoding method of the present invention;

Figure 3(a) is a schematic diagram of the traditional two-point intersection method;

Fig. 3 (b) is the schematic diagram of the improved two-point intersection method of the present invention;

Fig. 4 is a schematic diagram of the change of D2D user QoE level with iteration algebra;

Figure 5 is a schematic diagram of price changes for each resource block of a CUE user;

Figure 6 is a schematic diagram of the performance comparison of various algorithms.

Detailed ways

In order to better explain the present invention and facilitate understanding, the present invention will be described in detail below through specific embodiments in conjunction with the accompanying drawings.

All technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the technical field of the invention. The terms used herein in the description of the present invention are for the purpose of describing specific embodiments only, and are not intended to limit the present invention. As used herein, the term "and/or" includes any and all combinations of one or more of the associated listed items.

As shown in FIG. 1A, FIG. 1A shows a schematic flowchart of a resource allocation method provided by an embodiment of the present invention, and the method includes the following steps:

Step S1, the D2D resource allocator receives resource application requests for all D2D communication pairs sent by the D2D resource requester.

In this embodiment, the D2D resource allocator and the D2D resource applicant are located in the sending device of the current D2D communication pair, that is, the sending device.

Step S2, the D2D resource allocator uses a distributed game method to allocate resources for the current D2D communication pair according to the resource allocation vector R _n and the resource price vector F _m , and determines the D2D resource allocation vector R _d corresponding to all D2D communication pairs.

In this step, the resource allocation vector R _n may be a CUE-based resource allocation vector output by the CUE resource allocator in the base station received by the sending device in advance;

The resource price vector F _m may be the resource price vector output by the resource price allocator in the base station received by the sending device in advance.

Step S3, the CUE resource allocator located in the base station receives resource application requests of all CUEs within a preset time period sent by the CUE resource requester in the base station.

Step S4, the CUE resource allocator adopts a centralized resource allocation method to allocate resources to each CUE according to the resource allocation vector R _d and the resource price vector F _m , and forms resource allocation vector R _n from resources allocated by all CUEs.

In this step, the resource allocation vector R _d may be the resource allocation vector R _d determined by the D2D resource allocator sent by the base station from the D2D communication to the sending device.

Step S5, the convergence judger located in the base station determines that the QoE value of any CUE after resource allocation meets the usage requirements, then outputs the resources allocated in step S4, and sends confirmation information of resource allocation to the D2D resource allocator of the sending device.

Correspondingly, the above method further includes the following step S6 not shown in the figure;

S6. The convergence determiner determines that the QoE value of any CUE after resource allocation does not meet the usage requirements, then the resource price allocator adopts a resource price update strategy to update the resource price vector, and outputs the updated resource price vector;

Based on the updated resource price vector, the process from step S2 to step S5 is repeated until it is determined in step S5 that the QoE value of any CUE after resource allocation meets the usage requirement.

In a specific application, before step S1 in the above method,

The CUE resource allocator in the base station can randomly select channel resources that meet the constraint conditions for each CUE, and randomly allocate communication power; and the resource price allocator in the base station is based on the channel resource information used by the current CUE, the remaining channel resource information and/or Or the channel resource information used by the D2D communication pair randomly initializes a price for each channel.

It should be noted that, in this embodiment, the sending device of the D2D communication pair may be the requester of the D2D communication pair.

The method in this embodiment is a semi-distributed resource allocation method that combines genetic algorithm and game model. The genetic algorithm is mainly based on a centralized method to allocate appropriate communication resources for each CUE, and the game model is based on a distributed method. Communication pairs allocate communication resources. The above semi-distributed resource management framework can effectively adapt to the characteristics of resource management in heterogeneous networks, and effectively improve the QoE level of each user.

Step S2 in the above method may also include the following sub-steps S21 to S27:

S21. The sending device to which the D2D resource allocation belongs receives the resource price vector Fm and the resource allocation vector _Rn sent by the base station, and sets the running times to 0;

S22. The D2D resource allocator randomly selects an allocation strategy to allocate resources for the D2D communication pair in the D2D resource applicant of the sending device;

S23. The D2D resource allocator of the sending device acquires resource allocation information of all D2D communication pairs.

In particular, all D2D communication pairs within the preset range can learn channel resources used by each D2D communication pair through a sharing mechanism.

S24. The D2D resource allocator forms a D2D resource allocation vector R _d ^new according to the resource allocation information of all D2D communication pairs, and obtains a benefit function U _d ^new of each D2D communication pair;

S25. The D2D resource allocator randomly selects one D2D in the D2D communication pair with a probability of 1/D, selects another allocation strategy to allocate resources for the selected D2D, generates a resource allocation vector R _d ^rand , and obtains a benefit function U _d ^rand .

For example, in this embodiment, the above-mentioned sub-step S25 and sub-step S24 respectively use formula 1 to obtain their respective benefit functions;

Formula one:

limitation factor,

Where η _d is the communication power allocated by the D2D communication pair d, is the maximum communication power that can be used by D2D communication pair d, L is the maximum power level after discretization, and the restriction is used to ensure that each D2D communication pair can only allocate one channel, Indicates whether D2D communication pair d uses channel m, otherwise it is 0, f _m is the price of channel m; MOS ₃ (r _d ) is the MOS value obtained by each D2D communication pair.

S26. Based on the SBR mechanism, the D2D resource allocator selects a corresponding allocation policy among the resource allocation vector R _d ^rand and the resource allocation vector R _d ^new as the final resource allocation policy.

For example, sub-step S26 of this embodiment may include:

Use Formula 2 to calculate the strategy judgment probability p;

Where χ is the balance factor (χ>0), and For adopting the allocation strategy in sub-step S22 And the allocation strategy in sub-step S25 The benefit function obtained by calculating according to formula 1;

If P<=rand(0,1), select the allocation policy in sub-step S22 as the final resource allocation policy; otherwise, select the allocation policy in sub-step S25 as the final resource allocation policy.

S28. The transmission to which the D2D communication pair belongs is that the device broadcasts the current resource allocation strategy to all neighbor nodes;

S29, running times +1, and judge whether the maximum running times is reached, if yes, return to S30; otherwise, return to S22;

S30. The D2D resource allocator adopts the last selected resource allocation strategy to allocate resources for each D2D communication pair; and forms resource allocation vector R _d with resources allocated by all D2D communication pairs.

In addition, in order to better understand the content of FIG. 1A above, detailed description will be made in conjunction with FIG. 1B below. Both CUE user and CUE in the following description refer to traditional cellular users.

1. Description of each part of the resource allocation device:

(1) CUE resource allocation initializer: Located in the base station, it is mainly used to randomly select channel resources that meet the constraint conditions for each CUE, and randomly allocate communication power to them; the same channel that needs to meet the constraint conditions of the same BS cannot be allocated to the same channel at the same time. Two different CUEs, and the same CUE cannot occupy two channels at the same time.

Specifically, the set of CUEs in the entire network is defined as N={1, 2, 3..., N}. Then the initial resource allocation of all CUEs is defined as in is the channel initialization vector of each CUE, and L _N ={l ₁ ,l ₂ ,...,l _N } is the power initialization vector of each CUE.

(2) Resource price initializer _RN : located in the base station, randomly initializes a price for each channel

Assuming that the set of available channels of all BSs in the entire heterogeneous network is the same, M, the initial resource price vector is F _M ={f ₁ ,f ₂ ,…,f _M }, where f _m is the unit power used by channel m price. Alternatively, the resource price vector is updated according to the resource price update strategy.

(3) D2D resource requester: located in the terminal/sending device, used to apply for resources for each D2D communication pair, where for any D2D communication pair d, the resource application content is <IDd, QoEd>, d∈D.

It should be noted that normally, as long as the distance between two terminals is very close, the D2D direct communication technology can be used, and the D2D communication pair in this embodiment is predetermined.

(4) CUE resource applicant: located in the base station, used to apply for resources for each CUE, wherein for any CUE user n, the resource application content is <IDn, QoEn>, n∈N.

When a CUE user needs to apply for resources, start using the CUE resource applicant.

(5) D2D resource allocator: located in the terminal/sending device, based on the collected resource price and the resource application of each D2D communication pair, it executes a distributed game method to allocate resources for each D2D communication pair. The resource allocation vectors of all D2D communication pairs are expressed as in Spectrum vector assigned for each communication pair, A vector of communication power allocated for each communication pair.

(6) CUE resource allocator: located in the base station, based on the D2D resource allocation vector R _D , the resource requests of all CUEs, with the goal of maximizing the revenue from resource sales, implement a centralized resource allocation method, and allocate each CUE resources, and update all CUE resource allocation vectors Simultaneously output the final QoE levels aQoE _N of all CUEs.

The gray part in Figure 1B is the content of the resource application. For example, the resource application of the CUE only includes the ID number and QoE requirements of the CUE. After sending the above requirements to the resource allocation allocator, the resource allocator executes the resource allocation algorithm to apply allocate resources. QoE is a value set by the resource applicant after judging its own business.

(7) Convergence determiner: located in the base station, used to determine whether the entire resource allocation process has converged.

For example, the method for judging convergence is that resource prices remain unchanged after S consecutive rounds, that is, convergence is judged. S can be set by yourself.

(8) Resource price calculator: located in the base station, used to update the resource price vector F _M ={f ₁ , f ₂ ,...,f _M } according to the QoE value aQoE _N obtained by each CUE.

2. Resource allocation method

2.1 Semi-distributed resource allocation method:

The present invention also proposes a semi-distributed resource allocation method, the resource allocation method of which is shown in Algorithm 1.

Among them, line 7 to line 12 in Algorithm 1 act as resource price update method, and its main principle is as follows: For any CUE user n, if it allocates spectrum resources after t _n , the QoE value aQoE _n obtained by it cannot meet its QoE requirements, namely ( aQoE _n <τ _q ), then it can be concluded that the interference of D2D communication on the spectrum resource t _n is relatively large, because the price of the spectrum resource needs to be increased In order to limit the probability that the D2D communication pair in the network uses the frequency spectrum.

2.2 Distributed resource allocation method

For the resource allocation of D2D communication pairs, the present invention proposes a game-based distributed resource allocation algorithm, and the entire algorithm flow is shown in Algorithm 2 below.

The benefit function U _d of the D2D communication pair d is defined as the current QoE value obtained by the communication pair minus the cost of using the spectrum (the price of the strategy multiplied by the power), expressed as:

Where η _d is the communication power allocated by the D2D communication pair d, is the maximum communication power that can be used by D2D communication pair d, and L is the maximum power level after discretization. Restriction 1(a) is used to ensure that each D2D communication pair can only allocate one channel.

For D2D communication pair d, it adopts a random selection strategy The probability of being the current policy is

Where χ is the balance factor (χ>0), and to adopt the original strategy and new strategies The value of the benefit function calculated according to the formula (1). In this case, if Then it will be more likely to use the new strategy, and this SBR (Smoothed Better Response) mechanism based on probability selection can converge to Nash equilibrium with a higher probability.

2.3 Centralized resource allocation method

For CUE, its optimization goal is to obtain as high a resource benefit as possible while maintaining the QoE requirements of each CUE. Therefore, its optimization model can be defined as:

Among them, the optimization target of formula (3) is the overall revenue of resource sales, and the constraint condition (3a) is used to ensure that the obtained QoE value of each CUE user meets its demand value. Constraint condition (3b) is used to ensure that each CUE can be associated with at most one channel of one base station. in is the associated variable of CUE user n, if CUE user n is associated with channel m of base station s, then otherwise Constraint condition (3c) is used to ensure that each channel of each base station is allocated to only one CUE user at most, where is the set of all CUE users of association mechanism s. In the constraint condition (3d), l _s,m is the power allocated by the base station s on the channel m, where the present invention adopts a discrete power control mechanism, therefore, the allocated power is a power level. Specifically, when the allocated power level is l _s,m , its specific power is where L is the maximum power level, The maximum communication power allocated for base station s on each channel.

Aiming at the above optimization problem, the present invention proposes a centralized resource allocation method based on genetic algorithm, as shown in Algorithm 3.

Among them, for any individual A in the population, its coding scheme consists of two parts, which are the channel correlation vector and the power allocation vector Figure 2 shows the encoding scheme.

For the power allocation vector The traditional two-point crossover method is used. As shown in Figure 3(a). In addition, for the channel correlation vector We design an improved two-point intersection method, as shown in Fig. 3(b). Specifically, its crossover method is:

(1) Randomly select two points to intersect as c1 and c2;

(2) The genes between the two cross individuals A and B are retained to the next offspring individuals CA and CB;

(3) Insert the genes in individual B that are different from individual CA in turn into the remaining positions of individual CA, thereby generating offspring individual CA;

(4) Insert the genes in individual A that are different from individual CB into the remaining positions of individual CB in order to generate offspring individual CB;

In the above implementation method, the characteristics of CUE resource allocation and D2D resource allocation in heterogeneous networks are fully considered, and a semi-distributed resource allocation model and method are proposed, which can improve D2D as much as possible while maintaining the QoE requirements of CUE users. User's QoE level.

Referring to Fig. 4 and Fig. 5, Fig. 4 and Fig. 5 respectively show the change of QoE level of D2D users and the change of resource block price of CUE users. It can be seen that the algorithm proposed in the embodiment of the present invention can quickly converge to the optimal Excellent solution.

In addition, FIG. 6 shows the comparison of the final QoE obtained by the D2D user with the traditional MaxData and the Optimal algorithm based on exhaustive search. It can be seen that the algorithm proposed in the embodiment of the present invention is very close to the optimal solution.

Finally, it should be noted that: the above-described embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand : It is still possible to modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements to some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the technical solutions of the various embodiments of the present invention range.

Claims (9)

1. A resource allocation method supporting D2D communication in a heterogeneous network, characterized in that, comprising: Step S1, the D2D resource allocator receives the resource application request of the current D2D communication pair sent by the D2D resource requester; The D2D resource allocator and the D2D resource applicant are located in the sending device of the current D2D communication pair; Step S2, the D2D resource allocator uses a distributed game method to allocate resources for the current D2D communication pair according to the resource allocation vector R _n and the resource price vector F _m , and determines the D2D resource allocation vector R _d corresponding to all D2D communication pairs ; The resource allocation vector R _n is a CUE-based resource allocation vector output by the CUE resource allocator in the base station received by the sending device in advance; The resource price vector F _m is the resource price vector output by the resource price allocator in the base station received by the sending device in advance; Step S3, the CUE resource allocator located in the base station receives the resource application requests of all CUEs within the preset time period sent by the CUE resource requester in the base station; Step S4, the CUE resource allocator adopts a centralized resource allocation method to allocate resources for each CUE according to the resource allocation vector R _d and the resource price vector F _m , and forms the resources allocated by all CUEs into a resource allocation vector R _n ; The resource allocation vector R _{d is the resource allocation vector R d determined} by the D2D resource allocator sent by the base station to receive the D2D communication from the sending device; Step S5, the convergence judger located in the base station determines that the QoE value of any CUE after resource allocation meets the usage requirements, then outputs the resources allocated in step S4, and sends confirmation information of resource allocation to the D2D resource allocator of the sending device; If the convergence determiner determines that the QoE value of any CUE after resource allocation does not meet the usage requirements, the resource price allocator adopts the resource price update strategy to update the resource price vector, and outputs the updated resource price vector; Based on the updated resource price vector, the process from step S2 to step S5 is repeated until it is determined in step S5 that the QoE value of any CUE after resource allocation meets the usage requirement. 2. The method according to claim 1, wherein step S2 comprises: S21. The sending device to which the D2D resource allocation belongs receives the resource price vector Fm and the resource allocation vector _Rn sent by the base station, and sets the running times to 0; S22. The D2D resource allocator randomly selects an allocation strategy to allocate resources for the D2D communication pair in the D2D resource applicant of the sending device; S23. The D2D resource allocator of the sending device acquires resource allocation information of all D2D communication pairs; S24. The D2D resource allocator forms a D2D resource allocation vector R _d ^new according to the resource allocation information of all D2D communication pairs, and obtains a benefit function U _d ^new of each D2D communication pair; S25. The D2D resource allocator randomly selects one D2D in the D2D communication pair with a probability of 1/D, selects another allocation strategy to allocate resources for the selected D2D, generates a resource allocation vector R _d ^rand , and obtains a benefit function U _d ^rand ; S26. Based on the SBR mechanism, the D2D resource allocator selects a corresponding allocation strategy among the resource allocation vector R _d ^rand and the resource allocation vector R _d ^new as the final resource allocation strategy; S28. The transmission to which the D2D communication pair belongs is that the device broadcasts the current resource allocation strategy to all neighbor nodes; S29, running times +1, and judge whether the maximum running times is reached, if yes, return to S30; otherwise, return to S22; S30. The D2D resource allocator adopts the last selected resource allocation strategy to allocate resources for each D2D communication pair; and forms resource allocation vector R _d with resources allocated by all D2D communication pairs. 3. The method of claim 2, wherein, In sub-step S25 and sub-step S24, formula 1 is used respectively to obtain respective benefit functions; Formula one: limitation factor, Where η _d is the communication power allocated by the D2D communication pair d, is the maximum communication power that can be used by D2D communication pair d, L is the maximum power level after discretization, and the restriction is used to ensure that each D2D communication pair can only allocate one channel, Indicates whether D2D communication pair d uses channel m, otherwise it is 0, and f _m is the price of channel m; MOS ₃ (r _d ) is the MOS value obtained by each D2D communication pair; where M is the set of available channels; F _M = {f ₁ , f ₂ ,..., f _M }; where _RN is all CUE The resource allocation vector for . 4. The method according to claim 3, wherein the substep S26 includes: S261. Calculate the strategy judgment probability p by using Formula 2; Where χ is the balance factor (χ>0), and For adopting the allocation strategy in sub-step S22 And the allocation strategy in sub-step S25 The benefit function obtained by calculating according to formula 1; S261. If p<=rand(0,1), select the allocation strategy in sub-step S22 as the final resource allocation strategy; Otherwise, select the allocation policy in sub-step S25 as the final resource allocation policy. 5. The method according to claim 4, characterized in that, for each CUE, the optimization goal is to obtain higher resource income when the QoE usage requirement of each CUE is satisfied; Optimization model: Restrictions: Among them, constraint condition 1 is used to ensure that the obtained QoE value of each CUE meets the respective use requirements; constraint condition 2 is used to ensure that each CUE can be associated with at most one channel of one base station; is the associated variable of the nth CUE, if the nth CUE is associated with the channel m of the base station s, then otherwise Constraint three is used to ensure that each channel of each base station is allocated to only one CUE at most, where is the set of all CUEs associated with base station s; l _{s, m} in constraint condition 4 is the power allocated by base station s on channel m. 6. The method according to claim 5, wherein S4 comprises: S41. Randomly initialize the population and the number of iterations of the population G, and set the initial operation algebra of the population g=1; S42. Restoring the initialized individual so that it satisfies the constraints; S43. Calculate the population according to the optimization model The fitness value of each individual in ; S44. When g≤G, let and start from i=1 to R/2; S45, according to the roulette method from the population Select two individuals A and B in , and the channel correlation vectors in individual A and individual B and Use the improved binary crossover method and get a new channel correlation vector and S46. For the power distribution vectors in individual A and individual B and Use the traditional binary crossover method and get a new power allocation vector and S47. Will and Combined to generate offspring individual A, the and Combine to generate offspring individual B; Mutate the progeny individuals A and B with q _m probability; Repair the mutated individual to make it meet the constraints; Calculate the population according to the optimization model The fitness value of each individual in ; S48. Comparing the fitness values of the individuals in the population R and R' in turn, replacing the low fitness value individuals in the population R with the corresponding high fitness value individuals in R'. 7. The method according to claim 6, characterized in that, before step S1, The CUE resource allocator in the base station randomly selects channel resources that meet the constraint conditions for each CUE, and randomly allocates communication power; and the resource price allocator in the base station is based on the channel resource information used by the current CUE, the remaining channel resource information and/or The channel resource information used by the D2D communication pair randomly initializes a price for each channel. 8. The method according to claim 7, wherein the sending device of the D2D communication pair is the requester of the D2D communication pair; The channel resources used by each D2D communication pair within the preset range can be known by all the D2D communication pairs within the range through a sharing mechanism. 9. A resource allocation system supporting D2D communication in a heterogeneous network, characterized in that, The system includes: a base station, sending devices for multiple D2D communication pairs using D2D communication within the coverage of the base station, and multiple CUEs; Wherein, the sending device interacts with the base station, and the base station determines whether the QoE value of each CUE among the CUEs interacting with the base station meets the usage requirement according to the channel resource information occupied by the sending device for D2D communication; If not, the base station adjusts the channel resource price, and the sending device adapts to adjust the channel resource allocation information of the D2D communication pair that is applying for channel resources, and the base station adapts to adjust the channel resource allocation information of the CUE that is applying for channel resources, until the communication with the base station The QoE value of each CUE among the interacted CUEs meets usage requirements.