CN103427951B - With the data forwarding method that coding redundancy controls - Google Patents

Abstract

The invention discloses a kind of data forwarding method controlled with coding redundancy, belong to wireless network data retransmission technique field。This method is met the persistent period by utilizing the hybrid estimation method based on historical information to predict and estimates transmission capacity in real time, and then the degree of redundancy according to current network state information quantization encoding fused data exactly, complete coding fused data with maximum resource utilization rate for target to forward, specifically include following steps: step one: between pair, connection breaking wireless network initializes；Step 2: prediction node meets the persistent period；Step 3: estimate internodal transmission capacity；Step 4: data encoding chance judges and coding redundancy controls data and forwards。Coding fused data while ensureing reliable data transmission, can be carried out monitor in real time by the method, it is achieved promotes network resource utilization and minimizes the purpose of iterative redundant copy。

Description

With the data forwarding method that coding redundancy controls

Technical field

The invention belongs to wireless network data retransmission technique field, relate to a kind of data forwarding method controlled with coding redundancy。

Background technology

It is different from the data forwarding mechanism of tradition mobile ad-hoc network, between connection breaking wireless network make full use of the chance of meeting that node motion is brought, with more flexibly, " store-carry-forward " pattern realizes data transmission, it is adaptable to node is sparse, movement is frequent and connection has the applied environment of discontinuous nature。Similar with other wireless networks, a connection breaking wireless network also has the feature of resource-constrained, and therefore, data forwarding method is one of key technology of connection breaking wireless network, and rationally efficient data forwarding method can improve overall performance of network。

In recent years, research worker find to introduce in a connection breaking wireless network data repeating process network coding method can maximization network handling capacity, and upper largely reduce the impact that bottleneck link forwards, raising network transmission efficiency for data。Considering implementation complexity, the widely used random linear network encoding mode of a connection breaking wireless network forwards data, and its core concept is that base data replicas is constantly carried out linear weighted function fusion iteration by forward node, until destination node can recover initial data。But, for connection breaking wireless network between resource-constrained, this kind of mode will produce substantial amounts of iterative redundant copy, greatly occupy Internet resources。

For the network performance decline problem that iterative redundant copy causes, current existing coding data mechanism Network Based mainly has: (1) NTC mechanism is (referring to document Z.Song, J.Su, W.Peng, etal., NTC:TowardsEfficientNetworkCodinginDelayTolerantNetworks [C], InProc.InnovativeMobileandInternetServicesinUbiquitousCo mputing (IMIS), FifthInternationalConference, 2011.): this algorithm presets the ratio of coding fused data and base data replicas, when network encoding fused data and meeting decoding requirements, node will not continue to coding data, iterative redundant copy is under control；(2) HubCode (Hub-basedforwardingusingnetworkcoding) mechanism is (referring to document S.Ahmed, S.Kanhere, HUBCODE:hub-basedforwardingusingnetworkcodingindelaytole rantnetworks [C], WirelessCommunicationsandMobileComputings, 2011.): this algorithms selection has the node of high degree of communication as coding via node, only coding via node performs coding and decoding operation, and then complete data forwarding, thus reducing the purpose of network iterative redundant copy。But, both coding data forwarding mechanisms fail the data participating in iteration coding in transmitting procedure are carried out real-time management, when data carry node active degree higher time, these data will repeatedly participate in the coding fusion process of via node, take the Internet resources such as nodal cache, transmission opportunity for a long time, cause that limited Internet resources cannot obtain Appropriate application, and along with the network operation time is gradually increased, data iterative redundant copy amount also will rise therewith。

Summary of the invention

In view of this, it is an object of the invention to provide a kind of data forwarding method controlled with coding redundancy, in the method, node adopts the hybrid estimation method based on historical information to predict and meets the persistent period and estimate transmission capacity in real time, and then, degree of redundancy according to current network state information quantization encoding fused data exactly, completes coding fused data with maximum resource utilization rate for target and forwards。

For reaching above-mentioned purpose, the present invention provides following technical scheme:

A kind of data forwarding method controlled with coding redundancy, comprises the following steps: step one: between pair, connection breaking wireless network initializes；Step 2: prediction node meets the persistent period；Step 3: estimate internodal transmission capacity；Step 4: data encoding chance judges and coding redundancy controls data and forwards。

Further, in step one: when the network operation just starts, to node sets flag bit Flag all in network so that it is provide foundation, ordinary node flag bit Flag for data forwarding process_i=0, encode via node flag bit Flag_i=1；Record node Encounter Time t first respectively₀With time departure t₁, and the time duration X of meeting first of computing node₀=t₁-t₀。

Further, in step 2: utilize the hybrid estimation method based on historical information to predict that subsequent time node meets time duration X_i+1。

Further, for weight, the persistent period of meeting is weighted on average with α, 1-α (0 < α < 1), namelyWhereinRepresent that subsequent time node meets the estimated value of persistent period, X_iFor the actual value of current time Encounter Time, S_iAnd S_i-1Represent the exponential smoothing value of current time and previous moment and initial value S respectively₀For former seasonal effect in time series Section 1, i.e. Encounter Time first, α is smoothing factor。

Further, when persistent period estimated value of meeting is less than actual value time, the average of available history control information is to connecting Time Duration Error τ_iEstimate, ${\mathrm{\&tau;}}_i=\frac{{\mathrm{\&Sigma;}}_{k=1}^{i-1}{f}_k\left(t\right)}{\left|\right\{X_k\&GreaterEqual;{\hat{X}}_k,1\&le;k\&le;i-1\left\}\right|},$ Wherein1≤k≤i-1 is the difference of Encounter Time actual value and estimated value,The part Encounter Time representing more than estimated value adds up,Represent the estimated value number of times less than actual value；When estimated value is more than actual value, data are transmitted according to established rule, if disconnecting, the data for non-transmission success directly abandon and be not updated corresponding control information。

Further, in step 4, the Flag if two common via nodes meet_i=Flag_j=0, and in node i buffer memory, the destination node of data is j, i.e. M_{i_id}=j_id, then i is made directly data forwarding, simultaneously by M_{i_id}Coding number of times L be set to L=G；The Flag if common via node and coding via node meet_i=0 and Flag_j=1, then directly forward uncoded data M_i；If meeting, node is two coding via node Flag_i=Flag_j=1, the comparison according to summary vector, it is determined that need the data acquisition system of transmissionSize, the transmission capacity C according to estimating: ifData in forwarding cache successively, and record the coding number of times L of data respectively；IfCalculate data cached code machine meeting respectivelyAccording to transmission capacity size and code machine meetingDescending is sequentially completed data and forwards；If the persistent period of meeting of current time t prediction, less than t+1 moment actual measured value, namely also has remaining persistent period τ_iTime, then to τ_iIt is predicted, and continues in forwarding cacheIt is worth bigger data；So far, repeat above step, until total data is delivered complete in network。

Further, for moment t nodal cache data n, defining its code machine can be:Wherein Γ is the ttl value of this data n, and Γ-t is the residue life span of data n, and it is the exponential of λ that parameter is obeyed at node Encounter Time interval, L_nCoding number of times for data n。

Further, for current time t, the initial data coding number of times l of record is directly by accumulation calculating, i.e. l=l+1；For re-encoding data, first the coding number of times of record coding fused data is l₀=l₀+ 1, re-encoding number of times is done simple average value processing, i.e. l '=l₀/ n, wherein n is the initial data quantity merged；The coding number of times of all initial datas of classified statisticAs L >=L_maxDuring=G, code machine can set to 0。

The beneficial effects of the present invention is: the data forwarding method controlled with coding redundancy that the present invention proposes, can while ensureing reliable data transmission, coding fused data is carried out monitor in real time, it is achieved promote network resource utilization and minimize the purpose of iterative redundant copy。

Accompanying drawing explanation

In order to make the purpose of the present invention, technical scheme and beneficial effect clearly, the present invention provides drawings described below to illustrate:

Fig. 1 is the flow chart of the method for the invention；

Fig. 2 is that in the present invention, coding redundancy controls data forwarding mechanism flow chart；

Fig. 3 is that in the present invention, code machine can cognitive method。

Detailed description of the invention

Below in conjunction with accompanying drawing, the preferred embodiments of the present invention are described in detail。

Between there are two category nodes in connection breaking wireless network, i.e. coding via node and common via node, and all nodes random distribution in a network。According to the node type in network, data forwarding mechanism need to consider following three types, is between common via node respectively, between common via node and coding via node and between coding via node。

In the data forwarding mechanism controlled with coding redundancy, it is necessary first to the network that whole coding nodes is limited is initialized, and each node is arranged flag bit, ordinary node flag bit Flag_i=0, encode via node flag bit Flag_i=1；Record node Encounter Time t first simultaneously₀With time departure t₁, the time duration X of meeting first of computing node₀=t₁-t₀；Utilize and subsequent time node met time duration X based on the hybrid estimation method of historical information_i+1Estimate, namelyWherein, S_i、S_i-1Represent the exponential smoothing value of current time and previous moment and initial value S respectively₀For former seasonal effect in time series Section 1, i.e. Encounter Time first, α is smoothing factor。And then, utilize C=B X_i+1Estimate internodal transmission capacity, and the code machine according to each data can be that it reasonably arranges forwarding priority。The Flag if two common via nodes meet_i=Flag_j=0, and in node i buffer memory, the destination node of data is j, i.e. M_{i_id}=j_id, then i is made directly data forwarding, simultaneously by M_{i_id}Coding number of times L be set to L=G；The Flag if common via node and coding via node meet_i=0 and Flag_j=1, then directly forward uncoded data M_i；If meeting, node is two coding via node Flag_i=Flag_j=1, the comparison according to summary vector, it is determined that need the data acquisition system of transmissionRelatively transmission capacity estimated value C withSize: ifData in forwarding cache successively, and record the coding number of times L of data respectively；IfThen according to data cached code machine meetingAccording to transmission capacity size with descendingComplete to forward in advance choosing of data, and be sequentially carried out data forwarding。If the persistent period of meeting of current time t prediction, less than t+1 moment actual measured value, namely also has remaining persistent period τ_iTime, utilizeTo τ_iIt is predicted, and continues in forwarding cacheIt is worth bigger data。WhereinFor the difference of Encounter Time actual value Yu estimated value,Represent more than the part Encounter Time accumulated value of estimated value,Represent the estimated value number of times less than actual value。

In the method, realized data code machine meeting by the coding number of times of data in statistics nodal cachePerception。For adopting the node of network coding method, owing in network, the data of transmission are the coding fused data comprising initial data, but not initial data itself, therefore, data encoding number of times reflects its delivery state intuitively, and then, the degree of redundancy of coding fused data can be known by this parameter。In order to utilize limited transmission capacity fully, node needs the preferential coding data forwarding redundancy relatively low。Obviously, data participation forwarding and the chance of coding that redundancy is relatively low are relatively big, for moment t nodal cache data n, and its code machine meetingWherein Γ is the ttl value of this data n, and Γ-t is the residue life span of data n, and it is the exponential of λ that parameter is obeyed at nodes Encounter Time interval, L_nCoding number of times for data n。

Fig. 1 is the flow chart of the method for the invention, as shown in the figure, this method mainly comprises the steps that the network initializing coding via node limited amount, the estimation of inter-node transmission capacity, data encoding chance cognitive method and corresponding coding redundancy control data forwarding process。

Fig. 2 is that in the present invention, coding redundancy controls data forwarding mechanism flow chart, specifically, comprises the following steps:

1. initialize the network of coding via node limited amount。If node label position: Flag_i=0 is common via node；Flag_i=1 is coding via node；Record node Encounter Time t first respectively₀With time departure t₁, and the time duration X of meeting first of computing node₁=t₁-t₀；

The initialization of network: when the network operation just starts, to node sets flag bit Flag all in network so that it is provide foundation for data forwarding process；And computing node meets the persistent period first, as the initial value of the Duration Prediction that meets。

2., after netinit, utilize the node that obtains to meet first time duration X₁As the initial value of prediction, predict the time duration X of meeting of node subsequent time according to the hybrid estimation method based on historical information simultaneously_i+1, then, estimate internodal transmission capacity C。

Assume { X_i, i=1,2 ..., n} is that node meets duration time sequence, then any X in set_iIndependent uncorrelated, for weight, this time series is weighted consensus forecast with α, 1-α (0 < α < 1), namelyWhereinRepresent that subsequent time node meets the estimated value of persistent period, X_iFor the actual value of current time Encounter Time, S_iAnd S_i-1Represent the exponential smoothing value of current time and previous moment and initial value S respectively₀For former seasonal effect in time series Section 1, i.e. Encounter Time first, α is smoothing factor。

As noted above, node can estimate meeting the persistent period of subsequent time and other nodes at current time。In order to promote the order of accuarcy of estimated result further, the present invention adopts historical information that estimation difference is processed。When estimated value is less than actual value, utilize the average of history control information to connecting Time Duration Error τ_iEstimate,Wherein1≤k≤i-1 is the difference of Encounter Time actual value and estimated value,The part Encounter Time representing more than estimated value adds up,Represent the estimated value number of times less than actual value。

When estimated value is more than actual value, data are transmitted according to established rule, if disconnecting, the data for non-transmission success directly abandon and be not updated corresponding control information。

According to above-mentioned two situations, the available historical information of node adopts exponential smoothing time series models to estimate to meet the persistent period。When estimated result less than actual meet the persistent period time, node adopt the further estimation difference of conditional mean method。When estimated result more than actual meet the persistent period time, carry out data transmission according to established rule, the data of bust this no longer update control information, directly wait next time transmission opportunity。So far, complete node and meet the Approximate prediction of persistent period, namely ${\hat{X}}_{i+1}=\left\{\begin{array}{cc}{S}_i,& {\hat{X}}_{i+1}\&GreaterEqual;X_{i+1}\\ S_i+{\mathrm{\&tau;}}_i,& {\hat{X}}_{i+1}<X_{i+1}\end{array}\right..$

Internodal transmission capacity is estimated: the actual transmissions capacity B of link is the size of transmissible data volume in the node unit interval。Therefore, the communications capacity C=B X between node s and node r_i。

3. coding redundancy controls data forwarding process:

(1) Flag if two common via nodes meet_i=Flag_j=0, and in node i buffer memory, the destination node of data is j, i.e. M_{i_id}=j_id, then i is made directly data forwarding, simultaneously by M_{i_id}Coding number of times L be set to L=G；

(2) Flag if common via node and coding via node meet_i=0 and Flag_j=1, then directly forward uncoded data M_i；

(3) if the node that meets is two coding via node Flag_i=Flag_j=1, the comparison according to summary vector, it is determined that need the data acquisition system of transmissionSize, according to the transmission capacity estimated, comparesSize: ifData in forwarding cache successively, and record the coding number of times L of data respectively；IfCalculate data cached code machine meeting respectivelyAccording to transmission capacity size and descendingValue completes to forward in advance choosing of data, and is sequentially carried out data forwarding。

(4) if current time t prediction persistent period of meeting less than t+1 moment actual measured value, namely also have remaining persistent period τ_iTime, to τ_iIt is predicted, and continues in forwarding cacheIt is worth bigger data。

For moment t nodal cache data n, its code machine can beWherein Γ is the ttl value of this data n, and Γ-t is the residue life span of data n, and it is the exponential of λ that parameter is obeyed at nodes Encounter Time interval, L_nCoding number of times for data n。The present invention adopts random linear network encoding method, adopts the iteration method of operation in a network due to data, it is clear that the statistical result for encoding number of times is directly connected to the accuracy of code machine meeting cognitive method。The statistics of coding number of times needs multiple step format to carry out, and needs real-time update；Coding number of times need to distinguish counting for the particular make-up of coding fused data simultaneously。

The statistical following two situation of coding number of times: 1) node generates and meets coding via node after initial data and forward, and coding number of times is designated as original coding number of times l；2) when encoding fused data iteration again, it is designated as re-encoding number of times l '。For connection breaking wireless network between multi-source many purposes, in order to reduce the computation complexity of decoding further, the present invention adopts the mode in " generation " that data are encoded, define the initial data that for the purpose of same " generation " data, node is identical, " generation " be sized to G, represent the initial data quantity generated；Assuming at runtime, the sufficient and necessary condition of all data decoded reduction of energy being somebody's turn to do in " generation " is that the coding fused data quantity receiving linear independence must be at least G, defines the maximum coding number of times L of each initial data in same " generation "_max, then G=L_max。

Assuming current time t, owing to initial data direct coding is relatively big on the impact of decoding success rate, then the initial data recorded encodes number of times l directly by accumulation calculating, i.e. l=l+1；Coding again owing to encoding fused data makes its linear dependence bigger than initial data direct coding, and for re-encoding the statistics of number of times, first the coding number of times of record coding fused data is l₀=l₀+ 1, but for each initial data merged, any initial data merged due to coding fused data is identical on the impact of decoding gain, therefore re-encoding number of times should be done simple average value processing, to shorten the processing delay of node as far as possible, ensure the accuracy of code machine meeting perception, i.e. l '=l simultaneously₀/ n, wherein n is the initial data quantity merged。To sum up, the coding number of times of all initial datas of classified statisticAs L >=L_maxDuring=G, code machine can set to 0。

It is illustrated in figure 3 the detailed process that the code machine meeting cognitive method of the present invention performs in connection breaking wireless network between multi-source many purposes。Assuming that node a, b, c are for coding via node, S, R, P, T is source node, node for the purpose of D；S, R, P, T carry initial data { x respectively₁,x₂,x₃,x₄}。t₁In the moment, a receives S and the R base data replicas x forwarded₁,x₂If a still has buffer memory, it is directly stored, and send outline data vector；T₂In the moment, a and b meets, and a forwards F after himself carries data encoding₁, forward code coefficient simultaneously and record x₁,x₂Coding number of times；T₃To t₇Moment nodal operation all ibid, but at t₈In the moment, forward after the data encoding that selection code machine can be big, rather than directly by all data encodings on buffer memory；T₉In the moment, D receives decoding success during the coding fused data of 4 linear independences。

What finally illustrate is, preferred embodiment above is only in order to illustrate technical scheme and unrestricted, although the present invention being described in detail by above preferred embodiment, but skilled artisan would appreciate that, in the form and details it can be made various change, without departing from claims of the present invention limited range。

Claims (5)

1. the data forwarding method controlled with coding redundancy, it is characterised in that: comprise the following steps:

Step one: between pair, connection breaking wireless network initializes；

Step 2: prediction node meets the persistent period；

Step 3: estimate internodal transmission capacity；

Step 4: data encoding chance judges and coding redundancy controls data and forwards；

In step one: when the network operation just starts, to node sets flag bit Flag all in network so that it is provide foundation, common via node flag bit Flag for data forwarding process_i=0, encode via node flag bit Flag_i=1；Record node Encounter Time t first respectively₀With time departure t₁, and the time duration X of meeting first of computing node₀=t₁-t₀；

In step 2: utilize the hybrid estimation method based on historical information to predict that subsequent time node meets time duration X_i+1；

With α, 1-α, the persistent period of meeting is weighted on average for weight by 0 < α < 1, namelyWhereinRepresent that subsequent time node meets the estimated value of persistent period, X_iFor the actual value of current time Encounter Time, S_iAnd S_i-1Represent the exponential smoothing value of current time and previous moment and initial value S respectively₀For former seasonal effect in time series Section 1, i.e. Encounter Time first, α is smoothing factor。

2. the data forwarding method controlled with coding redundancy according to claim 1, it is characterised in that: when persistent period estimated value of meeting is less than actual value time, the average of available history control information is to connecting Time Duration Error τ_iEstimate,Wherein1≤k≤i-1 is the difference of Encounter Time actual value and estimated value,The part Encounter Time representing more than estimated value adds up,Represent the estimated value number of times less than actual value；When estimated value is more than actual value, data are transmitted according to established rule, if disconnecting, the data for non-transmission success directly abandon and be not updated corresponding control information。

3. the data forwarding method controlled with coding redundancy according to claim 2, it is characterised in that: in step 4: the Flag if two common via nodes meet_i=Flag_j=0, and in node i buffer memory, the destination node of data is j, i.e. M_{i_id}=j_id, M_{i_id}For data, j in node i buffer memory_idFor node j address；Then i is made directly data forwarding, simultaneously by M_{i_id}Coding number of times L be set to L=G, G and represent the initial data quantity generated；The Flag if common via node and coding via node meet_i=0 and Flag_j=1, then directly forward uncoded data M_i；If meeting, node is two coding via node Flag_i=Flag_j=1, the comparison according to summary vector, it is determined that need the data acquisition system of transmissionSize, the transmission capacity C according to estimating: ifData in forwarding cache successively, and record the coding number of times L of data respectively；IfCalculate data cached code machine meeting respectivelyAccording to transmission capacity size and code machine meetingDescending is sequentially completed data and forwards；If the persistent period of meeting of current time t prediction, less than t+1 moment actual measured value, namely also has remaining persistent period τ_iTime, then to τ_iIt is predicted, and continues in forwarding cacheIt is worth bigger data；So far, repeat above step, until total data is delivered complete in network。

4. the data forwarding method controlled with coding redundancy according to claim 3, it is characterised in that: for moment t nodal cache data n, defining its code machine can beWherein Γ is the ttl value of this data n, and Γ-t is the residue life span of data n, and it is the exponential of λ that parameter is obeyed at node Encounter Time interval, L_nCoding number of times for data n。

5. the data forwarding method controlled with coding redundancy according to claim 4, it is characterised in that: for current time t, the initial data coding number of times l of record is directly by accumulation calculating, i.e. l=l+1；For re-encoding data, first the coding number of times of record coding fused data is l₀=l₀+ 1, re-encoding number of times is done simple average value processing, i.e. l'=l₀/ n, wherein n is the initial data quantity merged；The coding number of times of all initial datas of classified statisticAs L >=L_maxDuring=G, code machine can set to 0。