CN106713477A - Internet of Vehicles network dynamic evolution method based on population competition - Google Patents
Internet of Vehicles network dynamic evolution method based on population competition Download PDFInfo
- Publication number
- CN106713477A CN106713477A CN201710001115.6A CN201710001115A CN106713477A CN 106713477 A CN106713477 A CN 106713477A CN 201710001115 A CN201710001115 A CN 201710001115A CN 106713477 A CN106713477 A CN 106713477A
- Authority
- CN
- China
- Prior art keywords
- network
- node
- topology
- car networking
- feature
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Withdrawn
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/12—Protocols specially adapted for proprietary or special-purpose networking environments, e.g. medical networks, sensor networks, networks in vehicles or remote metering networks
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L45/00—Routing or path finding of packets in data switching networks
- H04L45/02—Topology update or discovery
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention relates to the field of Internet of Vehicles, in particular to an Internet of Vehicles network dynamic evolution method based on population competition. The Internet of Vehicles network dynamic evolution method comprises the steps of: step 1, constructing Internet of Vehicles network node topological characteristics, wherein similarity of Internet of Vehicles nodes are excavated, a node topological characteristic is formed by a plurality of nodes with the similarity, and the real-time performance of a network system is improved in an Internet of Vehicles large-scale complex network real-time scene; step 2, establishing a stable and orderly network S measurement scale method; and step 3, dynamically evolving an Internet of Vehicles network. The Internet of Vehicles network dynamic evolution method deals with the diversity of application requirements and diversity of environment, and provides a guarantee for the steady state required for application of the Internet of Vehicles large-scale network.
Description
Technical field
The present invention relates to car networking field.
Background technology
In new car networking large scale network system, how on the basis of the real-time network that interconnects for building, protect
Stable state is held for application layer provides real time data, so that support applications layer intelligent Service, is that car networking large scale network is moved
The difficult point that state mechanism of Evolution research institute faces.
Car networking high dynamic characteristic causes that state-maintenance is very challenging.At present in vehicle self-organizing network VANET
In, researcher has carried out correlative study to route maintenance, achieves certain achievement, is mainly manifested in the three classes route of VANET
In agreement, including unicast, geographical multicast and broadcast.
Aim sequence distance vector route (Destination-Sequenced Distance-Vector Routing,
DSDV it is) the table- driven VANET routing plans based on bellman-ford algorithm, each entry in the routing table includes one
Individual sequence number, if there is link, the sequence number is even number, is otherwise odd number, solves the problems, such as route loop.DSR
Agreement (Dynamic Source Routing, DSR) is a kind of on-demand routing protocol, it is intended to needed for by eliminating Table driven mode
The periodicity table new information wanted is limiting the VANET control bandwidth that is consumed of message.In Ad-hoc demand distance vectors route
, it is necessary to the network node of connection is connected by broadcast transmission in (Ad-hoc On Demand Distance Vector, AODV)
Request, other nodes forward the message, and record they are which node to receive this message from, and back create interim road
By the node to initiation request.Take off will loyalty to wait in the VANET networks based on cooperative mechanism, it is considered to the communication that cooperative mechanism is produced
Gain calculates link transmission delay, and integrated link number of collisions and transmission delay devise MCCM (Multi- as route criterion
Channel Contention-aware Cooperative Metirc), substituted in AODV and judged as route using hop count
Foundation, devises a kind of distributed routing protocols MCCR (Multi-Channel Contention-aware
Cooperative Routing)。
Greedy edge stateless Routing Protocol (Greedy Perimeter Stateless Routing for
Wireless Networks, GPSR), data forwarding is carried out by the destination address of routing node position and packet.GPSR
Instant neighborhood information only according to node in network topology is forwarded, and greedy cannot be forwarded when packet enters one
Region when, the region that algorithm can be peripherally is route, but it is not particularly suited for urban road scene.Coordinate at greedy edge
Routing Protocol (Greedy Perimeter Coordinator Routing, GPCR) agreement is improved by GPSR agreements,
The main crossing part for considering street, the function restriction of direction of transfer is selected in the vehicle of street intersections by packet, rather than
The vehicle at crossing can only be forwarded according to the selected direction in last crossing with greedy pattern, it is adaptable to be route between city vehicle
Communication.Li Yuanzhen etc. proposes a kind of routing algorithm suitable for City scenarios VANET, is forwarded by the competition using timer
The suitable next-hop node of policy selection, the outstanding degree on the overtime duration and this node geo-location of timer is inversely proportional.
The selection of next fork in the road, the small traffic density of prioritizing selection physical length are carried out using dijkstra's algorithm in fork in the road region
There is packet higher to submit rate and relatively low in big street, its simulation comparison with GPSR the and PBRP agreements with caching
Data delay.SALEET H etc. propose the geographic routing protocol (Intersection-based based on intersection
Geographical Routing Protocol, IGRP), to solve the QoS routing issues of VANET in City scenarios, calculating
It is better than GPSR and GPCR in amount.
Geographical source-routed protocol (Geographic Source Routing) forwards packet, the road according to forward-path
Footpath is based on coordinate position and position on map calculates, but can not solve the too low caused partially connected of traffic density and ask
Topic.The geographical stateless VANET routes of the proposition such as XIANG Yong (Geographic Stateless VANET Routing,
GeoSVR), it is combined with numerical map and a kind of improved limitation forwarding algorithm by by node location, overcomes insecure nothing
Line channel problems.Song is superfine to propose a kind of DRES (Distributed Real-time delay Evaluation
Scheme) mechanism, the network delay information of real-time is obtained for vehicle, and is devised a kind of using the base for carrying forwarding mechanism
In the VANET Routing Protocols of distributed real time information.The agreement selects the routing mode in footpath using source, and vehicle is according to each section net
The estimation of network time delay, the most short forward-path of data transfer time delay is calculated using dijkstra's algorithm on the electronic map.
NZOUONTA J etc. propose based on road occupation vehicular traffic (Road-Based using Vehicular Traffic,
RBVT) Routing Protocol, gathers real-time vehicle transport information, creates and is made up of the linking-up road intersection of higher network connection probability
Path.
Vehicle auxiliary data transmission agreement (Vehicle-Assisted Data Delivery, VADD) employs carrying and turns
Hair mechanism, the delay of packet transmission is calculated using predictable vehicle mobility, finds the road of next forwarding packet
Road.Preferentially detect (Location First Probe, L-VADD), orientation preferentially and detect (Direction First in position
Probe, D-VADD) can be used to select optimal path with three kinds of different retransmission protocols such as hybrid VADD (H-VADD).
Geographical Multicast Routing (Geocast routing) is substantially a location-based QoS routing, and its target is
Packet is sent to the every other node in a geographic area specified, referred to as association area, many VANET from source node
Application program can benefit from it.BACHIR A etc. propose geographical multicast protocol (the Inter-Vehicles Geocast in workshop
Protocol, IVG), broadcast warning message is to the vehicle in all danger zones based on time delay algorithm on highway.
KIHL M etc. propose a geographical multicast protocol for Robust distributed, and for inter-vehicle communication, its target is to pass packet
The vehicle positioned at specific static state geographic area is sent to, vehicle is received or packet discard according only to its current position.
DURRESI M etc. propose a urgent broadcast agreement BROADCOMM on the highway of subregion based on Geographic routing,
Sensor on automobile constantly collects important information, and detecting any emergency will directly trigger broadcast, the association
Discuss in terms of message broadcast time delay with routing cost better than similar based on the Routing Protocol for flooding.During the multi-hop Radio Broadcasting Agreements of city
Prolong with routing cost aspect better than similar based on the Routing Protocol for flooding.City multi-hop Radio Broadcasting Agreements (Urban Multi-Hop
Broadcast protocol, UMB) be in order to solve the integrity problem of broadcast storm, many hop broadcastings of concealed nodes and urban district,
Sender's node selects broadcaster node farthest upwards to be turned in the case of without any priori topology information, as far as possible
Send out and reply data bag, it is new by the repeater initialization positioned at intersection when the path Shang You intersections that the message is propagated
Directional broadcasting.TONGUZ O etc. propose multi-hop broadcast agreement DV-CAST in VANET, while being applied to intensive and sparse
Traffic scene.
The research on VANET Routing Protocols is concentrated mainly on before existing research, from route-type, forwarding strategy, recovers plan
Slightly, the aspect such as route maintenance and applicable scene is specifically compared, as shown in table 1.From existing domestic and international present Research point
Analysis, effectively keeps the dynamic evolution method of steady ordered state essentially for small yardstick closed loop system in current car networking research
System, does not consider under extensive new car intranet network environment as the change of demand and environment keeps stable state to meet
The completion of current task.Therefore, in order to preferably support car networking application development, it is necessary to propose new suitable for following car networking
Dynamic evolution mechanism required for Intelligent Service.
The content of the invention
The present invention breaks through the Research Thinking of existing small yardstick closed-loop system, from car networking large-scale network environment, structure
Car networking network node topological characteristic is made, on this basis, by Lyapunov stability theory, network stabilization bar is provided
Part, with reference to Species Competition, proposes network stabilization state transition method.Above-mentioned problem in science is solved, setting up a set of can support that car joins
Net large scale network access theoretical system and method, for Internet provides theory support, for application layer provides real time data
Ensure, while also for future car the Internet services intelligent development establishes theoretical and application foundation.
Car networking is a net for complexity being made up of mutually coordinated between some network element/subsystems, infiltration and restriction
Network system, the faint change of network internal any one network element/subsystem can all cause the respective change of dynamic network, with former
What subsystem is always influenceed and regulation by external web environment again.The network system of external web environment and dynamic construction
Between system and property and the intensity of the effect that influences each other of network internal is again extremely complex.Species Competition is the ecosystem
In by continuous competitive evolution between population, promote inferior position population extinction, dominant population to retain.Found by analysis, car networking
In this complicated unordered chaos state to Competitive Thought between the dynamical evolution process of steady ordered state and population more coincide
The thought of present invention Species Competition from ecology, tries hard to description car networking system and changes Near The Critical Point in stable state
The condition of phase transformation and behavior, and degree of stability to system is the reason for be described and analyze its change and mechanism, from new angle
The car networking dynamical evolution problem of the complicated form of degree close examination, with the extensive open loop of thought guidance car networking of Species Competition, cross-domain
Dynamic evolution method design, so as to provide the new method of solve problem for car networking system, this joins to lifting China in car
There is important technical support meaning in the core competitiveness in net field.
Present invention aim at a kind of car networking network dynamic evolution method based on Species Competition is disclosed, from new angle
The car networking dynamical evolution problem of the complicated form of close examination, with the extensive open loop of thought guidance car networking of Species Competition, cross-domain
The design of dynamic evolution method, so as to provide the new method of solve problem for car networking system.
The present invention needs technical scheme to be protected, is characterized as below：
A kind of car networking network dynamic evolution method based on Species Competition, it is characterised in that specific method includes as follows
Step：
In step 1, car networking network node topological characteristic construction
Assuming that in Training scene k, from time Tx(x=1,2 ..., r) take q node in the node of collection, by pre- place
It is Q to manage and carry out the node new feature after deep learning1,Q2,…,Qi,…,Qq.First, the region structure for being covered in q node
M different node topology feature is made, is usedRepresent, and specify that it is start node topological characteristic collection S (0), i.e.,WhereinFor node topology concentrates i-th node topology feature, for example,Represent
In Q1,Q2,Q3,Q4,Q5Middle selection Q1,Q4,Q5Three node topology features of the node composition with general character,Table
Show in Q1,Q2,Q3,Q4,Q5Middle selection Q4,Q5Two node topology features of the node composition with general character.Here, we use 0,
1 represents that node whether there is in node topology feature respectively.
Next, using Species Competition thought, being carried out to the node topology feature in start node topological characteristic collection S (0)
Topology screening, topology are replaced, topology mutation, build node topology feature set of future generation, as shown in Figure 1.Here, present invention difference
Topology screening, topology are replaced, topology mutation is defined as follows：Topology screening refers to by random selection or by calculating target
Function selects some node topology features, in then directly applying to node topology feature set of future generation；It refers to work as that topology is replaced
The exchange of the local topology that may be carried out between any 2 node topologies in front nodal point topological characteristic collection, such as present node is opened up
Flutter feature set S (g)=10011,11011,01110,10001In the 1st node topology feature (10011) with the 2nd
(11011) node topology feature carries out localized swaps, and the 3rd node topology feature (01110) and the 4th (10001) section
Point topological characteristic carries out localized swaps, obtains new node topology for S (g+1)={ 11011,10011,01101,10010 }；Open up
It refers to the mutation carried out inside a node topology to flutter mutation, can be expressed as the addition of node topology feature interior joint or leave
The change of posterior nodal point topological characteristic, such as S (g)={ 10011 (additions), 11011(leaving),01110 (additions), 10001 (plus
Enter) }, it is S (g+1)={ 10111,11010,11110,10101 } that new node topological characteristic integrates after mutation.
For the quality of the node topology feature of evaluation structure is estimated, it is necessary to construct object function F (P), i.e.,：
F (P)=- ∑0 ＜ i, j≤q{[f(Qi,Qj)-g(Qi,Qj)+t(Qi,Qj)]*Gij*Bq[i]} (1)
Wherein P represents a node topology feature, and q represents the number of P interior joints, GijRepresenting between node i and node j is
It is no sensible, Bq[i] represents BqWhether the value of middle i-th bit, i.e., i-th node belongs to P.F () represents two node new feature Qi,QjIt
The incidence relation of interphase interaction is worth bigger explanation and introduces G to the influence degree of network coverage sizeijNetwork coverage is got over
Greatly；G () represents two node new feature Qi,QjInterphase interaction incidence relation to the influence degree of network transfer delay, value
Bigger explanation introduces GijNetwork transfer delay can be made smaller；T () represents two node new feature Qi,QjInterphase interaction association
The influence degree of relation pair network stabilization, is worth bigger explanation and introduces GijNetwork stabilization can be made higher；F (P) represents the node
The influence of dynamic network normal transmission state of the topological characteristic to being formed, is worth the smaller influence to normal transmission state smaller.
The node topology feature with relatively low objective appraisal value is concentrated to be used as the node topology spy of S (g+1) S (g)
Collection, after being then mutated by the topological screening in some generations, topology replacement, topology, if node topology concentrates each node to open up
The objective appraisal value for flutterring feature tends to convergence, just obtains the node topology feature set S in Training scene kk。
Finally, using the data set of multiple Training scenes, by node topology latent structure method, car networking is finally given
The node topology feature set of different time and different location Training scene, is that car networking is dynamically drilled around node topology feature set
The research of change method provides place mat.
The present invention utilizes car networking network node topological characteristic constitution step so that can be kept between car networking network node
Metastable connection；The covering of certain area can be reached, and influences the normal transmission state of dynamic network；Node topology feature
Dynamic network forming process is substantially reduced as car networking network structure, so as to improve the real-time of network interconnection intercommunication.
Using node topology feature and its measurement complete in the new node topology characteristic set for building, from overall angle, there is provided
Unified network Model and its calculation and analysis method, effectively analyze the network performance under high dynamic environment, solve
The access interconnection and interflow problem of car networking large scale network.
Step 2, the foundation of steady ordered network S measurements
Car networking is a complex network form for high dynamic change for cross-domain especially for extensive open loop, because
This, stability is not only one of its key property, and has influence on handling capacity, propagation delay time, performance, the effect of car networking network
All more important performance indications such as rate.Whether can quickly have in the car networking network of this high dynamic for current task R
Effect complete, it is necessary to one remain stabilization network system situation, the present invention this network state is defined as three classes, including
Stable state, nearly stable state, and away from stable state, the transformational relation between three is stable stateNearly stable stateAway from stable state, wherein, network stabilization state representation task R can perform；Nearly stable state represents that current task R can not
Perform, it is necessary to be self-regulated according to node topology feature, if stable state can be reached, continue executing with task R, otherwise, network
State evolution is, away from stable state, now to need to obtain new node from the car networking large scale network in actual scene again
Information, reconfigures new stabilizing network and continues executing with current task R, three kinds of shapes by car networking network process Evolution Theory model
State transfer process is as shown in Figure 2.
UtilizeThe foundation of steady ordered network S measurements, as heterogeneous networks in measurement car networking Dynamic Evolution
Sx (x=1,2 ..., the n) standard whether state is stablized.
Step 3, car networking network dynamic evolution
Car networking is a net for complicated form being made up of mutually coordinated between several species node, infiltration and restriction
Network system.Similarly, the faint change of node in the car networking node topology feature of present invention design can also cause network stabilization
The respective change of property, the stability of network directly determines whether current task R can perform.Corresponding to three kinds of network stabilization
State change, the present invention is defined as this dynamic evolution mechanism of car networking：Self adaptation → self-regulationSelf-healing, specifically such as
Under：First, each self-information of Real-time Collection node under car networking large-scale network environment, by network process Evolution Theory model
A network for steady ordered is set up, starts to perform current task R.During stable state, network implementation adaptation mechanism, with
Network is kept to be in stable state；Secondly, when network evolution begins to deviate from stable state to nearly stable state, it is necessary to implement certainly
Regulation mechanism, is adjusted to network stable state and continues executing with current task R；3rd, when network is in the non-thread of nearly stable state
Property area, small randomness fluctuation can cause network state self enlarge-effect, when network stabilization is gradually to away from stabilization
State infinite approach is, it is necessary to implement self-healing mechanism, its target is so that network evolution tends to nearly stable state, until may reach
State network to steady ordered continues executing with current task R.In extreme circumstances, network system is in away from stable state,
Showing self-healing mechanism can not solve current network systems state, and network stabilization has been collapsed, it is necessary to big rule of Real-time Collection again
Mould network node information, builds new stabilizing network and continues executing with current task R.Car networking network dynamic mechanism of Evolution such as Fig. 3 institutes
Show.
Using self adaptation, self-regulation, self-healing mechanism during car networking dynamical evolution, so as to begin in sensible sex chromosome mosaicism
Stable state is kept to provide safeguard eventually；Can supporting node or node topology feature missing when car networking dynamic evolution
Method, information flow remains flowing in real time and rationally convergence under realizing car networking large-scale network environment.
To sum up, technical solution of the present invention research method, the thought of Species Competition from ecology tries hard to description car connection
Condition and behavior of the net system in stable state change Near The Critical Point phase transformation, and degree of stability to system is described and divides
The reason for analysing its change and mechanism, from the car networking dynamic evolution problem of the new complicated form of angle close examination, with Species Competition
The extensive open loop of thought guidance car networking, the design of cross-domain dynamic evolution method, so as to be asked for car networking system provides solution
The new method of topic, this has important technical support meaning to lifting China in the core competitiveness in car networking field.
Subordinate list explanation
The VANET of table 1 correlation Routing Protocol comparative results
Brief description of the drawings
Fig. 1 node topology latent structure processes
Network stabilization State Transferring in Fig. 2 car networking Dynamic Evolutions
Fig. 3 car networking network dynamic evolving mechanisms
Fig. 4 node topologies feature and its measurement PxAnd Py
Fig. 5 node topology characteristic sets
Fig. 6 steady ordered network S and equalization point solution procedure
Fig. 7 car networking dynamical evolution method solution procedurees
Fig. 8 is the inventive method flow chart.
Specific embodiment
Technical solution of the present invention is further described below by way of accompanying drawing.
It is summarized as follows
Step 1, car networking network node topological characteristic building method, in order in the real-time field of car networking large-scale complex network
Under scape, the real-time of network system is improved, it is necessary to general character between wheeled digging machine networked node, and by some nodes with general character
Collectively constitute a node topology feature.For this reason, it may be necessary to prior (1,2 ..., n) research node under different off-line Training scene
Topological characteristic building method, obtains node topology feature set, so that for car networking network dynamic evolution method provides network structure.
Step 2, steady ordered network S measurement methods are set up, specific method comprises the following steps：
Step 21. according to node topology characteristic set, with reference to the current cross-domain actual scene of car networking large scale network open loop,
Network system S and Liapunov (LyaPunov) function V (N) are solved, is used to illustrate that network system S is stable；
Step 22. solves the equalization point (transition threshold) of steady ordered network S so that dynamic evolution system is in its equalization point
Neighbouring asymptotically stability, stability WSs of the network S in its equalization point is analyzed using Krasovsky (Krasovski) method, and
As the measurement scale of state in network dynamic evolutionary process (stabilization, nearly stabilization and away from stabilization) conversion.
Step 3, car networking network dynamic evolution method
The polytropy of the diversity and environment of car networking application demand, it is necessary to remain that the stabilization of network could meet and work as
The need for preceding task.Accordingly, it would be desirable to the dynamical evolution method of car networking complex network form is studied, so that for car networking is extensive
Stable state required for the application of network provides safeguard.
Describe in detail by the following examples
Embodiment
Specific implementation process of the invention, as shown in figure 8, including following 3 big steps：
First big step, node topology latent structure
For the purpose for reaching analysis network, predictive behavior, instructing Dynamic Evolution Model to design, it is necessary first to joined according to car
The node general character that net network node is obtained, structure node topological characteristic.Here, definition node topological characteristic of the present invention is by certain
The topological structure of number destination node connection composition so that the connection that can be kept relative stability between these nodes, reaches certain area
The covering in domain, and the normal transmission state of dynamic network is influenceed, the network of network consisting process evolution theoretical model is called again
Component.In fig. 4, by 1 roadside infrastructure node R SU, 1 3G cellular network node and 6 vehicle node C2x,C2q,
C2g,C2k,C2m,C2nThe region of covering, by node topology latent structure method, can construct various different node topologies special
Levy, for example, by after deep learning, according to four node Rs SU, C2x,C2k,C2nGeneral character and three nodes 3G, C2g,C2nBe total to
Property constructs two kinds of different node topology features respectively.It is compared to the individual node in car networking, node topology feature conduct
Car networking network structure substantially reduces dynamic network forming process, so as to improve the real-time of network interconnection intercommunication.Correspondence
Ground, definition node topological characteristic measurement of the present invention, represents all node (C of the node topology feature11,…,C1k) between keep
The relative time span being stably connected with and the dynamic network normal transmission state influence degree to being formed, use variable P=F
(C11,…,C1k) represent.In fig. 4, it is assumed that by node R SU, C2x,C2k,C2nThe node topology feature of composition can be completely covered
The region, and think that the node topology that this four nodes are constituted is characterized in most rational covering, then represent the node topology
Kept relative stability between all nodes in feature connection time it is most long, and to formed dynamic network normal transmission state
Influence is minimum, its node topology characteristic measure Px=F (RSU, C2x,C2k,C2n) represent；It is compared to Px, in addition by node 3G,
C2g,C2nThe node topology feature of composition, the time vice-minister of the connection that kept relative stability between node, and the dynamic to being formed
The influence time of network normal transmission state is small, uses Py=F (3G, C2g,C2n) represent.Px is less than Py in Fig. 4, illustrates the region by node
The node topology feature of RSU, C2x, C2k, C2n construction is most reasonable, by node 3G, the node topology feature time of C2g, C2n construction
It.
And so on, the data set under n different Training scenes, the node topology for constructing different time and different location is special
Levy, and integrated its is node topology characteristic set under complete car networking large-scale network environment, as shown in Figure 5.
Second largest step：Steady ordered network S and equalization point are solved
For this reason, it may be necessary to according to car networking network process Evolution Theory model and current vehicle intranet network actual scene, research
One state network S of steady ordered, as heterogeneous networks Sx (x=1,2 ..., n) shape in measurement car networking Dynamic Evolution
The standard whether state is stablized, to meet current task R the need for.
In order to solve the dynamic evolution method of car networking large scale network, it is necessary to build a network for steady ordered first
S is used as the scale for weighing State Transferring in Dynamic Evolution.According to node topology characteristic set, with reference to the big rule of current car networking
The cross-domain actual scene of lay wire network open loop, solves network system S and Liapunov (LyaPunov) function V (N), is used to
Bright network system S is stable；Solve the equalization point (transition threshold) of steady ordered network S so that dynamic evolution system exists
Asymptotically stability near its equalization point, stabilizations of the network S in its equalization point is analyzed using Krasovsky (Krasovski) method
Property WS, and as the measurement scale of state in network dynamic evolutionary process (stabilization, nearly stabilization and away from stabilization) conversion.Stabilization
Ordered network S and equalization point solution procedure are as shown in Figure 6.
The third-largest step：Car networking dynamical evolution method solution procedure, as shown in Figure 7.
All nodal informations are gathered from the cross-domain real-time network scene of the extensive open loop of car networking, using off-line training scene
The node topology characteristic set of lower structure, constructs the node topology feature P of current real-time network1,P2,…,Pl..., match energy
Enough cover the node topology feature P of currently practical network1,P2,…,Pl, so as to form initial dynamic stability network S1={ P1,
P2,…,Pl}。
With the movement of vehicle node, network S1Stability it is on a declining curve, whenWhen, due to network S1Section
Leaving for respective nodes may produce influence to its state in point topological characteristic, and the present invention utilizes Species Competition thought, implements
Adaptation mechanism.Here, present invention definition adaptation mechanism is as follows：When network is single or some node topology features in a certain section
When point or multiple nodes leave, topological change is carried out by screening node topology characteristics algorithm so that current network is still protected
Hold stable state.For example, the node topology feature P in a certain coverageu={ Q1,Q2,Q3,QlIn node Q2Leave,
Suboptimum node topology feature P is found by screening node topology characteristics algorithmv={ Q1,Qd,Q3Replace Pu, and Pu∩Pv={ Q1,
Q3, network can still keep stable state to perform current task R.
As network S1Start to develop and deviate stable state, ifDue to some start node topological characteristics in network
Node leave or disintegrate so that the connective of some regions produces influence in network, and the present invention utilizes Species Competition thought,
Implement Self-adjusting Mechanism.Here, Self-adjusting Mechanism is defined as follows：When in network some node topology features be destroyed, it is necessary to
Topological change is carried out by screening node topology characteristics algorithm, or is the regional choice by node topology feature dispatching algorithm
Other node topology features so that current network reaches stable state.For example, it is assumed that present steady state network S1Opened up by node
Flutter feature P1,P2,P3,P2,P5Constitute, i.e. S1={ P1,P2,P3,P2,P5, wherein P5={ Q1,Q2,Q3,Q5, work as P5Disintegrating causes
The connectivity of region of its covering is destroyed, it is necessary to implement Self-adjusting Mechanism.First, selection suboptimum node topology feature P6={ Q2,
Q3,Q6Replace P5To cover current region (P5∩P6={ Q2,Q3), if current suboptimum node topology feature is also destroyed,
Need to realize new node topological characteristic P using node topology feature dispatching algorithmkRegion overlay, andUntil
New network S2StabilityCurrent task R can be continued executing with.
As network S1After several times developing and implementing self adaptation/self-regulation repeatedly, new network system SkStability gradually to
Away from stable state infinite approach, ifNeed to implement self-healing mechanism.Here, the present invention defines self-healing mechanism
It is as follows：When great deal of nodes topological characteristic has been destroyed, it is necessary to pass through effective interpolation algorithm or node topology feature reconstruction in network
Method, is connective destroyed regional structure new node topological characteristic so that current network state connects to nearly network stabilization state
Closely.For example, it is assumed that current network Sk={ P1,P2,P3,P4,P5, if node topology feature P2,P3,P4It is now all unsatisfactory for cover
Cover area condition, network connectivty is destroyed, it is necessary to implement self-healing mechanism.First, the present invention is inserted by node topology feature
Mend and restructing algorithm, construct different node topology feature (wherein P1,P5Keep it is constant), and be combined into m kinds can cover work as
Preceding network SkThe network of overlay areaUse initial network setRepresent, wherein
For the network that i-th kind of node topology combinations of features is formed.To implementing competition mechanism between heterogeneous networks in SN (0), carry out g+1 times
Collection of network SN (g+1) is obtained after iteration, if obtaining a network S from SN (g+1)(g+1), its stabilityClose to WS,
Self-healing success is then thought, otherwise it is assumed that current network SkCurrent task R can not be continued executing with, it is necessary to resurvey current big rule
The cross-domain network real time node information of mould open loop, builds new dynamic stability network.Car networking dynamic evolution method solution procedure such as Fig. 7
It is shown.
Innovative point
One of innovation：The problem that interconnects for car networking in access, it is proposed that node topology latent structure method.
Interconnected aspect in car networking, current achievement in research is conceived to small yardstick closed-loop system and is between individual node substantially
Interconnect, it is impossible to fundamentally improve car networking real-time.The present invention is answered for the extensive open loop of car networking is cross-domain
Miscellaneous form, using Species Competition thought, builds node topology feature set such that it is able to the real-time requirement that rapid build interconnects
Network.
The two of innovation：It is access middle according to current task holding real-time stabilization network problem for car networking, it is proposed that one
Cover effective car networking dynamical evolution method for solving.Under car networking high dynamic network environment, stabilization how is remained
It is to solve the access another key issue of car networking the need for network state is to meet current task.The present invention is scientifically explained
Condition and rule that car networking dynamical network system is evolved, developed are released, it is three kinds that this complicated form network of car networking is dissected
Transformation mechanism：Self adaptation (stable state) → self-regulation (nearly stable state)Self-healing (away from stable state), and utilization kind
Group's Competitive Thought annotates whole evolution process.
Specification subordinate list
Table 1.
Claims (1)
1. a kind of car networking network dynamic evolution method based on Species Competition, it is characterised in that specific method includes following step
Suddenly：
Step 1, car networking network node topological characteristic construction
Assuming that in Training scene k, from time Tx(x=1,2 ..., r) take q node, by pretreatment simultaneously in the node of collection
The node new feature after deep learning is carried out for Q1,Q2,…,Qi,…,Qq；First, the regional structure m for being covered in q node
Individual different node topology feature, usesRepresent, and specify that it is start node topological characteristic collection S (0), i.e.,WhereinFor node topology concentrates i-th node topology feature；
Next, using Species Competition thought, topology is carried out to the node topology feature in start node topological characteristic collection S (0)
Screening, topology are replaced, topology mutation, build node topology feature set of future generation；Respectively to topology screening, topology replacement, topology
Mutation is defined as follows：Topology screening refers to select some node topologies special by random selection or by calculating target function
Levy, in then directly applying to node topology feature set of future generation；It refers to any 2 in present node topological characteristic collection that topology is replaced
The exchange of the local topology that may be carried out between individual node topology；
For the quality of the node topology feature of evaluation structure is estimated, it is necessary to construct object function F (P), i.e.,：
F (P)=- ∑0 ＜ i, j≤q{[f(Qi,Qj)-g(Qi,Qj)+t(Qi,Qj)]*Gij*Bq[i]} (1)
Wherein P represents a node topology feature, and q represents the number of P interior joints, GijRepresent whether lead between node i and node j
Reach, Bq[i] represents BqWhether the value of middle i-th bit, i.e., i-th node belongs to P.F () represents two node new feature Qi,QjBetween phase
The incidence relation of interaction is worth bigger explanation and introduces G to the influence degree of network coverage sizeijNetwork coverage is bigger；g
() represents two node new feature Qi,QjInterphase interaction incidence relation to the influence degree of network transfer delay, be worth bigger
Illustrate to introduce GijNetwork transfer delay can be made smaller；T () represents two node new feature Qi,QjInterphase interaction incidence relation
To the influence degree of network stabilization, it is worth bigger explanation and introduces GijNetwork stabilization can be made higher；F (P) represents the node topology
The influence of dynamic network normal transmission state of the feature to being formed, is worth the smaller influence to normal transmission state smaller；
The node topology feature with relatively low objective appraisal value is concentrated to be used as the node topology feature set of S (g+1) S (g),
Then after being mutated by the topological screening in some generations, topology replacement, topology, if node topology concentrates each node topology feature
Objective appraisal value tend to convergence, just obtain the node topology feature set S in Training scene kk；
Finally, using the data set of multiple Training scenes, by node topology latent structure method, car networking is finally given different
Time and the node topology feature set of different location Training scene；
Step 2, the foundation of steady ordered network S measurements
This network state is defined as three classes, including stable state, nearly stable state, and away from stable state, three it
Between transformational relation beWherein, network stabilization state representation task R
It is executable；Nearly stable state represents that current task R can not be performed, it is necessary to be self-regulated according to node topology feature, if energy
Stable state is reached, task R is continued executing with, otherwise, network state develops into away from stable state, now needs again from reality
New node information is obtained in car networking large scale network in scene, by car networking network process Evolution Theory model again structure
Make new stabilizing network and continue executing with current task R；
UtilizeThe foundation of steady ordered network S measurements, as heterogeneous networks Sx (x in measurement car networking Dynamic Evolution
=1,2 ..., the n) standard whether state is stablized；
Step 3, car networking network dynamic evolution
Corresponding to three kinds of state changes of network stabilization, this dynamic evolution mechanism of car networking is defined as： It is specific as follows：
First, each self-information of Real-time Collection node under car networking large-scale network environment, by network process Evolution Theory mould
Type sets up a network for steady ordered, starts to perform current task R；
During stable state, network implementation adaptation mechanism, to keep network to be in stable state；
Secondly, when network evolution begins to deviate from stable state to nearly stable state, it is necessary to implement Self-adjusting Mechanism, network is adjusted
It is whole to continue executing with current task R to stable state；
3rd, when network is in the inelastic region of nearly stable state, a small randomness fluctuation can cause network state self to be put
Big effect, when network stabilization gradually to away from stable state infinite approach, it is necessary to implement self-healing mechanism, its target is so that
Network evolution tends to nearly stable state, until the state network for being likely to be breached steady ordered continues executing with current task R；
In extreme circumstances, network system is in away from stable state, shows that self-healing mechanism can not solve current network systems
State, network stabilization has been collapsed, it is necessary to Real-time Collection large-scale network node information again, builds new stabilizing network and continue to hold
Row current task R.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710001115.6A CN106713477A (en) | 2017-01-03 | 2017-01-03 | Internet of Vehicles network dynamic evolution method based on population competition |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710001115.6A CN106713477A (en) | 2017-01-03 | 2017-01-03 | Internet of Vehicles network dynamic evolution method based on population competition |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106713477A true CN106713477A (en) | 2017-05-24 |
Family
ID=58905767
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710001115.6A Withdrawn CN106713477A (en) | 2017-01-03 | 2017-01-03 | Internet of Vehicles network dynamic evolution method based on population competition |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106713477A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107196835A (en) * | 2017-05-31 | 2017-09-22 | 同济大学 | The connection base component building method that car networking large scale network interconnects |
| CN108024205A (en) * | 2017-10-27 | 2018-05-11 | 北京理工大学 | Car networking moving advertising transmission method based on deep learning |
| CN108391249A (en) * | 2018-01-24 | 2018-08-10 | 长安大学 | A kind of traffic perception route method applied to car networking |
-
2017
- 2017-01-03 CN CN201710001115.6A patent/CN106713477A/en not_active Withdrawn
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107196835A (en) * | 2017-05-31 | 2017-09-22 | 同济大学 | The connection base component building method that car networking large scale network interconnects |
| CN107196835B (en) * | 2017-05-31 | 2020-08-14 | 同济大学 | Construction method of communication base component for interconnection and intercommunication of large-scale internet of vehicles |
| CN108024205A (en) * | 2017-10-27 | 2018-05-11 | 北京理工大学 | Car networking moving advertising transmission method based on deep learning |
| CN108024205B (en) * | 2017-10-27 | 2020-07-24 | 北京理工大学 | Internet of vehicles mobile advertisement propagation method based on deep learning |
| CN108391249A (en) * | 2018-01-24 | 2018-08-10 | 长安大学 | A kind of traffic perception route method applied to car networking |
| CN108391249B (en) * | 2018-01-24 | 2020-06-02 | 长安大学 | Traffic sensing routing method applied to Internet of vehicles |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106961707B (en) | Multifactor decision making Routing Protocol based on connectivity in a kind of VANET | |
| Saleet et al. | Intersection-based geographical routing protocol for VANETs: A proposal and analysis | |
| Ramamoorthy et al. | An enhanced hybrid ant colony optimization routing protocol for vehicular ad-hoc networks | |
| Qureshi et al. | Road aware geographical routing protocol coupled with distance, direction and traffic density metrics for urban vehicular ad hoc networks | |
| Belamri et al. | A survey on QoS routing protocols in Vehicular Ad Hoc Network (VANET) | |
| CN105722176B (en) | There is the connectivity method of the large scale scale heterogeneous network of the car networking of infrastructure in City scenarios | |
| Sutariya et al. | An improved AODV routing protocol for VANETs in city scenarios | |
| Rahimi et al. | A hybrid geographic-DTN routing protocol based on fuzzy logic in vehicular ad hoc networks | |
| Tee et al. | Survey of position based routing for inter vehicle communication system | |
| Arianmehr et al. | HybTGR: a hybrid routing protocol based on topological and geographical information in vehicular ad hoc networks | |
| Bouras et al. | A GPSR enhancement mechanism for routing in VANETs | |
| Gupta et al. | A survey on vehicular ad hoc networks | |
| CN106713477A (en) | Internet of Vehicles network dynamic evolution method based on population competition | |
| Smys et al. | Efficient self-organized backbone formation in mobile ad hoc networks (MANETs) | |
| Sharma et al. | A contemporary proportional exploration of numerous routing protocol in VANET | |
| Ghanbarzadeh et al. | Reducing message overhead of AODV routing protocol in urban area by using link availability prediction | |
| Choksi et al. | Power constrained performance evaluation of aodv, olsr and dsdv routing protocols for vehicular ad-hoc networks | |
| Saini et al. | Implementation and performance analysis of AODV routing protocol in VANETs | |
| Qureshi et al. | Study of efficient topology based routing protocols for vehicular ad-hoc network technology | |
| Aujla et al. | Comparative Analysis of AODV, DSR, GRP, OLSR and TORA by varying Number of Nodes with FTP and HTTP Applications over MANETs | |
| Sarkar et al. | VANET routing protocols in real-world mobility tracing | |
| Khamayseh et al. | Network size estimation in VANETs. | |
| Sbayti et al. | A new routing method based on ant colony optimization in vehicular ad-hoc network | |
| Dharani Kumari et al. | EGRP: Enhanced geographical routing protocol for vehicular adhoc networks | |
| Rondinone et al. | Exploiting multi-hop connectivity for dynamic routing in VANETs |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| WW01 | Invention patent application withdrawn after publication | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20170524 |