Summary of the invention
The purpose invented herein, which is to provide one kind, preferentially to transmit high value according to the value of information in car networkingThe car networking synthetical information evaluating method of information.
Specifically, the present invention provides a kind of car networking synthetical information evaluating method, includes the following steps:
Step 100 carries out unified data characteristics parameter characterization to all information in current car networking, and it is special to quantify each parameterLevy the weight shared in scaling information value；
Step 200 carries out estimation calculating by property parameters and information dimension of the weight to every information, obtains each informationBased on the information value size under specified services parameter；
Step 300, the threshold value for setting filter information value, abandon the information of not up to threshold value, and qualification is believedBreath, which saves to data to send, to be collected；
Step 400, the information for sending concentration to data are analyzed, and exception information therein is cleaned, meanwhile, it is rightEmpty information and missing information carry out information and fill a vacancy operation, and the information that can not be handled is marked；
Step 500, then by information qualified after processing according to its be worth the big minispread of weight with determine information passSend sequence.
In an embodiment of the invention, the weight calculation mode in the step 100 is as follows:
Step 111. determines all information with information content from a new perception period；
Step 112. utilizes the potential feature calculation of environmental information and the sensing region of selection most information content；
Step 113. obtains the detected value of each sensor in sensing region and forms set, according to any one sensorDetected value and corresponding detection accuracy and bulk properties detected value obtain the entropy of the detected value；
Any one information can be obtained according to the errored message degree of sensor and the entropy of corresponding detected value in step 114.Weight.
In an embodiment of the invention, in the step 111, determine that there is letter in the perception period new from oneThe information mode of breath amount is as follows:
Wherein, SnFor the information content after determination, j is perception period, CjFor whole space time unit set, SjWhen to have perceivedDummy cell set, and haveK is the information content quantity of selection, and Cn is all information content, Ω (Xs∪v；XCn\v) it is space-timeThe space time unit Cn of unit S ∪ v mutual information between v.
In an embodiment of the invention, step 113 medium entropy to obtain process as follows:
If x1、x2…...xn, it is the detected value of sensor in sensing region, takes any n detected value composition set, d1,d2... ... dnFor the precision of respective sensor, the overall attribute value that detects isIf m*n matrix, wherein xj is j-thAttribute value, the then entropy of any detected value are as follows:
In an embodiment of the invention, the calculation formula of weight is as follows in the step 114:
Its k is constant, k=1/lnm, cjFor errored message degree, and cj=1-Ej, the entropy of detected valueWjFor the weight of j-th of information.
In an embodiment of the invention, in the step 200, the process for obtaining information value size is as follows:
Total number of events in step 211, a cycle determining first, using the probability that wherein everything part occurs as the thingThe characteristic parameter value of part, determines the information content of everything part；
Step 212, the variation probability that current event is determined according to the variation speed of information content, recycle the weight of event i.e.The information value calculation formula of current information can be obtained.
In an embodiment of the invention, in the step 211, the mode for obtaining the information content is as follows:
The event space for defining information source first is S={ S1,...Si,...,SM, wherein the total number of event is M, Si={xi,x2,...xiIndicate an event, xi,x2,...xiThe value of each characteristic parameter is respectively indicated, then event SiInformation content is；
Ii=-logpi
Wherein, piThe probability occurred for event i.
In an embodiment of the invention, change the formula of probability in the step 212 are as follows:
pij=Γ (pi,t)
If information matrix is ∏, then γ is ΠiCharacteristic value (γ1, γ2... γn) constitute diagonal matrix, t be perceptionPeriod.
In an embodiment of the invention, the calculation formula of the information value H in the step 212 are as follows:
Wherein, pijFor the probability of event change, wiFor the significance level weight of event, M is the total number of event.
In an embodiment of the invention, in the step 500, the value weight size calculation formula of data informationIt is as follows:
wi=f (αi,βi)
Wherein, the threshold value of information value size isIt is information value weight ΩiWith event SiFeature ginsengAmount, if information matrix is Πi, γ is ΠiCharacteristic value (γ1, γ2... γn) constitute diagonal matrix, η is each characteristic value pairFeature vector (the η answered1, η2... ηn) composition matrix, target is to SiDesirability βiIt indicates,It choosesCorresponding feature vector (the η of preceding i characteristic value1, η2... ηn) composition characteristic matrix αi, and f indicates wiIt is about αiAnd βiLetterNumber.
The present invention constructs car networking information value comprehensive evaluation model based on information value, to bulk information intoThe unified data characteristics parameter characterization of row, quantization (data) the feature weight shared in scaling information value, and foundationThe threshold value of information value selects the information of transmission, at (the meeting business demand) information for reaching threshold valueReason, and the information of not up to threshold value (not meeting business demand) is then abandoned, and using the data transmission mechanism of communication for coordination, is mentionedThe utilization rate of high-transmission resource has been finally reached the purpose for promoting network overall transfer performance.
Specific embodiment
The present invention for network environment in car networking dynamic is changeable, complicated variety of physical environment and data transmission seaThe features such as amount property and real-time, constructs the car networking synthetical information evaluating model based on information value, makes all kinds of letters in car networkingBreath value reaches balanced with transmission, plays important supporting role to promote traffic efficiency, safety etc..In addition, car networking networkState changes at random, in network entity have real-time dynamic reorganization characteristic, cause it is very high to networked-induced delay requirement, becauseThis, makes full use of limited resources, is task first to meet the preferential transmission of high price value information.
As shown in Figure 1, in an embodiment of the invention, a kind of car networking synthetical information evaluating method is provided, withPreferential transmission is worth higher information within the limited time, and steps are as follows for execution:
Step 100 carries out unified data characteristics parameter characterization to all information in current car networking, and it is special to quantify each parameterLevy the weight shared in scaling information value；
Information therein includes OBD data, nearby vehicle information, voice data and the video counts that car-mounted terminal acquires in real timeTraffic information etc. is set according to, road.
As shown in Fig. 2, the obtaining step of weight is as follows:
Step 111. determines all information with information content from a new perception period；
Method of determination is as follows:
Wherein, SnFor the information content after determination, j is perception period, CjFor whole space time unit set, SjWhen to have perceivedDummy cell set, and haveK is the information content quantity of selection, and Cn is all information content, Ω (Xs∪v；XCn\v) it is space-timeThe space time unit Cn of unit S ∪ v mutual information between v.
Step 112. utilizes the potential feature calculation of environmental data and the sensing region of selection most information content；
Step 113. obtains the detected value of each sensor in sensing region and forms set, according to any one sensorDetected value and corresponding detection accuracy and bulk properties detected value obtain the entropy of the detected value；
Its medium entropy to obtain process as follows:
If x1、x2…...xn, it is the detected value of sensor in sensing region, takes any n detected value composition set, influenceThe principal element of the data precision is sensor accuracy and error, and sensor accuracy is only related with sensor itself, and value is solidFixed, error is generally caused by environment and other equipment, main to consider history deviation and overall equal deviation.d1, d2... ... dnForThe precision of respective sensor, the overall attribute value that detects areAssuming that m*n matrix, wherein xjFor j-th of attribute value, thenThe entropy of any detected value are as follows:
Any one information can be obtained according to the errored message degree of sensor and the entropy of corresponding detected value in step 114.Weight.
The calculation formula of weight is as follows:
Its k is constant, k=1/lnm, cjFor errored message degree, and cj=1-Ej, the entropy of detected valueWjFor the weight of j-th of information.
Step 200 carries out estimation calculating by property parameters and information dimension of the weight to every information, obtains each informationBased on the information value size under specified services parameter；
Wherein, the process for obtaining information value size is as follows:
Total number of events in step 211, a cycle determining first, using the probability that wherein everything part occurs as the thingThe characteristic parameter value of part, determines the information content of everything part；
The mode for obtaining information content is as follows:
The event space for defining information source first is S={ S1,...Si,...,SM, wherein the total number of event is M, Si={xi,x2,...xiIndicate an event, xi,x2,...xiThe value of each characteristic parameter is respectively indicated, then event SiInformation content is；
Ii=-log pi
Wherein, piThe probability occurred for event i.
Step 212, the variation probability that current event is determined according to the variation speed of information content, recycle the weight of event i.e.The information value calculation formula of current information can be obtained.
Because the value of information is related with the significance level of information and complexity, complexity is again fast with the variation of informationIt is slow related, so the formula of variation probability are as follows:
pij=Γ (pi,t)
If information matrix is Π, then γ is ΠiCharacteristic value (γ1, γ2... γn) constitute diagonal matrix, t be perceptionPeriod.
The calculation formula of information value H are as follows:
Wherein, pijFor the probability of event change, wiFor the significance level weight of event, M is the total number of event.
In addition, it can include the step of being screened to the information after determining information value:
Characteristic parameter matrix and target first by everything part after probability extracts obtain the desirability of the eventTo the weight expression formula for indicating information importance level；
Weight expression formula is substituted into value formula by information source again and obtains the initial information value of event, then output is extremely believedPlace, adjustment self-demand degree feeds back to information source again after initial information value is analyzed in the stay of two nights, and information source is based on the feedback signalOptimize weight expression formula and be then forwarded to the stay of two nights after recalculating, the stay of two nights feeds back new desirability to information source, weight again after receivingMultiple process, until obtaining the information value standard for meeting pre-provisioning request.
Step 300, the threshold value for setting filter information value, abandon the information of not up to threshold value, and qualification is believedBreath, which saves to data to send, to be collected；
In this step, it is all to save all qualified data informations to data transmission collection, just jumps toStep 400, preservation processing is not influenced on abandoning for the data information for not meeting threshold value during processing.
Step 400, the information for sending concentration to data are analyzed, and exception information therein is cleaned, meanwhile, it is rightEmpty information and missing information carry out information and fill a vacancy operation, and the information that can not be handled is marked；
Here cleaning refers to that individual information deviates most of data, then is tentatively judged as it and obtains adopting for bias dataIt is abnormal to collect node, which is excluded.
Step 500, then by information qualified after processing according to its be worth the big minispread of weight with determine information passSend sequence.
As shown in Fig. 2, the value weight size calculation formula of information is as follows:
wi=f (αi,βi)
Wherein, the threshold value of information value size isIt is information value weight ΩiWith event SiFeature ginsengAmount, if information matrix is Πi, γ is ΠiCharacteristic value (γ1, γ2... γn) constitute diagonal matrix, η is each characteristic value pairFeature vector (the η answered1, η2... ηn) composition matrix, target is to SiDesirability βiIt indicates,It choosesCorresponding feature vector (the η of preceding i characteristic value1, η2... ηn) composition characteristic matrix αi, and f indicates wiIt is about αiAnd βiLetterNumber.
Present embodiment based on information value, constructs car networking information value overall merit mould by this upper stepType can be realized effective transmission of emergency message.Unified data characteristics parameter characterization is carried out to bulk information, is quantified (data)The feature weight shared in scaling information value, and selected according to information of the threshold value of information value to transmissionIt selects, (the meeting business demand) information for reaching threshold value is handled, and not up to threshold value (not meeting business demand)Information is then abandoned, and using the data transmission mechanism of communication for coordination, is improved the utilization rate of transfer resource, has been finally reached promotion netThe purpose of network overall transfer performance.
So far, although those skilled in the art will appreciate that present invention has been shown and described in detail herein multiple showsExample property embodiment still without departing from the spirit and scope of the present invention, still can according to the present disclosure directlyDetermine or deduce out many other variations or modifications consistent with the principles of the invention.Therefore, the scope of the present invention is understood that and recognizesIt is set to and covers all such other variations or modifications.