CN106454920A - Resource allocation optimization algorithm based on time delay guarantee in LTE (Long Term Evolution) and D2D (Device-to-Device) hybrid network - Google Patents

Abstract

The invention discloses a resource allocation optimization algorithm based on time delay guarantee in an LTE (Long Term Evolution) and D2D hybrid network, and belongs to the technical field of wireless communication. The resource allocation optimization algorithm comprises the following steps: I, establishing a communication system for a D2D user and an LTE user existing as a pair in an LTE single cell; II, allocating resources to the D2D user and the LTE user in the communication system, and performing mathematical modeling under the condition that system throughput is maximized and user time delay is lower than threshold time delay; III, solving a mathematical model by adopting a resource allocation algorithm based on particle swarm optimization to obtain final positions of particles when the system throughput is maximum; and IV, performing simulation verification on the resource allocation algorithm based on the particle swarm optimization in order to lower system time delay effectively. The resource allocation optimization algorithm has the advantages that under an LTE network and a D2D communication hybrid network framework, on the premise that the user time delay does not exceed a time delay threshold, the overall throughput of the system is maximized, and reasonable allocation and optimization of wireless resources are realized.

Description

A kind of resource allocation optimization algorithm in LTE and D2D hybrid network based on Delay Guarantee

Technical field

The invention belongs to based on Delay Guarantee in wireless communication technology field, specifically a kind of LTE and D2D hybrid network Resource allocation optimization algorithm.

Background technology

At present, terminal direct connection（D2D）Communication technology allows two equipment close together, is not being affected by base station or is only receiving Direction communication is realized in the case of limited impact, the technology can significantly improve throughput of system and spectrum efficiency, and user's letter Make data that core net is required no, directly transmit between user, with good Research Prospects.Yet with the mobile industry in part Business is more sensitive to time delay, such as instant video communication and interactive game etc., and postponing a meeting or conference when long causes the using experience degree of user Decline, it is impossible to ensure the QoS of user.

Existing LTE cellular network has also been difficult to meet the growing mobility demand of user.

For the problems referred to above, the D2D communication technology based on low time delay can effectively meet demand of the user to real-time, Lifting communication QoS, the D2D for being ensured using low time delay under the scene of part are replaced LTE cellular communication, form LTE and D2D mixing group Net, can effectively improve spectrum efficiency, alleviate the pressure of mobile network.But, under LTE cellular network, deployment D2D network is also deposited In many challenges, such as D2D user is to discovery procedure, resource management between D2D user and LTE user etc..Mixed in LTE and D2D In charge-coupled net, there are problems of how Radio Resource distributes and to each other.

In existing LTE and D2D mixed networking mode, the hybrid network such as model selection, resource allocation is mainly adopted at present Radio resource management techniques.

Document [1]：The D2D multicast resource allocative decision with QoS guarantee based on OFDMA system.[C] // personal indoor Wireless communication system, the 24th international symposium of 2013IEEE, 2013:12383-2387. document [2]：Under LTE-A network Model selection // cordless communication network the meeting [C] of D2D communication, 2010IEEE.IEEE, 2010:1-6. and document [3]：In honeybee The wireless council of resource-sharing prioritization scheme [J] .IEEE for communicating for D2D under nest network, 2011,10 (8):2752- 2763. mode selection problem that have studied D2D user and phone user's share spectrum resources respectively, although mode selection algorithm is examined The link-quality of D2D and honeycomb and the disturbed condition under every kind of possible shared model are considered, have finally selected meeting Cellular Networks Highest transfer rate can be provided under network SINR restrictive condition.But, the algorithm of the pattern does not have on consideration multiplexing honeycomb Row or the difference of downlink resource, the also impact not by interference to overall performance of network is taken into account.

Document [4]：Under cellular network, [C] //IEEE vehicle technology conference .2009 is perceived based on the interference of D2D communication: 1-5. proposes a kind of Resource Allocation Formula for perceiving based on interference, and D2D carries out Interference Detection to up-link first, the Later Zhou Dynasty, one of the Five Dynasties Base station is fed back information to phase property, and base station is D2D user resource allocation, to reduce phone user with this according to the information for receiving Interference to D2D user.But, the Resource Allocation Formula could not realize the maximization of throughput of system, also fail to realize providing The reasonable fairness in source distribute to user.

Sequential and Resource Allocation Formula [C] // radio communication and network conference that document [5] is communicated based on D2D, 2013IEEE.IEEE,2013:134-139. proposes a kind of while considering that the resource of the fairness of throughput of system and user is divided Join algorithm, but, the resource allocation algorithm cannot perceive Link State, it is impossible to realize the requirement of low time delay.

D2D communication [J] under document [6] LTE-A network. modern science and communication technology, 2010,47 (12):42- The D2D Communication Jamming management method of shared model under LTE network is have studied in 49., however, the management method could not equally be distinguished Consider the uplink and downlink resource of cellular network.

Under document [7] 5G cellular network, D2D communication facing challenges and following developing direction [J] the .IEEE communication are miscellaneous Will, 2014,52 (5):Although the algorithm that 86-92. is proposed can reach highest speed rates, the algorithm does not account for power Control.

Document [8] is used for resource allocation performance evaluation [C] //GLOBECOM Workshops of D2D communication, 2011IEEE.IEEE,2011:358-362. proposes a kind of flexible resource multiplex for having combined model selection and power distribution Strategy, can minimize overall power consumption, but not make throughput of system maximum as far as possible.

To sum up, resource allocation research algorithm of the prior art does not all account for the impact that time delay communicates to D2D, is therefore It is that good Power Control and very high throughput of system is reached, but the QoS of user cannot be ensured, it is impossible to which satisfaction is When the business demand higher to delay requirement such as video.

Content of the invention

The present invention is have studied and is asked based on the resource allocation of time delay under the framework that LTE cellular network and D2D communication coexist Topic, by carrying out rational resource allocation come maximum system throughput to LTE user and D2D user；By will be based on time delay The resource allocation of guarantee is abstract for mixed integer nonlinear programming problem, in order to realize relatively low complexity, it is proposed that a kind of LTE With the resource allocation optimization algorithm in D2D hybrid network based on Delay Guarantee.

Comprise the following steps that：

Step one, be directed to the mono- cell of LTE, while exist match D2D user and LTE user, set up communication system.

Communication system includes：One base station, K₁Individual LTE user and K₂To D2D user couple；LTE user and D2D user couple Share N number of Resource Block.The downlink resource of communication multiplexing LTE network between D2D user couple, and a D2D user is to taking one Resource Block, different D2D users are to being multiplexed identical Resource Block.

K₁Individual LTE user is using setRepresent, K₂To D2D user to adopting Represent；

Step 2, resource allocation is carried out to the D2D user couple in communication system and LTE user, maximizing system throughput Amount and user's time delay carry out mathematical modeling under conditions of being less than thresholding time delay；

By the formula of resource allocation maximum system throughput, as follows：

Mathematical modeling is as follows：

The distribution condition of Resource Block n is indicated, whenWhen represent that Resource Block n is allocated to k-th user,When Then represent that Resource Block n is not allocated to k-th user.

K-th LTE user or k-th D2D user are represented to the transfer rate on Resource Block n；Formula is as follows：

B represents the bandwidth of Resource Block n；Γ (Γ≤1) represents the gap under truth with Shannon capacity；Represent kth The SINR of individual LTE user or k-th D2D user to receiving terminal on Resource Block n；Formula is as follows：

Represent that base station distributes to the transmission power of k-th LTE user on Resource Block n；Base station is on all Resource Block Transmission power sum no more than base station maximum transmission power limit

Represent the channel gain from base station to k-th LTE user on Resource Block n；

Represent and k-th LTE user being assigned to of Resource Block n or k-th D2D user is removed to rear, on Resource Block n The jamming power that other all users cause.

K-th LTE user or the noise power of k-th D2D user couple that expression Resource Block n is assigned to；

Represent k-th D2D user to the transmission power on Resource Block n；Any one D2D user is in all resources Transmission power sum on block is limited no more than the transmission power of D2D user

Represent the channel gain from the transmitting terminal of k-th D2D user couple to receiving terminal on Resource Block n；

l_kRepresent the queue length of k-th LTE user or k-th D2D user to transmitting terminal；

Represent the maximum delay thresholding that each user will meet.

The resource allocation algorithm of step 3, employing based on particle group optimizing obtains throughput of system to mathematics model solution Each particle final position when maximum；

According to the resource allocation problem based on time delay, discrete particle resource allocation is represented the company for being mapped to resource allocation On continuous domain, restrictive condition is converted into penalty function, proposes fitness function and solved with particle swarm optimization algorithm.

Comprise the following steps that：

Step 301, the position of each particle is initialized, speed, history optimal solution and globally optimal solution；

Location sets X of i-th particle on Resource Block_iIt is expressed as：I=1, 2 ..., S, S are the sum of particle；N=1,2 ..., 2N, N are the quantity of Resource Block.

For i-th particle, by the position of the particleAnd speedInitialized using the random number between 0 and 1, History optimal solution pbest_i0 is initialized to globally optimal solution gbest.

Step 302, be directed to i-th particle, update speed and the position of the particle in the t time iteration；

More new formula is as follows：

W is the inertia weight factor for controlling Particle velocity；c₁And c₂It is the normal amount of two Studying factors, r₁And r₂It is to take Value stochastic variable between zero and one.

pbest_iT () represents i-th particle in the t time iteration, make throughput of system maximization obtain the position of optimal value Put, i.e. the history optimal solution of the particle；

History optimal solution pbest of i-th particle_iT (), is represented with following formula：

Wherein, τ is iterationses, x_iT () is the position at i-th particle place；F () is object function.

Gbest (t) represents the t time iteration in the range of all particles, and making throughput of system obtain optimal value must own The position of particle, i.e. globally optimal solution, are represented with following formula：

Step 303, i-th particle is updated after position vector be divided into two parts, and will be whole per partly all be decoded into Corresponding Resource Block is respectively allocated to LTE user and D2D user couple as Resource Block sequence number by number；

By being decoded to the position vector that particle updates, obtain LTE user's sequence number corresponding to each Resource Block and The sequence number of D2D pair, and then judge that each Resource Block has been allocated to D2D user or LTE user.

For i-th particle, the location sets vector X on Resource Block_iIt is divided into two parts With

Decoding formula is as follows：

Represent the sequence number that Resource Block n is assigned to LTE user；Value represent Resource Block n and be assigned to The sequence number of D2D pair；Floor () is represented and is rounded downwards, K₁Represent quantity and the K of LTE user₂Represent the quantity of D2D user couple.

Value be K₁+K₂+ 1 expression Resource Block n is not allocated to D2D user couple.

Step 304, for D2D user and the LTE user of Resource Block is distributed, by base station and the transmitting terminal total work of D2D pair Rate is evenly distributed on each shared Resource Block, and resource allocation problem is converted into the maximization system under restrictive condition Throughput problem.

Restrictive condition is：

Step 305, fitness function is introduced to each particle, the maximization system under restrictive condition in step 304 is gulped down The amount of telling problem, is converted into non-limiting problem, and calculates the value of fitness function Fitness；

P∈R⁺It is compensating factor.For penalty function；

Step 306, judge the more new position of i-th particle whether can make fitness function value maximum, if it is, enter step Rapid 307, otherwise, return to step 302 updates i-th particle history optimal solution position pbest_i；

Step 307, the history optimal solution according to each particle, finding out makes all particle positions of fitness function value maximum Put, update globally optimal solution gbest.

Each particle final position when obtaining throughput of system maximum according to globally optimal solution, resource allocation reaches most Excellent.

Step 4, simulating, verifying is carried out to the resource allocation algorithm based on particle group optimizing, it is achieved that radio resource allocation Reasonably optimizing.

It is an advantage of the current invention that：

1）, a kind of resource allocation optimization algorithm in LTE and D2D hybrid network based on Delay Guarantee, in LTE network and Under D2D communication hybrid network framework, on the premise of ensureing user's time delay less than time delay thresholding, maximization system is integrally handled up Amount.

2）, a kind of resource allocation optimization algorithm in LTE and D2D hybrid network based on Delay Guarantee, in relatively low algorithm Good systematic function is obtained in the case of complexity, and calculated Resource Allocation Formula is existed with optimum Resource Allocation Formula Only has the gap of very little in performance.

3）, a kind of resource allocation optimization algorithm in LTE and D2D hybrid network based on Delay Guarantee, can effectively improve The QoS of D2D communication user in LTE network.

Description of the drawings

Fig. 1 is the model of communication system figure that the present invention sets up；

Fig. 2 is the resource allocation optimization algorithm flow process in a kind of present invention LTE and D2D hybrid network based on Delay Guarantee Figure；

Fig. 3 is method flow of the present invention using the resource allocation algorithm based on particle group optimizing to mathematics model solution Figure；

Fig. 4 is variation diagram of the throughput of system with D2D user to quantity under three kinds of algorithms of the present invention；

Fig. 5 is variation diagram of the average user time delay with D2D user to quantity under three kinds of algorithms of the present invention；

Fig. 6 be under two kinds of algorithms of the present invention average D2D speed with D2D user to the distance between variation diagram；

Fig. 7 be under two kinds of algorithms of the present invention average D2D time delay with D2D user to the distance between variation diagram.

Specific embodiment

Below in conjunction with the accompanying drawings the specific implementation method of the present invention is described in detail.

Resource allocation optimization algorithm in a kind of LTE and D2D hybrid network based on Delay Guarantee, is ensureing a fixed response time In the case of carry out the reasonable distribution of resource, using the resource allocation policy based on particle group optimizing, by the dynamic to Resource Block Scheduling and the simple distribution of power, reach the purpose of maximum system throughput；In being studied a question using time delay as with The restrictive condition that family need to meet.

As shown in Fig. 2 comprising the following steps that：

The mono- cell of LTE of step one, the D2D user for presence pairing simultaneously and LTE user, sets up communication system.

Overall network scene is as shown in figure 1, the communication system under LTE single-cell environment includes：One base station, K₁Individual LTE user and K₂To D2D user couple；D2D user has been completed pairing, LTE user and D2D user in a shared mode Share N number of Resource Block.

Only consider to D2D to and LTE user resource allocation process, between D2D user couple communication multiplexing LTE network under Row resource, and Resource Block can only be by a D2D to taking, and different D2D users are to being multiplexed identical Resource Block.So Would not there is interference between D2D user couple, only have base station in network to the interference of D2D user couple and D2D to being used phase With the interference that the LTE user of Resource Block brings.

K₁Individual LTE user is using setRepresent, K₂To D2D user to adopting Represent；Further, sequence number k=1 can be used, 2 ..., K unifying identifier LTE user and D2D user couple, wherein K=K₁+K₂.False If all downlink channel condition informations can be known in base station, such base station just can flexible Resources allocation between users.

Step 2, resource allocation is carried out to the D2D user couple in communication system and LTE user, maximizing system throughput Amount and user's time delay carry out mathematical modeling to resource allocation problem under conditions of being less than thresholding time delay；

On the premise of ensureing user time delay less than thresholding time delay, to D2D user to and LTE user carry out resource allocation Maximum system throughput, and mathematical modeling is carried out to resource allocation problem.

By the formula of resource allocation maximum system throughput, as follows：

Mathematical modeling is as follows：

A binary variable is represented, the distribution condition of Resource Block n is indicated, whenWhen represent Resource Block n be allocated K-th user is given,When then represent that Resource Block n is not allocated to k-th user, it is nonetheless possible to being allocated to other User.

Assume that the maximum transmission power in base station isThe transmission power of D2D user couple is limited

K-th LTE user or k-th D2D user are represented to the transfer rate on Resource Block n；Formula is as follows：

B represents the bandwidth of Resource Block n；Γ (Γ≤1) represents the gap under truth with Shannon capacity；

If D2D user is assigned to the Resource Block of LTE user to occupying, then D2D user is to will be by from base The interference that stands.LTE user will be subject to and it shares the D2D user of same resource block to disturbing.In general, channel capacity Can be calculated using shannon formula, but be to be beyond one's reach under reality, therefore true feelings are represented with Γ (Γ≤1) Gap under condition with Shannon capacity.

Represent the SINR of k-th LTE user or k-th D2D user to receiving terminal on Resource Block n；Formula is as follows：

Represent that base station distributes to the transmission power of k-th LTE user on Resource Block n；Base station is on all Resource Block Transmission power sum no more than base station maximum transmission power limit

Represent the channel gain from base station to k-th LTE user on Resource Block n；

Represent and k-th LTE user being assigned to of Resource Block n or k-th D2D user is removed to rear, on Resource Block n The jamming power that other all users cause.

K-th LTE user or the noise power of k-th D2D user couple that expression Resource Block n is assigned to；

Represent k-th D2D user to the transmission power on Resource Block n；Any one D2D user is in all resources Transmission power sum on block is limited no more than the transmission power of D2D user

Represent the channel gain of the transmitting terminal of the upper k-th D2D user couple of Resource Block n to receiving terminal；

l_kRepresent the queue length of k-th LTE user or k-th D2D user to transmitting terminal, and the arrival rate clothes for wrapping From Poisson distribution；

Represent the maximum delay thresholding that each user will meet.

Formula（3）Be object function, represent that asked a question target is maximum system throughput, will all users owning Transfer rate on Resource Block is added.Restrictive condition（4）Ensure that each Resource Block can only at most be assigned to a LTE and use Family；Restrictive condition（5）Ensure that each Resource Block can only at most be assigned to a D2D user couple.Formula（3-6）Represent resource Whether block n is allocated to k-th LTE user or k-th D2D user couple.Restrictive condition（7）Limit base station transmitting terminal Maximum transmission power；Restrictive condition（8）Limit maximum transmission power of the D2D user to transmitting terminal.Restrictive condition（9）Describe Each user will meet maximum delay thresholdingAbove optimization problem is that a mixed integer nonlinear programming is asked Topic, the problem has very big solution space.

Power distribution and resource allocation two parts are contained in constructed model, in order to divide power distribution and resource Join decoupling, it is assumed that the transmitting terminal of D2D user couple and base station distribute power averaging on the Resource Block for oneself being used, and this is one The individual simple and power distribution strategies with practicality.Afterwards without prejudice to restrictive condition（7）With（8）In the case of only consider Resource block assignments problem.

The resource allocation algorithm of step 3, employing based on particle group optimizing obtains throughput of system to mathematics model solution Each particle final position when maximum；

Particle swarm optimization algorithm is initially proposed by J.Kennedy and R.Eberhart, in standard particle colony optimization algorithm, The position of each particle represents a potential solution of optimization problem, and defines an object function to assess different solutions Fitness.The population of total S particle is moved in the solution space that M is tieed up, and finding to make object function obtain optimum Globally optimal solution.In each iteration, each particle adjusts the speed of oneself to follow its history optimal solution and current institute The globally optimal solution of discovery, finally enables particle reach globally optimal solution.

Standard particle colony optimization algorithm is generally used to solve the optimization problem of continuous solution space, but this is not particularly suited for this calculation Resource allocation problem in method based on time delay, the problem of representation of particle of the present invention only considers indicator variableThe variable-value is Discrete, it is possible to use discrete particle swarm optimization algorithm solving this problem, therefore according to the resource allocation based on time delay Problem, represents discrete particle resource allocation and is mapped on the continuous domain of resource allocation, restrictive condition is converted into compensation letter Number, proposes fitness function and is solved with particle swarm optimization algorithm.

As shown in figure 3, comprising the following steps that：

Step 301, the position of each particle is initialized, speed, history optimal solution and globally optimal solution；

The present invention represents the position of each particle using one by the elementary composition vector of 2N, wherein each element Value is between zero and one.The position of particle represents the resource block assignments situation of LTE user and D2D user couple.By LTE user and The resource block assignments of D2D user couple carry out combined optimization, independent dynamically distributes in the case of fixing with the distribution of LTE user resources block The Resource Block of D2D user couple is compared, and is obtained in that more preferable systematic entirety energy.

For N number of Resource Block and S particle, location sets X of i-th particle on Resource Block_iIt is expressed as：I=1,2 ..., S, S are the sum of particle；N=1, 2 ..., 2N, N are the quantity of Resource Block.

For i-th particle, by the position of the particleAnd speedInitialized using the random number between 0 and 1, History optimal solution pbest_i0 is initialized to globally optimal solution gbest.

Step 302, be directed to i-th particle, update speed and the position of the particle in the t time iteration；

More new formula is as follows：

W is the inertia weight factor for controlling Particle velocity；In initial standard particle colony optimization algorithm, inertia weight because Son is to maintain constant, but studies through further, it is thus proposed that the inertia weight factor can gradually subtract with the iteration of algorithm Little, affect algorithm performance to prevent particle to be absorbed in locally optimal solution in algorithm early stage, the inertia weight factor can be taken relatively Big value so that speed maintains original trend and huge change will not occur, in the algorithm later stage in order to make result more steady Rapidly restrain and greatly reduce the probability for occurring vibrating in convergence process, the inertia weight factor can be taken less Value is easier to change and can rapidly adapt to different situations so as to allow speed.Decreasing strategy with regard to the inertia weight factor also has People is studied, and the performance for summing up decreasing strategy by many experiments is the descending based on concave function from high to low successively Can, than good, the successively decreasing and being better than successively decreasing based on convex function based on linear function based on successively decreasing for linear function.Due to poor performance It is not very big, the present invention adopts linear function decreasing strategy.

c₁And c₂It is the normal amount of two Studying factors, c is generally set₁=c₂=2.r₁And r₂Be value between zero and one Stochastic variable.

pbest_iT () represents i-th particle in the t time iteration, make throughput of system maximization obtain the position of optimal value Put, i.e. the history optimal solution of the particle；Represented with following formula：

Wherein, τ is iterationses, x_iT () is the position at i-th particle place；F () is object function.

Gbest (t) is represented in the t time iteration in the range of all particles, obtains throughput of system maximization optimum It is worth the position of all particles, that is, globally optimal solution；

Step 303, i-th particle is updated after position vector be divided into two parts, and will be whole per partly all be decoded into Corresponding Resource Block is respectively allocated to LTE user and D2D user couple as Resource Block sequence number by number；

By being decoded to the position vector that particle updates, obtain LTE user's sequence number corresponding to each Resource Block and The sequence number of D2D pair, and then judge that each Resource Block has been allocated to D2D user or LTE user.

For i-th particle, the location sets vector X on Resource Block_iIt is divided into two parts WithThere are two elements in a vector, i.e.,WithCorresponding to Resource Block n.Obtain It is right respectively that the sequence number of the LTE user of Resource Block n and D2D user couple can pass throughWithDecode to obtain.

Decoding formula is as follows：

K₁Represent quantity and the K of LTE user₂Represent the quantity of D2D user couple；

Represent the sequence number that Resource Block n is assigned to LTE user；Span is from 0 to K₁；Round downwards, so taking Less than K₁+ 1, whenValue be to represent Resource Block n to be not allocated to LTE user for 0.

Value represent the sequence number that Resource Block n is assigned to D2D pair；Span is from K₁+ 1 arrives K₁+K₂+1.

Floor () is represented and is rounded downwards；K₁+K₂+ 1 be right side boundary extreme value take less than,Value be K₁+K₂+ 1 table Show that Resource Block n is not allocated to D2D user couple.

After in the method that have found expression particle and the position decoding of particle being become the result of resource block assignments, due to Power distribution problems employ base station and general power is evenly distributed to each shared Resource Block by the transmitting terminal of D2D user couple On solution, so except restrictive condition（9）Outside all of restrictive condition all meet.The money for being proposed before final Source assignment problem becomes in restrictive condition（9）Under maximum system throughput problem.

Step 304, for D2D user and the LTE user of Resource Block is distributed, by base station and the transmitting terminal total work of D2D pair Rate is evenly distributed on each shared Resource Block, and resource allocation problem is converted into the maximization system under restrictive condition Throughput problem.

Restrictive condition is：

Step 305, fitness function is built to each particle, the maximization system under restrictive condition in step 304 is gulped down The amount of telling problem, is converted into non-limiting problem, and calculates the value of fitness function Fitness；

Restrictive condition is eliminated by increasing introducing penalty function to object function, just calculate the value of last fitness function It is the solution of resource allocation problem.

Penalty function is expressed as：

Final fitness function is shown below：

P∈R⁺It is compensating factor；Penalty function plays very important in terms of guiding particle walks out non-feasible zone as early as possible Effect, for a feasible solution, the value of penalty function should be 0, then the value of fitness function is exactly resource allocation problem Solution.

Step 306, judge the more new position of i-th particle whether can make fitness function value maximum, if it is, enter step Rapid 307, otherwise, return to step 302 updates i-th particle history optimal solution position pbest_i；

Step 307, the history optimal solution according to each particle, finding out makes all particle positions of fitness function value maximum Put, update globally optimal solution gbest.

Globally optimal solution gbest, is represented by the following formula：

Each particle final position when obtaining throughput of system maximum according to globally optimal solution, resource allocation reaches most Excellent.

According to above step, based on particle group optimizing resource allocation algorithm overall flow using following natural language come Description：

Step 4, simulating, verifying is carried out to the resource allocation algorithm based on particle group optimizing, it is achieved that radio resource allocation Reasonably optimizing.

Emulated with ergodic algorithm by the resource allocation algorithm based on particle group optimizing, being randomly assigned algorithm, than Compared with performance, verify that resource allocation algorithm of the present invention can effectively reduce Time Delay of Systems further.

The simulating scenes of the present invention be set as a radius be 500 meters, system bandwidth be 3MHz LTE cell, have 15 Available resource block.Phone user and D2D user are to being uniformly distributed in whole cell range, and the quantity of wherein phone user is 3, The quantity of D2D user couple changes between 2-7 is individual according to different situations.

The path loss related from distance according to circumstances has two kinds of different computational methods, and the path between base station and user is damaged Consumption is using formula L (d)=128.1+37.6log₁₀D is calculating, and the path loss of D2D connection uses formula L (d)=148+40log₁₀d To calculate, it is to be measured in units of km wherein apart from d.The time delay thresholding of each user is set toBase station Transmission power maximum be 36dBm, user equipment maximum transmission power be 17dBm.While the power spectral density of noise sets For -174dBm/Hz.Crucial system emulation parameter is as shown in table 1.

Table 1

Parameter	Value
		Number of cells	1
Radius of society	500m
		Phone user's number	3
D2D is to number	2-7
		System bandwidth	3MHz
Base station maximum transmission power	36dBm
		User's maximum transmission power	17dBm
Noise power spectral density	-174dBm/Hz
		User's time delay thresholding	100ms

Parameter setting further with regards to particle swarm optimization algorithm is as follows, and iterationses are set to T=100, quantity S of particle =20, two Studying factors are set to c₁=c₂=2, and inertia weight factor w is from 0.95 to 0.4 linear minimizing.

Next resource allocation algorithm by the present invention based on particle group optimizing and traversal resource allocation algorithm, and at random Resource allocation algorithm is compared.Because the present invention is likely to be converging on local optimum based on the resource allocation algorithm of particle group optimizing Therefore it be compared, to obtain the performance between locally optimal solution and globally optimal solution by solution with traversal resource allocation algorithm Difference.Traversal resource allocation algorithm is calculated to each resource distribution mode, is selected one and is met time delay in all users The resource distribution mode of maximum system throughput in the case of restrictive condition.Random resource allocation algorithm be by a Resource Block A user is randomly assigned to, until all of Resource Block has all been allocated, this resource allocation algorithm does not consider any restriction Condition.

Be respectively compared throughput of system and the average user time delay of three kinds of algorithms, wherein the transmitting terminal of D2D user couple and The distance of receiving terminal is 50 meters, and the quantity of D2D user couple increases to 7 from 2.

As shown in Figure 4, it can be seen that the throughput of system ratio traversal resource based on the resource allocation algorithm of particle group optimizing The throughput of system of allocation algorithm is lower, because traversal resource allocation algorithm take into account every kind of resource allocation conditions, The solution for obtaining is globally optimal solution, and the resource allocation algorithm based on particle group optimizing is in the case that iterationses are restricted Obtained is locally optimal solution.The complexity ratio of resource allocation algorithm resource allocation algorithm based on particle group optimizing is traveled through in addition Complexity much higher, the carried algorithm of the present invention significantly reduces the complexity of algorithm in the case of sacrificial system handling capacity Degree.

Because the target of this algorithm is maximum system throughput, no matter the quantity of D2D user couple be how many, as far as possible The resource of system is taken full advantage of, as can be seen from the figure overall system throughput does not have very big with the increase of D2D pair Change, only has some increases.And the increase with D2D user to quantity, it can be seen that the resource based on particle group optimizing Between allocation algorithm and traversal resource allocation algorithm, the difference of handling capacity is gradually increased, because the resource allocation when number of users increases Situation becomes many, compares with situation of number of users when less, and the local for being solved based on the resource allocation algorithm of particle group optimizing is most Excellent solution and globally optimal solution gap can be increased.With regard to random resource allocation algorithm, as can be seen from the figure its throughput of system The all random change with user's average delay, and its performance is all poorer than other algorithms, because random resource allocation algorithm Simply each Resource Block is randomly assigned to a user, ensures the performance of system without any mechanism.

As shown in figure 5, during the average user of resource allocation algorithm based on particle group optimizing and traversal resource allocation algorithm Prolong very close to and postponing a meeting or conference during average user and with D2D user, the increase of quantity is increased.Because when D2D user is to quantity During increase, the resource assigned to by each user is reduced compared with before, can be obtained with transfer rate while reducing, when therefore Prolonging correspondingly to increase.In addition due to characteristic which is randomly assigned when using random resource allocation algorithm, it is possible to not It is that each user is assigned Resource Block, particularly when the quantity of D2D user couple increases.When a user is not divided When being fitted on Resource Block, the handling capacity of the user would is that 0, and time delay would is that infinity.When this happens, will not The data of this user are counted in the middle of the calculating of overall system performance, in case it is infinitely-great situation time delay occur.

Fig. 6 and Fig. 7 illustrate when D2D user to the distance between change when, D2D user is to Mean Speed and average delay Variation tendency, the quantity of wherein D2D user couple is set to 3.Random resource allocation algorithm is not because clearly become here Law and do not consider.

As shown in Figure 6, it can be seen that when the distance between the transmitting terminal of D2D user couple and receiving terminal increase, D2D is used The Mean Speed at family pair quickly reduces at the beginning, and the speed for reducing afterwards gradually becomes slow.When distance in short-term, D2D user To transmitting terminal and receiving terminal between channel conditions fine, resource allocation algorithm of the present invention will be mainly by will in this case Resource block assignments carry out maximum system throughput to the good user of channel conditions.And work as transmitting terminal and the receiving terminal of D2D user couple The distance between when being gradually increased, due to deteriorated channel conditions, the resource for distributing to D2D user couple can be reduced, so D2D user Mean Speed can be reduced.On the other hand, as each user has to meet the restriction thresholding of time delay, so distributing to The resource of D2D user must provide for enough transfer rates to ensure user's time delay not over threshold value, although after therefore Channel situation between D2D user couple continues to be deteriorated, and the trend that D2D user is reduced to Mean Speed can be than shallower.And work as D2D user to the distance between increase when, between resource allocation algorithm based on particle group optimizing and traversal resource allocation algorithm D2D user Mean Speed gap can be reduced because when D2D user to the distance between than larger when, in order to maximize be System handling capacity, the resource that can distribute to D2D user couple accordingly can be reduced, and only maintain enough transfer rates, and motility is due to money Source is reduced and is reduced, the gap between such globally optimal solution and locally optimal solution and D2D user to the distance between relatively hour Comparing will be less.

As shown in Figure 7, it can be seen that the average delay of D2D user couple increases with the increase of distance, based on population Between the resource allocation algorithm of optimization and traversal resource allocation algorithm, the gap of average delay gradually increases with the growth of distance Plus, this is because D2D user is being gradually reduced to the speed that can reach.

The present invention introduces the resource allocation policy based on particle group optimizing, the strategy in the hybrid network of LTE and D2D On the premise of throughput of system is ensured, resource allocation solution the resource allocation based on particle group optimizing is mapped to and has been calculated Method, the algorithm achieves the dynamic dispatching of resource and the simple distribution of power so that throughput of system is maximized, and effectively reduces logical The purpose of letter D2D time delay.

Claims (3)

1. the resource allocation optimization algorithm in a kind of LTE and D2D hybrid network based on Delay Guarantee, it is characterised in that include as Lower step：

Step one, be directed to the mono- cell of LTE, while exist match D2D user and LTE user, set up communication system；

Communication system includes：One base station, K₁Individual LTE user and K₂To D2D user couple；K₁Individual LTE user is using setRepresent, K₂To D2D user to adoptingRepresent；

Step 2, resource allocation is carried out to the D2D user couple in communication system and LTE user, in maximum system throughput and User's time delay carries out mathematical modeling under conditions of being less than thresholding time delay；

By the formula of resource allocation maximum system throughput, as follows：

$max\underset{k}{\mathrm{\Σ}}\underset{n}{\mathrm{\Σ}}{x}_k^n{r}_k^n$

Mathematical modeling is as follows：

$\left\{\begin{array}{c}\underset{k\∈C}{\mathrm{\Σ}}{x}_k^n\≤1\\ \underset{k\∈D}{\mathrm{\Σ}}{x}_k^n\≤1\\ x_k^n=\{0,1\}\\ \underset{n}{\mathrm{\Σ}}\underset{k\∈C}{\mathrm{\Σ}}{x}_k^n{P}_{BS}^n\≤P_{BS}^{\max }\\ \underset{n}{\mathrm{\Σ}}{x}_k^n{P}_k^n\≤P_d^{\max }\\ \frac{l_k}{\underset{n}{\mathrm{\Σ}}{x}_k^n{r}_k^n}\≤D_k^{threshold}\end{array}\right.$

The distribution condition of Resource Block n is indicated, whenWhen represent that Resource Block n is allocated to k-th user,Shi Zebiao Show that Resource Block n is not allocated to k-th user；

K-th LTE user or k-th D2D user are represented to the transfer rate on Resource Block n；

Represent that base station distributes to the transmission power of k-th LTE user on Resource Block n；Base station sending out on all Resource Block The transmission power for power sum being penetrated no more than base station maximum is limited

Represent k-th D2D user to the transmission power on Resource Block n；Any one D2D user is on all Resource Block Transmission power sum no more than D2D user transmission power limit

l_kRepresent the queue length of k-th LTE user or k-th D2D user to transmitting terminal；

Represent the maximum delay thresholding that each user will meet；

The resource allocation algorithm of step 3, employing based on particle group optimizing obtains throughput of system maximum to mathematics model solution When the final position of each particle；

Comprise the following steps that：

Step 301, the position of each particle is initialized, speed, history optimal solution and globally optimal solution；

Location sets X of i-th particle on Resource Block_iIt is expressed as：S Sum for particle；N=1,2 ..., 2N, N are the quantity of Resource Block；

For i-th particle, by the position of the particleAnd speedInitialized using the random number between 0 and 1, history Optimal solution pbest_i0 is initialized to globally optimal solution gbest；

Step 302, be directed to i-th particle, update speed and the position of the particle in the t time iteration；

More new formula is as follows：

$v_i^n(t+1)=w\×v_i^n\left(t\right)+c_1\×r_1\×\left({\mathrm{pbest}}_i\right(t)-x_i^n(t\left)\right)+c_2\×r_2\×\left({\mathrm{gbest}}_i\right(t)-x_i^n(t\left)\right)$

$x_i^n(t+1)=x_i^n\left(t\right)+v_i^n(t+1)$

W is the inertia weight factor for controlling Particle velocity；c₁And c₂It is the normal amount of two Studying factors, r₁And r₂It is value 0 And the stochastic variable between 1；

pbest_iT () represents i-th particle in the t time iteration, make throughput of system maximization obtain the position of optimal value, i.e., The history optimal solution of the particle；

History optimal solution pbest of i-th particle_iT (), is represented with following formula：

${\mathrm{pbest}}_i\left(t\right)=\underset{\mathrm{\τ}}{\min }f\left(x_i\right(t\left)\right),\mathrm{\τ}\≤t$

Wherein, τ is iterationses, x_iT () is the position at i-th particle place；F () is object function；

Gbest (t) represents the t time iteration in the range of all particles, makes throughput of system obtain optimal value and obtains all particles Position, i.e. globally optimal solution, represented with following formula：

$gbest\left(t\right)=\underset{i}{min}{\mathrm{pbest}}_i\left(t\right)=\underset{i,\mathrm{\τ}}{\min }f\left(x_i\right(t\left)\right),\mathrm{\τ}\≤t$

Step 303, i-th particle is updated after position vector be divided into two parts, and will all be decoded into integer per part and make For Resource Block sequence number, corresponding Resource Block is respectively allocated to LTE user and D2D user couple；

For i-th particle, the location sets vector X on Resource Block_iIt is divided into two partsWith

Decoding formula is as follows：

$D\left(x_i^n\right)=floor(x_i^n\×(K_1+1\left)\right),x_i^n\∈(0,1)$

$D\left(x_i^{N+n}\right)=floor(x_i^{N+n}\×(K_2+1\left)\right)+K_1+1,x_i^{N+n}\∈(0,1)$

Represent the sequence number that Resource Block n is assigned to LTE user；Value represent Resource Block n and be assigned to D2D pair Sequence number；Floor () is represented and is rounded downwards, K₁Represent quantity and the K of LTE user₂Represent the quantity of D2D user couple；

Value be K₁+K₂+ 1 expression Resource Block n is not allocated to D2D user couple；

Step 304, for D2D user and the LTE user of Resource Block is distributed, the transmitting terminal general power of base station and D2D pair is put down All it is assigned on each shared Resource Block, resource allocation problem is converted into the maximization system throughput under restrictive condition Amount problem；

Restrictive condition is：

Step 305, fitness function is introduced to each particle, the maximum system throughput problem under restrictive condition is converted For non-limiting problem, and calculate the value of fitness function Fitness；

$Fitness=\underset{k}{\mathrm{\Σ}}\underset{n}{\mathrm{\Σ}}{x}_k^n{r}_k^n-P\underset{k}{\mathrm{\Σ}}{\[min(0,D_k^{threshold}-\frac{l_k}{\underset{n}{\mathrm{\Σ}}{x}_k^n{r}_k^n})\]}^2$

P∈R⁺It is compensating factor；For penalty function；

Step 306, judge the more new position of i-th particle whether can make fitness function value maximum, if it is, enter step 307, otherwise, return to step 302 updates i-th particle history optimal solution position pbest_i；

Step 307, the history optimal solution according to each particle, finding out makes all particle positions of fitness function value maximum, more New globally optimal solution gbest；

Step 4, simulating, verifying is carried out to the resource allocation algorithm based on particle group optimizing, it is achieved that the conjunction of radio resource allocation Reason optimizes.

2. resource allocation optimization algorithm as claimed in claim 1 in a kind of LTE and D2D hybrid network based on Delay Guarantee, Characterized in that, LTE user and D2D user are to sharing N number of Resource Block in described step one；Communicate between D2D user couple multiple With the downlink resource of LTE network, and a D2D user, to taking a Resource Block, different D2D users are to being multiplexed identical Resource Block.

3. resource allocation optimization algorithm as claimed in claim 1 in a kind of LTE and D2D hybrid network based on Delay Guarantee, Characterized in that, described in step 2：

$r_k^n=B{\log }_2(1+{\mathrm{\ΓS}}_k^n)$

B represents the bandwidth of Resource Block n；Γ represents the gap under truth with Shannon capacity；Γ≤1；Represent k-th LTE The SINR of user or k-th D2D user to receiving terminal on Resource Block n；Formula is as follows：

$S_k^n=\left\{\begin{array}{cc}\frac{P_{BS}^n{g}_{BSk}^n}{I_k^n+{\mathrm{\σ}}_k^n}& k\∈C\\ \frac{P_k^n{g}_{kk}^n}{I_k^n+{\mathrm{\σ}}_k^n}& k\∈D\end{array}\right.$

Represent the channel gain from base station to k-th LTE user on Resource Block n；

Represent and remove k-th LTE user being assigned to of Resource Block n or k-th D2D user to rear, on Resource Block n other The jamming power that all users cause；

K-th LTE user or the noise power of k-th D2D user couple that expression Resource Block n is assigned to；

Represent the channel gain from the transmitting terminal of k-th D2D user couple to receiving terminal on Resource Block n.