CN103973789A - VANET clustering method combining historical credit of vehicle with current state of vehicle - Google Patents

Abstract

The invention relates to the technical field of mobile communication and discloses a VANET clustering method combining the historical credit of a vehicle with the current state of the vehicle. According to the method, the historical credit R of the vehicle is modeled according to the transmission bandwidth, queue length and time for acting as the cluster head of the vehicle in a historical period; the current state function C of the vehicle is modeled according to the current state information, including the node degree and the speed and position of the vehicle relative to neighboring vehicles, of the vehicle; a utility function U=alpha R + (1-alpha)C is modeled according to the historical credit and the current state function of the vehicle, a candidate cluster head the corresponding utility function of which is the largest is selected to serve as a target cluster head, and then an adjacent user of the target cluster head is selected as a cluster member. By the adoption of the method, a VANET topological structure can be kept stable relatively, and service transmission performance can be optimized.

Description

A kind of VANET clustering method of combining the historical credit of vehicle and current state

Technical field

The present invention relates to the clustering method of vehicle self-organizing network, belong to vehicle-carrying communication technical field.

Background technology

In recent years, the intelligent transportation system based on vehicle self-organizing network VANET (Vehicular Ad-hoc NETwork) is subject to domestic and international extensive concern.VANET adopts advanced communication and network technology, can provide between vehicle and vehicle and roadside infrastructure between data communication.VANET combining global positioning system (GPS) and wireless communication technology, as WLAN (wireless local area network) (WLAN), cellular network etc., for the data access service of two-forty is provided in the vehicle of high-speed moving state, and information interaction between support unit, become support vehicles driving safety, the effective technology of high-speed data communication, intelligent traffic administration system and car entertainment is provided.

In VANET, the factors such as the fast moving characteristic of vehicle and access point AP (Access Point) limited coverage area cause part vehicle to carry out direct communication with AP, can be by adopting relay vehicle (RV) cooperation to forward the transfer of data realizing between source vehicle (SV) and AP.For application-specific scene, as dense distribution vehicle region, can be formed bunch by the nearer vehicle of geographic distance, by selecting bunch head at each bunch, support bunch interior nodes direct communication, bunch intermediate node is carried out relay forwarding by bunch head, can effectively reduce route control information expense, improve user data transmission efficiency, and realize Internet resources and efficiently utilize.But node high-speed motion causes link property dynamic change, each node available resources and business demand to exist the characteristics such as larger difference all VANET Clustering Algorithm to be proposed to new challenge.How to consider network in VANET, node and traffic performance, efficient clustering method reasonable in design, is problem demanding prompt solution.

The cluster scheme of the traffic characteristics design vehicle of existing research based on vehicle at present, document [SheaChristine, Behnam Hassanabadi and Shahrokh Valaee, Mobility-based clustering inVANETs using affinity propagation, IEEE GLOBECOM, 2009] a kind of method of carrying out cluster based on vehicle mobile has been proposed.At a bunch formation stages, by utilizing distributed affine algorithm, minimize the mobility between bunch head and bunch member, realize efficient cluster.Document [Cui Gang etc., the VANET cluster-dividing method of consideration vehicular traffic characteristic, publication number 102307373A, open day on January 4th, 2012] has proposed a kind of VANET cluster-dividing method based on vehicular traffic characteristic.By each vehicle node position, speed, connectedness and the behavior of travelling are carried out to simple weighted, select optimal objective bunch head, realize the optimization cluster of VANET.Document [Liu Zhengyu etc., a kind of VANET sub-clustering maintaining method based on adjacent vehicle distances, publication number 102883263A, open day on January 16th, 2013] a kind of method of carrying out VANET sub-clustering maintenance based on adjacent vehicle distances proposed, with the direct safe distance of the vehicle that travels on highway as with reference to foundation, by safe distance interdependence between calculating assessment car and car, and then the vehicle with certain safe correlation is divided into same cluster, determine respective cluster head and bunch member.

Above clustering method is only considered the characteristic such as car speed, position, does not consider the factor such as candidate cluster head Resource Availability and business efficiency of transmission, may cause that network topology stability is lower and data transmission performance is seriously limited,

Summary of the invention

The above-mentioned defect and the deficiency that exist for prior art VANET clustering method, the present invention considers the current state characteristic such as network and each vehicle relative velocity, distance and degree of communication, and serve as the historical credit rating of the factor modeling candidate cluster head such as bunch time based on average available bandwidth, queue length and history, by optimizing candidate cluster head current state and historical credit weighted sum, determine best cluster scheme, to realize the relatively stable and CAR SERVICE transmission performance optimization of network state.

The technical scheme that the present invention solves the problems of the technologies described above employing is.

Consider the historical credit rating of network and each vehicle current state and candidate cluster head, optimize and select bunch head and respective cluster member.A kind of VANET clustering method of combining the historical credit of vehicle and current state is proposed.Concrete technical scheme comprises: self current state information is sent to roadside unit RSU by vehicle, and RSU stores each car status information in its bunch of table, and the cluster flag CF of the each vehicle of juxtaposition is 0, a note bunch mark m=1; RSU is according to historical credit information R and the current state information C of vehicle, call the utility function value that vehicle cluster utility function U=α R+ (1-α) C calculates this vehicle, selecting the vehicle of corresponding utility function value maximum is bunch head, putting a bunch CF for a vehicle is 1, bunch member identifies ID_CH for (m, 0); If this bunch vehicle neighbor node number N _mbe less than or equal to the greatest member who allows in each bunch and count N _max, all neighbor nodes of this bunch of head are its bunch of member, and putting a bunch CF for member's vehicle is 1, and bunch member's mark of each neighbor node is followed successively by (m, i), i=1,2 ... N _m; If N _m>N _max, select the less N of utility function in neighbor node _maxindividual node is its bunch of member, and putting a bunch CF for member's vehicle is 1, and bunch member's mark of each neighbor node is followed successively by (m, i), i=1, and 2 ... N _max; RSU further checks the CF in bunch table, and the cluster utility function of the vehicle that calculating CF is 0, makes m=m+1, repeats the above-mentioned process of determining bunch head and bunch member thereof, until the CF position of all vehicles is 1; RSU sends bunch notification message to each bunch of head and bunch member, wherein, and 0≤α≤1.For i vehicle, according to the time of serving as bunch head in period of history T , determine a bunch factor the historical transmission bandwidth of normalization based on vehicle normalization queue length and a bunch factor, the historical credit function of modeling vehicle i is: wherein, w _b, w _lbe respectively the weights of corresponding transmission bandwidth and queue length, α _r, δ _rslope and the central value of corresponding historical credit function curve respectively.According to a hop neighbor interstitial content N of vehicle i _i, determine the vehicle i normalization node number of degrees $D_i^n=\left\{\begin{array}{cc}\frac{N_i}{N_{\max }}& N_i\≤N_{\max }\\ 1& N_i>N_{\max }\end{array}\right.,$ I=1,2, L, N, based on relative velocity and the relative position of the normalization node number of degrees, vehicle node and neighbor node, the current state function of modeling candidate vehicle i: i=1,2, L, N, wherein , be respectively relative velocity trust value and the relative position trust value of vehicle i, w _v, w _pbe respectively the weights of corresponding speed trust value and position trust value, α _t, δ _tslope and the central value of corresponding current state function curve respectively.According to the total quantity M of vehicle i transmitting-receiving bag in the T period _iand the big or small S giving out a contract for a project _b, call formula calculate the mean transmission bandwidth of vehicle i in period of history T, according to formula determine the historical transmission bandwidth of normalization of vehicle i, wherein, for the neighbours' vehicle within the scope of a jumping of vehicle i, for the maximum bandwidth of vehicle j.RSU receives when information of vehicles, and the each packet queuing model of modeling is M/M/1 queuing system, packet the time of advent interval and single window of being exponential distribution service time do not refuse system, the maximum queue length allowing according to vehicle i and average queue length calculating normalized average queue length is the relative velocity trust value of vehicle i is the speed difference DELTA v of all vehicles within the scope of itself and a jumping _ij, j ∈ Ω _i, j ≠ i, Δ v _ij=v _i-v _j, obey Δ v _i normal distribution, Δ v _ijprobability density function be: $f_{{\mathrm{\Δv}}_{\mathrm{ij}}}\left(x\right)=\frac{1}{\sqrt{2\mathrm{\π}}{\mathrm{\σ}}_{\mathrm{ij}}^{\mathrm{\Δv}}}{e}^{-\frac{{(x-m_{\mathrm{ij}}^{\mathrm{\Δv}})}^2}{2{\left({\mathrm{\σ}}_{\mathrm{ij}}^{\mathrm{\Δv}}\right)}^2}},$ Wherein $m_{\mathrm{ij}}^{\mathrm{\Δv}}=m_i^v-m_j^v,$ with be respectively vehicle i, the average speed of j, with be respectively vehicle i, the variance of j speed.Δ v _ijbe less than threshold speed v _thprobability

Vehicle i and one are jumped the distance d of all vehicles in scope _ij, obedience probability density is normal distribution, the position trust value of vehicle i is the distance d of itself and all hop neighbor vehicles _ij, j ∈ Ω _i, j ≠ i is all less than distance threshold d _thprobability ${T_i}^d=\underset{j\∈{\mathrm{\Ω}}_i,j\≠i}{\mathrm{\Π}}{\∫}_{-d_{\mathrm{th}}}^{d_{\mathrm{th}}}{f}_{d_{\mathrm{ij}}}\left(x\right)\mathrm{dx}.$

The present invention considers the cluster utility function value of the historical credit information of vehicle and current state modeling vehicle, selects vehicle that corresponding utility function value is larger as a bunch vehicle, and then determines bunch member's vehicle of each bunch.Adopt this method to carry out VANET cluster, can realize the relatively stable and CAR SERVICE transmission performance optimization of network topology structure.

Brief description of the drawings

Fig. 1 is the cluster scene of VANET of the present invention;

Fig. 2 is cluster flow chart of the present invention.

Embodiment

Clearer for the object, technical solutions and advantages of the present invention are expressed, below in conjunction with accompanying drawing, the present invention is described in further detail.

Fig. 1 is the VANET cluster scene that the present invention is based on historical credit and current state, and the deployment roadside unit (RSU) that each section is equal is responsible for all vehicles in this section to manage.

Fig. 2 is the flow chart of the VANET clustering method based on historical credit and current state of invention proposition, specifically comprises the following steps:

201: each vehicle sends oneself state information to RSU, and RSU is store car information in its bunch of table, calculates the utility function value of each vehicle, the cluster mark (CF) that each vehicle is set is 0, note initial cluster mark m=1.

Wherein, in bunch table (in table one), institute's canned data mainly comprises following information:

Vehicle ID: vehicles identifications, each vehicle has a unique ID.

Cluster mark (Cluster Flag, CF): characterize this node cluster whether, CF 1 represents cluster is 0 to represent not cluster.

Current state information: the present speed, position and the neighbor list thereof that mainly comprise this node vehicle.

Historical information: mainly comprise in the past in one-period time T historical transmission bandwidth, available bandwidth, the queue length of this vehicle and serve as time of bunch head.

Table one: bunch table canned data

202:RSU is according to historical transmission bandwidth, the queue length of vehicle and serve as time of bunch head, and the historical credit function of modeling candidate cluster head vehicle i is: wherein, for the historical bandwidth of normalization of vehicle i, that is: for the neighbours' vehicle within the scope of vehicle i mono-jumping, for the mean transmission bandwidth of vehicle i in period of history T, M _ifor the total quantity of the transceiving data bag of vehicle i within the T period, S _bfor the size of packet, for meeting j ≠ i's is all maximum.

for the historical queue length of normalization of vehicle i, and be respectively the average queue length of vehicle i and the maximum queue length allowing.According to the different business characteristic of vehicle node, its packet queuing model of modeling is M/M/1 queuing system, and the packet average arrival rate of supposing vehicle i is λ _i, average service rate is μ _i, its average queue length

for a bunch factor, wherein, for vehicle i serves as time of bunch head, w in period of history T _b, w _l, α _r, δ _rbe constant, wherein, w _b, w _lbe respectively the weights of corresponding historical transmission bandwidth and queue length, α _r, δ _rslope and the central value of corresponding historical credit function curve respectively.

203:RSU is according to the relative velocity of the vehicle node number of degrees, vehicle node and neighbor node and relative position, the current state function of modeling candidate cluster head vehicle i wherein for the normalization node number of degrees of vehicle i, that is: $D_i^n=\left\{\begin{array}{cc}\frac{N_i}{N_{\max }}& N_i\≤N_{\max }\\ 1& N_i>N_{\max }\end{array}\right.,$ Wherein, N _ifor a hop neighbor interstitial content of vehicle i.

for the relative velocity trust value of vehicle i.The each car speed of modeling is separate normally distributed random variable, and the speed of i vehicle is obeyed i=1,2, L, N, can obtain vehicle i, the relative speed Δ v of j _ij=v _i-v _j, obey ${\mathrm{\Δv}}_{\mathrm{ij}}~N(m_{\mathrm{ij}}^{\mathrm{\Δv}},{\left({\mathrm{\σ}}_{\mathrm{ij}}^{\mathrm{\Δv}}\right)}^2)$ Normal distribution, wherein $m_{\mathrm{ij}}^{\mathrm{\Δv}}=m_i^v-m_j^v,$ ${\left({\mathrm{\σ}}_{\mathrm{ij}}^{\mathrm{\Δv}}\right)}^2={\left({\mathrm{\σ}}_i^v\right)}^2+{\left({\mathrm{\σ}}_j^v\right)}^2,$ Δ v _ijprobability density function be: $f_{{\mathrm{\Δv}}_{\mathrm{ij}}}\left(x\right)=\frac{1}{\sqrt{2\mathrm{\π}}{\mathrm{\σ}}_{\mathrm{ij}}^{\mathrm{\Δv}}}{e}^{-\frac{{(x-m_{\mathrm{ij}}^{\mathrm{\Δv}})}^2}{2{\left({\mathrm{\σ}}_{\mathrm{ij}}^{\mathrm{\Δv}}\right)}^2}},$ The speed trust value of modeling vehicle i is the speed difference DELTA v of all vehicles within the scope of itself and a jumping _ij, j ∈ Ω _i, j ≠ i is all lower than given speed threshold value v _thprobability, ${T_i}^v=\underset{j\∈{\mathrm{\Ω}}_i,j\≠i}{\mathrm{\Π}}{\∫}_{-v_{\mathrm{th}}}^{v_{\mathrm{th}}}{f}_{{\mathrm{\Δv}}_{\mathrm{ij}}}\left(x\right)\mathrm{dx}.$

for the relative position trust value of vehicle i, modeling vehicle i and one are jumped the distance d of all vehicles in scope _ij, obedience probability density is normal distribution, the position trust value of vehicle i is the distance d of itself and all hop neighbor vehicles _ij, j ∈ Ω _i, j ≠ i is all less than distance threshold d _thprobability, ${T_i}^d=\underset{j\∈{\mathrm{\Ω}}_i,j\≠i}{\mathrm{\Π}}{\∫}_{-d_{\mathrm{th}}}^{d_{\mathrm{th}}}{f}_{d_{\mathrm{ij}}}\left(x\right)\mathrm{dx}.$

204: based on historical credit function and current state function, i vehicle cluster utility function of modeling is U _i=α R _i+ (1-α) C _i.

205:RSU select CF be 0 and the vehicle of effectiveness maximum as a bunch head, the CF of this vehicle is set to 1, bunch head/member mark is designated as (m, 0), the adjacent vehicle number of remembering this bunch of head is N _m.

206: judge this bunch vehicle neighbor node number and bunch in allow greatest member's number, i.e. N _m≤ N _max.

207: if so, this bunch all adjacent vehicle are its bunch of member, and a bunch head/member who identifies successively each vehicle is designated (m, i), i=1,2 ... N _i, putting each bunch of member CF is 1.

208: if not, select the less N of effectiveness in all adjacent vehicle _maxindividual vehicle, as its bunch of member, a bunch head/member who identifies successively each vehicle is designated (m, i), i=1,2 ... N _max, putting each bunch of member CF is 1.

209: judge the vehicle that whether has promising cluster of all vehicles, the CF of all vehicles is 1, if so, all clusters of all vehicles, finish algorithm.

210: if not, have the not vehicle of cluster, bunch mark m=m+1, goes to 205, until all clusters of all vehicles.

Claims (6)

1. combine the VANET clustering method of the historical credit of vehicle and current state for one kind, it is characterized in that: self current state information is sent to roadside unit RSU by vehicle, RSU stores each car status information in its bunch of table, and the cluster flag CF of the each vehicle of juxtaposition is 0, note bunch mark m=1; RSU is according to historical credit information R and the current state information C of vehicle, call the utility function value that vehicle cluster utility function U=α R+ (1-α) C calculates this vehicle, selecting the vehicle of corresponding utility function value maximum is bunch head, putting a bunch CF for a vehicle is 1, bunch member identifies ID_CH for (m, 0); If this bunch vehicle neighbor node number N _mbe less than or equal to the greatest member who allows in each bunch and count N _max, all neighbor nodes of this bunch of head are its bunch of member, and putting a bunch CF for member's vehicle is 1, and bunch member's mark of each neighbor node is followed successively by (m, i), i=1,2 ... N _m; If N _m>N _max, select the less N of utility function in neighbor node _maxindividual node is its bunch of member, and putting a bunch CF for member's vehicle is 1, and bunch member's mark of each neighbor node is followed successively by (m, i), i=1, and 2 ... N _max; RSU further checks the CF in bunch table, and the cluster utility function of the vehicle that calculating CF is 0, makes m=m+1, repeats the above-mentioned process of determining bunch head and bunch member thereof, until the CF position of all vehicles is 1; RSU sends bunch notification message to each bunch of head and bunch member, wherein, and 0≤α≤1.

2. VANET clustering method according to claim 1, is characterized in that: for i vehicle, according to the time of serving as bunch head in period of history T , determine a bunch factor the historical transmission bandwidth of normalization based on vehicle normalization queue length and a bunch factor, the historical credit function of modeling vehicle i is: wherein, w _b, w _lbe respectively the weights of corresponding transmission bandwidth and queue length, α _r, δ _rslope and the central value of corresponding historical credit function curve respectively.

3. VANET clustering method according to claim 1, is characterized in that, according to a hop neighbor interstitial content N of vehicle i _i, determine the vehicle i normalization node number of degrees $D_i^n=\left\{\begin{array}{cc}\frac{N_i}{N_{\max }}& N_i\≤N_{\max }\\ 1& N_i>N_{\max }\end{array}\right.,$ I=1,2, L, N, based on relative velocity and the relative position of the normalization node number of degrees, vehicle node and neighbor node, the current state function of modeling candidate vehicle i: i=1,2, L, N, wherein , be respectively relative velocity trust value and the relative position trust value of vehicle i, w _v, w _pbe respectively the weights of corresponding speed trust value and position trust value, α _t, δ _tslope and the central value of corresponding current state function curve respectively.

4. according to the VANET clustering method described in claim 2 or 3, it is characterized in that: according to the total quantity M of vehicle i transmitting-receiving bag in the T period _iand the big or small S giving out a contract for a project _b, call formula calculate the mean transmission bandwidth of vehicle i in period of history T, according to formula determine the historical transmission bandwidth of normalization of vehicle i, wherein, for the neighbours' vehicle within the scope of a jumping of vehicle i, for the maximum bandwidth of vehicle j.

5. according to the VANET clustering method described in claim 2 or 3, it is characterized in that: when RSU receives information of vehicles, the each packet queuing model of modeling is M/M/1 queuing system, be packet the time of advent interval and single window of being exponential distribution service time do not refuse system, the maximum queue length allowing according to vehicle i and average queue length calculating normalized average queue length is

6. VANET clustering method according to claim 3, is characterized in that: the relative velocity trust value of vehicle i is the speed difference DELTA v of all vehicles within the scope of itself and a jumping _ij, j ∈ Ω _i, j ≠ i, Δ v _ij=v _i-v _j, obey Δ v _i normal distribution, Δ v _ijprobability density function be:

$f_{{\mathrm{\Δv}}_{\mathrm{ij}}}\left(x\right)=\frac{1}{\sqrt{2\mathrm{\π}}{\mathrm{\σ}}_{\mathrm{ij}}^{\mathrm{\Δv}}}{e}^{-\frac{{(x-m_{\mathrm{ij}}^{\mathrm{\Δv}})}^2}{2{\left({\mathrm{\σ}}_{\mathrm{ij}}^{\mathrm{\Δv}}\right)}^2}},$ Wherein $m_{\mathrm{ij}}^{\mathrm{\Δv}}=m_i^v-m_j^v,$ ${\left({\mathrm{\σ}}_{\mathrm{ij}}^{\mathrm{\Δv}}\right)}^2={\left({\mathrm{\σ}}_i^v\right)}^2+{\left({\mathrm{\σ}}_j^v\right)}^2,$ with be respectively vehicle i, the average speed of j, with be respectively vehicle i, the variance of j speed, Δ v _ijbe less than threshold speed v _thprobability ${T_i}^v=\underset{j\∈{\mathrm{\Ω}}_i,j\≠i}{\mathrm{\Π}}{\∫}_{-v_{\mathrm{th}}}^{v_{\mathrm{th}}}{f}_{{\mathrm{\Δv}}_{\mathrm{ij}}}\left(x\right)\mathrm{dx}.$