CN104753644A - Delay-tolerant and disruption-tolerant network transmission system on basis of network coding - Google Patents

Abstract

The invention discloses a delay-tolerant and disruption-tolerant network transmission system on the basis of network coding. Local network environments and message correlation attributes are balanced, models for accepting delay-tolerant and disruption-tolerant network coding messages are built on the basis of game theories, and network information monitoring and sensing, message accepting games, coding and decoding are implemented by the delay-tolerant and disruption-tolerant network transmission system. The current local network environments and the message correlation attributes are acquired by means of monitoring and sensing at first, and information is transmitted to the message accepting games; game tree construction procedures are completed by the accepting games on the basis of message forwarding effectiveness computing according to revenue functions of participants, and then refined Nash equilibrium of sub-games is solved according to backward induction processes and is delivered for coding and decoding; to-be-accepted messages are coded and broadcast by means of coding and decoding according to game results, and neighbor nodes are decoded according to the game results. The delay-tolerant and disruption-tolerant network transmission system has the advantages that the adaptability of the network coding to the different network environments and different message types can be enhanced, and the network transmission efficiency can be improved.

Description

The appearance of coding Network Based holds circuit network transmission system late

Technical field

The appearance that the present invention relates in computer science holds circuit network field late, particularly relates to the transfer of data held and hold coding Network Based in circuit network late.

Background technology

Rong Chi holds circuit network (Delay Tolerant Network, DTN) because its feature such as topology mutability, network survivability, chance transmission, node isomery etc. obtains giving more sustained attention of academia and industrial quarters, deep space, military affairs, the extreme environment such as have been applied at present under water.DTN is unconnected network, there is not end-to-end path, rare link makes repeater-transmitter can be extremely valuable, there is a large amount of Messages-Waiting in nodal cache to forward simultaneously, when node obtain a valuable repeater-transmitter can time, it always expects in of short duration link Lifetime, to have forwarded all message to the opposing party.But generally, be limited to node motion speed and data transmission rate, shorter Lifetime often can only send a small amount of message, after even only sending a piece of news, link just disconnects, namely a message request of a neighbor node can only be met, now, in finite time, how to send information as much as possible and become study hotspot.

Network code (Network Coding) is as the data processing method be applied in network data transmission process, effectively can improve the throughput of network, reduce the consumption of transfer of data energy, strengthen network reliability, especially in the DTN network that network condition is extremely limited.Thus the appearance of coding Network Based is held circuit network transmission technology late and has been attracted the concern of domestic and international research institution and scholar.The feature of network code is single message load capacity, multiple messages amount of information, and hold circuit network related mechanism late by means of appearance, a coded message can meet multiple message request of multiple neighbor node simultaneously, improves network transmission efficiency.

Summary of the invention

The present invention holds circuit network efficiency of transmission for target late to improve to hold, and merge the thought of dynamic game, design and Implement the data transmission system of coding Network Based, its allomeric function module as shown in Figure 1.Specifically comprise:

1, the monitoring perception of the network information

Information is monitored perception and is comprised monitoring neighbours' messaging list and perception message attributes, the former obtains neighbor node messaging list by heartbeat machine contracting system, draws each neighbours' message request situation, the latter utilizes the message association attributes perceived to quantize message and forwards effectiveness, and both all rely on the broadcast mechanism holding and hold circuit network late.

2, decision-making is received based on game theoretic message

Receive the game stage in message, based on the information obtained on last stage, consider and be decoded into power and encoding and decoding time overhead, computing cost, at utmost meet neighbours' message request.Message receives game to be subordinate to Complete Information Dynamic Game, and component comprises participant, the actionable space of participant, the strategy set of the action sequence of all participants, the income of single participant or game total revenue.Complete Information refers to that the essential information of each message can be obtained by other message, comprises and forwards effectiveness, actionable space, revenue function etc.; Dynamically refer to order that message does decision-making have priority point, message is when determining whether to add coding with reference to the decision of preceding one, and the decision of self affects the decision of continue, and then affects single income and total revenue.So just can be created as game theoretic model by holding the coded message holding circuit network late problem of receiving.

3, encoding and decoding

Decision-making Game is received to filter out optimal candidate code set from message bag to be sent set, they are combined as an encoded packets by chance coding, sent by broadcast mechanism, neighbor node is monitored and is determined whether receive, if it is in conjunction with own message queue decoding and coding bag, asked message is obtained.

Compared with prior art, innovation of the present invention is:

1) add the network information and monitor perceptional function, network code is by network environment influence, and message receives game to make decision based on monitoring the network information perceived, and can strengthen the adaptability of network code to dynamic network, improve efficiency of transmission.

2) design based on game theoretic receiving, refusal revenue function combine closely message attributes and network environment, be applicable to different messages type and network state, the present invention dynamically makes the decision-making of coded message receiving or refusal according to current network conditions.

Accompanying drawing explanation

Fig. 1 holds circuit network encoding function module map late based on game theoretic appearance.

Fig. 2 is that messaging list monitors schematic diagram.

Fig. 3 is that message receives game theory.

Fig. 4 is that message receives game beta pruning.

Fig. 5 is encoding and decoding schematic diagram.

Embodiment

1. the monitoring perception of the network information

The network information is monitored and is depended on broadcast mechanism.Node periodically sends heartbeat packet to the neighbor node in oneself range of signal, monitors the heartbeat packet of neighbor node simultaneously.The message numbering list that the field that heartbeat packet comprises is node serial number, node memory stores up and timestamp.By heartbeat packet mechanism, each node maintains an information of neighbor nodes storehouse, records each neighbor node message sink situation, sees Fig. 2.Neighbours A, C, D of sending node B broadcast respective heartbeat packet, and B monitors, and is then stored to information of neighbor nodes storehouse.Information bank realizes periodic refreshing function, and after exceeding certain hour, heartbeat packet is eliminated, and shows that corresponding node is no longer neighbor node.

The forwarding effectiveness that network information perception is used for sensing network state, quantizes network parameter, calculates message.Following parameter is periodically gathered: network node sum L, message TTL life cycle in network operation process _i, message copy number n _i(T _i), received the number of nodes m of this message _i(T _i), node approach frequency λ, message priority pri _i, neighbor node is to the demand degree r of message _{neighbor (i)}/ r _neighbor, then calculate the forwarding effectiveness of message.Parameters implication is in table 1.

Table 1 network information perceptual parameters

Message forwards effectiveness and is determined by parameter in table 1, considered network size, node met characteristic, the message sum of the neighbor node number of message distribution situation in a network, message residue life span, message priority, request message and neighbor node request, using these parameters as the factor, the message under computing network coding environment forwards effectiveness:

$U_i=\frac{r_{\mathrm{neighbor}\left(i\right)}}{r_{\mathrm{neighbor}}}(1-\frac{m_i\left(T_i\right)}{L-1})\mathrm{\λ}{R}_i\exp (-\mathrm{\λ}{n}_i\left(T_i\right)R_i-{\mathrm{pri}}_i)$

2. receive decision-making based on game theoretic message

Message receives Decision-making Game to be defined as G={P, and A, D, U}, P, A, D and U are expressed as the four factor of this game---and participant, actionable space, strategy, revenue function, wherein participant is massage set P={PR _i, actionable space A={Admit, Reject}, represent that message is accepted or refuses, tactful D={a _ithe action sequence taked by all message forms, revenue function U comprises two parts, is the revenue function of single message respectively and receives game integral benefit function, U={u _i, Σ ui}.

Revenue function in receiving Decision-making Game process, in table 2, only extracts two adjacent participants here, the income of two participants under displaying Order Strategic Decision, wherein PR _iprior to PR _jaction.First row represents PR _iaction, the first row represents PR _jat PR _ithe action that action basis is taked, first element representation PR in special instruction (receive, receive) _ipR during selection " receiving " _jselection, second element representation PR _ipR during selection " refusal " _jselection.

Table 2 dynamic game gain matrix table

Gain matrix comprises receives income fu-fu _min+ c+p-f and refusal income fu _min-fu, with participant PR _ifor example illustrates each part of revenue function:

(1) fu _i: participant PR _imessage forward value of utility, do not repeat them here;

(2) fu _min: the minimum message of all participants forwards effectiveness, fu _min=min{fu _i;

(3) c _i: node is to receiving PR _ithe payoff of encoded packets, c _i=α * use _i, use _ireceive PR _ithe nodes that rear encoded packets is useful.Useful standard is: if encoded packets is made up of n message bag, comprises m message bag wherein in nodal cache, is just referred to as useful during and if only if n=m+1.If certain neighbor node request PR _i, and PR _ialso be accepted and add encoded packets, owing to containing PR in encoded packets _iinformation, be valuable concerning this encoded packets this neighbor node, at this moment neighbor node can give PR _icertain award, i.e. payoff;

(4) p _i: receive PR _ithe newly-increased potential value of encoded packets, p _i=β * poten _i, poten _ithink to receive PR _ithe nodes of encoded packets potentially useful.If certain neighbor node does not ask PR _i, that is contain PR in neighbor node _ialthough, receive PR _iafter encoded packets concerning this neighbor node, do not increase amount of information, if but after the follow-up message having it to ask is accepted, encoded packets contributes to decoding effort, therefore neighbor node needs to PR _icertain award, namely potential value is rewarded, can be seldom even if reward;

(5) f _i: node is to receiving PR _ithe penalty value of encoded packets, f _i=γ * punish _i, punish _ithink to receive PR _ithe failed nodes of encoded packets decoding.If certain neighbor node request PR _kand PR _i, but PR _kbe included in encoded packets, if at this moment encoded packets receives PR _ithis neighbor node can be made successfully not decode, and encoded packets becomes useless by useful, and now this neighbor node can give PR _iapply certain penalty value.

Dynamic game uses game tree representation more, and message receives game theory to see Fig. 3.In dynamic game, action always has sequencing, and message receives game message to be forwarded effectiveness order from high to low as receiving game order.Namely the process building game theory is the process that message carries out receiving according to the order of sequence, after forerunner's participant of participant makes decisions, this participant calculates the revenue function that in oneself actionable space, all decision-makings are corresponding, namely income and refusal income is received, game theory is expanded one deck downwards, follow-up participant by that analogy, until game theory arrive destination node, the now income of destination node record from root to leaf, namely each participant indicates the income of making decision by this path.

After the modeling of receiving Controlling model, determine the result of game, namely sub-game refined Nash equilibrium solution is asked, adopt reverse induction, namely reverse be progressively trace back to first participant from last participant, namely conclude is from multiple behavior branch, pick out optimum branching according to participant income, cuts off remaining branch, sees Fig. 4.Branch composition payoff remaining in game theory, determines to receive according to the branch of each participant and still refuses this message, finally all message composition coded message set P be accepted.

3. encoding and decoding

Message enters coded message set P according to receiving the Nash Equilibrium Solution of game, all message packages in set are synthesized an encoded packets and broadcast away by coding module, all neighbor nodes can listen to, only have this encoded packets just can receive oneself useful neighbor node and open decoding schema, see Fig. 5.

Suppose p ₁, p ₂..., p _nthe message in candidate code set P, coding module according to generate encoded packets p, here g ₁..., g _nrepresent the code coefficient that each message is corresponding, value is 1, and the operation method of Message Payload is step-by-step XOR.Why adopt coefficient be 1 linear coding scheme be because message encode in upper hop, in down hop and decodable code, participate in coding message number fewer under normal circumstances, linear network encoding can effectively save node calculate energy expense.Suppose p ₁, p ₂..., p _nmessage relevant to P in certain neighbor node message queue, decoder module according to solve origination message bag p _k, g here ₁..., g _nbe still 1.

Claims (5)

1. the appearance of coding Network Based holds circuit network transmission system late, specifically comprises the monitoring perception acquisition of the network information, message receives game, encoding and decoding; It is characterized in that:

It is obtain current local network context and message association attributes that the network information is monitored with perception object, and these information is passed to message receiving game.Message receives game based on the network information obtained, each participant is taked to receive or refusal action, and calculates corresponding receiving or refusal income, and then completes the structure of whole game theory, then ask Nash Equilibrium Solution according to reverse induction, and consign to encoding and decoding.Encoding and decoding according to payoff by be accepted message coding, broadcast, the decoding of last neighbor node;

Wherein, call access control decision-making adopts Game Theory, comprises following aspect:

1) message receives Decision-making Game to be defined as G={P, and A, D, U}, P, A, D and U are expressed as the four factor of this game---and participant, actionable space, strategy, revenue function, wherein participant is massage set P={PR _i, actionable space A={Admit, Reject}, represent that message is accepted or refuses, tactful D={a _ithe action sequence taked by all message forms, revenue function U comprises two parts, is the revenue function of single message respectively and receives game integral benefit function, U={u _i, Σ ui};

2) the receiving revenue function of participant is fu-fu _min+ c+p-f, refusal revenue function is fu _min-fu, fu _iparticipant PR _imessage forward value of utility, fu _minthe minimum message forwarding effectiveness of all participants, c _ithat node is to receiving PR _ithe payoff of encoded packets, p _ireceive PR _ithe newly-increased potential value of encoded packets, f _ithat node is to receiving PR _ithe penalty value of encoded packets;

3) message receives game message to be forwarded the extended order of sequence as game theory of effectiveness.After forerunner's participant of participant makes decisions, this participant calculates the revenue function that in actionable space, all decision-makings are corresponding, namely income and refusal income is received, game theory is expanded one deck downwards, follow-up participant by that analogy, until game theory arrives destination node, the now income of destination node record from root to leaf, namely each participant indicates the income of making decision by this path;

4) message is asked to receive the sub-game refined Nash equilibrium solution of game, adopt reverse induction, bottom-uply cut off the less branch of income in each participant all action branch, leave the branch of Income Maximum, until reach the root node of game theory, branch composition payoff, determines to receive according to branch and still refuses certain message, finally all message composition coded message set P be accepted.

2. the system as claimed in claim 1, is characterized in that, calculates message when forwarding effectiveness, can make utility function be: $U_i=\frac{r_{\mathrm{neighbor}\left(i\right)}}{r_{\mathrm{neighbor}}}(1-\frac{m_i\left(T_i\right)}{L-1}){\mathrm{\λR}}_i\exp ({-\mathrm{\λn}}_i\left(T_i\right)R_i-{\mathrm{pri}}_i)$

3. the system as claimed in claim 1, is characterized in that, node utilizes a transmission opportunity transmission message to carry out a message exactly and receives gambling process, completes choosing coded message set.

4. the system as claimed in claim 1, is characterized in that, message coding function is decoding functions is wherein g ₁..., g _nvalue is 1, p is encoded packets, p ₁..., p _nfor uncoded message.

5. a computer program, it realizes the system as claimed in claim 1.