CN105072031A - Delay-tolerant disruption-tolerant network routing method based on link transmission capacity - Google Patents

Abstract

A DTN network routing method based on link transmission capacity is used for improving the message delivery rate, and uses link transmission capacity calculation and message priority as a core method, and solves the problem of message routing in the DTN network. The method comprises: defining the concept of a virtual link to represent links among nodes in the DTN network; defining the capacity to represent transmission capacity of the virtual link, and designing a computing method of the capacity, a storage mechanism and a transmission updating method; defining forwarding capacity of the message on the basis of capacity definition, and designing drop precedence of the message and forwarding precedence of a (message, node) 2-tuple. When message forwarding is performed, the 2-tuple with high precedence is selected to forward according to the precedence of the (message, node) 2-tuple; and when buffer application is performed, the drop of the message is determined according to the drop precedence of the message, and the message with high drop precedence is selected to drop. By using a rational information selecting mechanism, the method of the invention improves the delivery rate of the messages in the DTN network

Description

A kind of appearance based on link transmission ability holds circuit network method for routing late

Technical field

The present invention relates to the Message routing field that appearance holds circuit network (Delay-andDisrupted-TolerantNetwork, DTN) late, particularly relate to a kind of DTN internet message method for routing based on link transmission ability.

Background technology

Be different from legacy network, DTN network is usually operated in constrained environment.Which dictates that and comprise a lot of characteristics that DTN network has legacy network and do not possess: intermittent connections, high latency, chance are transmitted etc.Traditional routing method all cannot be applicable to this network environment, therefore needs to re-start research and design to the routing algorithm in DTN network.

Up to the present, researcher has proposed many DTN network route methods, can be divided into following three classes according to routing strategy:

(1) based on the method for routing of contact probability:

Based on the method for routing of contact probability, according to internodal contact historical information, predict the probability that two nodes will carry out contacting sometime in future again.Message, when carrying out message and forwarding, is transmitted to the node that the likelihood ratio oneself of those contact target nodes is large by node, thus improves the probability that message is forwarded to destination node.The existing method for routing based on probability comprises: ProphetRouter, MaxPropRouter and DelegationRouter etc.

(2) based on the method for routing of node relationships

Based on the method for routing of node relationships, according to internodal motion model, contact model etc., predict internodal relation, then forward according to internodal relation.Being mainly used in mobile social networking at present, by predicting the social property of node and social relationships, message being forwarded to the node in close relations with destination node, thus reach and improve message and be forwarded to the probability that target receives you.The existing method for routing based on node relationships comprises: SocialFeature-basedMulti-pathRouting and AnOptimalProbabilisticForwardingProtocol etc.

(3) based on the method for routing of node density

Based on the method for routing of node density, according to the density of local nodes in DTN network, node region higher for local density is organized into one bunch, in then the repeating process of message being converted into bunch and bunch between two parts.This type of method for routing, general by choosing a Leader node for each bunch, responsible message bunch in and bunch between forwarding.When having message to forward in bunch, first message is transmitted to Leader node, is then responsible for forwarding by Leader node.The existing method for routing based on node density comprises: CAMR, EIMD etc.

Various method for routing has its advantage and applicable scene above, but in DTN network, they are not perfect method for routing.And these method for routing are all the selections carrying out route from the angle of node, do not consider the directive function of link in network to route, are not very sufficient to the utilization of the network information.After abstract definition is carried out to the link between DTN nodes, the network information can be utilized more fully, thus improve the efficiency of network route.

Summary of the invention

The present invention, to improve internet message delivery rate for target, with link transmission ability and message prioritization for core concept, solves the Message routing problem in DTN network.Specifically comprise:

1., in order to describe the transmission link in DTN network, propose the concept of " virtual link ".The concept of link " capacity " is defined, represents and distinguish the transmittability of virtual link.The computational methods of definition capacity, storage mode and propagation and update method.

2. define the concept of message forwarding capacity, storage mode and storage organization.The loss priority of definition message and forwarding priority, and on this basis, devise the routing forwarding strategy based on link transmission ability.

Compared with prior art, innovation of the present invention is: with the priority of DTN network link transmission ability and message for core concept, improves DTN internet message delivery rate.Be embodied in:

1. the definition of virtual link embodies the characteristic of " Route Selection and next-hop node are selected " in DTN network on the one hand, also embodies the transmission link asymmetry of DTN network on the other hand.The link transmission ability in DTN network in virtual link and non-DTN network has been distinguished in the definition of capacity, to reflect in DTN network transmittability end to end.

2. the definition of forwarding capacity embodies the current delivery status of message, and the message loss priority defined on this basis embodies the possibility that message is dropped when carrying out buffering area application.When the forwarding priority of message is considered to set up multiple link, how to carry out the selection of message and link.On the basis of message priority definition, the forwarding decision process of route is then converted into the process of message and link selection.

Accompanying drawing explanation

Fig. 1 virtual link and capacity describe

Fig. 2 memory configuration

Fig. 3 forwarding capacity data structure

Embodiment

Consult Fig. 1.Fig. 1 describes the DTN network comprising two virtual links.In DTN network, the forwarding of message carries out decision-making according to next-hop node, and next-hop node is higher to the transmittability of destination node, also just means that message or may transfer to destination node more sooner.Source node is only concerned about how soon can next-hop node or be forwarded to destination node by message, and be indifferent to next-hop node and how message be forwarded to destination node, therefore the path of these three kinds of node compositions of source node, next-hop node and destination node can be regarded as one " virtual link ", source node selects next-hop node according to the transmittability of this " virtual link ".

Consider the virtual link that node A (source node), Node B (next-hop node) and node C (destination node) form.Adopt dotted line to connect between figure interior joint, be due to node between connection be not lasting, dotted line connects and represents two nodes and set up in the past and connect or will connect in the future.This is also the reason why link being called " virtual ", namely there is not continuous link in DTN network, and link is sudden and of short duration.In addition, there is a slice " cloud " in the connection between Node B and node D, this is because source node is indifferent to next-hop node and how message is forwarded to destination node, only needs to know and message can be forwarded to destination node.

Use vl (A, B, C) to represent a virtual link in the present invention, its source node is A, and next-hop node is B, and destination node is C.The existence of virtual link is with good conditionsi, and first definition can reach node and the Node Concepts that can go directly below:

● can node be reached: if message can be forwarded to node Y by nodes X, so node Y is reached at the node of nodes X.The definition that can reach node is asymmetrical.Even node Y is reached at the node of nodes X, does not illustrate that nodes X is also reached at the node of node Y.

● can go directly node: if nodes X directly can contact with node Y, so node Y is the gone directly node of nodes X.The definition of node of can going directly is symmetrical.Namely, if node Y is the gone directly node of nodes X, so nodes X is also the gone directly node of node Y.Understand intuitively, after nodes X and node Y meet, being that nodes X directly contacts with node Y, is also that node Y directly contacts with nodes X.In addition, if node Y is the gone directly node of nodes X, so node Y must be reached at the node of nodes X.Otherwise it is quite different.

Afterwards, we can introduce the necessary and sufficient condition that virtual link exists, that is: the gone directly node of the necessary and sufficient condition that virtual link vl (A, B, C) exists to be Node B be node A and node C are reached at the nodes of Node B.Necessary and sufficient condition can be known thus, and the existence of virtual link is also asymmetrical.When virtual link vl (A, B, C) exists time, we do not know whether vl (C, B, A) exists.In addition, there is the virtual link that a class is special, i.e. vl (A, B, B).Vl (A, B, B) represents a virtual link be only made up of two nodes, and on this virtual link when forwarding messages, message is directly forwarded to destination node by source node.Existence for vl (A, B, B) is identical with the requirement of different virtual link, and Node B must be the gone directly node of node A, and obviously Node B is reached at the node of oneself.

In order to carry out the forwarding of message according to " virtual link ", need to define the transmittability of virtual link.The transmittability of definition " virtual link " is " capacity ".Nodes X to the capacity definition of node Y is: the data volume that can transfer to node Y in the unit interval from nodes X, represents with C (X, Y).As can be seen from the definition of capacity, capacity is oriented, and namely nodes X is different to capacity and the node Y of node Y to the capacity of nodes X, i.e. C (X, Y) ≠ C (Y, X), this meets the asymmetry of DTN network capacity.If represent network with graph structure, so cable network is exactly a non-directed graph, and DTN network is then a directed graph.C (A, B, C) is adopted to represent the capacity of virtual link vl (A, B, C), (vl (A is supposed by the known C of the asymmetry of virtual link (A, B, C) ≠ C (C, B, A), B, C) all exist with vl (C, B, A)).

In order to calculate the computational methods of derivation capacity, first consider the situation of wall scroll virtual link.Node A to node D exists one " virtual link "---vl (A, B, D).Consider that node A needs the message of forwarding 1 byte to node D, so node A needs message to be transmitted to Node B, and then is forwarded to node D by Node B.According to the definition of capacity, node A is forwarded to Node B required time and is node B forwards this message so node A forwards 1 byte message and to node C required time is, thus the capacity that can obtain " virtual link " vl (A, B, D) is:

$C(A,B,D)=\frac{1}{\frac{1}{C(A,B,B)}+\frac{1}{C(B,D)}}=\frac{C(A,B,B)\×C(B,D)}{C(A,B,B)+C(B,D)}$

Consider the situation of two virtual links again, be then generalized to the situation of arbitrary finite bar " virtual link ".Still consider that node A needs the message of transmission 1 byte, the time so transferring to node D by " virtual link " vl (A, B, D) is the time being transferred to node D by " virtual link " vl (A, C, D) is but node A is different to the capacity of Node B and node C, namely C (A, B, B) and C (A, C, C) are different.In practical application scene, the path that node is more prone to selection capacity larger forwards, and therefore can use with the secondary weights representing vl (A, B, D) and vl (A, B, D), so node A transmits 1 byte and to average time of node D is:

$\frac{C(A,B,B)}{C(A,B,B)+C(A,C,C)}\×\frac{1}{C(A,B,D)}+\frac{C(A,C,C)}{C(A,B,B)+C(A,C,C)}\×\frac{1}{C(A,C,D)}$

Thus the capacity obtaining egress A to node D is:

$C(A,D)=\frac{1}{\frac{C(A,B,B)}{C(A,B,B)+C(A,C,C)}\×\frac{1}{C(A,B,D)}+\frac{C(A,C,C)}{C(A,B,B)+C(A,C,C)}\×\frac{1}{C(A,C,D)}}$

Be generalized to ordinary circumstance, suppose that nodes X can arrive node Y, wherein i-th node Z that can go directly by n the node that can go directly _ithat representative, the capacity so between nodes X to node Y is:

$C(X,Y)=\frac{1}{{\mathrm{\Σ}}_{i=1}^n\frac{C(X,Z_i,Z_i)}{{\mathrm{\Σ}}_{j=1}^{n}C(X,Z_j,Z_j)}\×\frac{1}{C(X,Z_i,Y)}}$

$\begin{array}{c}=\frac{{\Σ}_{j=1}^{n}C(X,Z_j,Z_j)}{{\Σ}_{i=1}^{n}C(X,Z_i,Z_i)\×\frac{C(X,Z_i,Z_i)+C(Z_i,Y)}{C(X,Z_i,Z_i)C(Z_i,Y)}}\\ =\frac{{\Σ}_{j=1}^{n}C(X,Z_j,Z_j)}{n+{\Σ}_{i=1}^n\frac{C(X,Z_{i,},Z_i)}{C(Z_i,Y)}}\end{array}$

Consult Fig. 2.Fig. 2 describes the storage organization of capacity, comprises two list structures: capacities chart (NodeCapacityTable, NCT) and virtual link capacities chart (VirtualLinkCapacityTable, VLCT).Y in node capacity table _irepresent the destination node of this record, C (X, Y _i) represent source node X to destination node Y _icapacity.In virtual link capacities chart, Z _irepresenting the destination node of this record, is the node that can go directly of nodes X, C (X, Z _i, Z _i) represent link vl (X, Z _i, Z _i) capacity, bt _ithe settling time of representative record.Without any record in showing time initial, when running into a node, node sets up record for this reason.Therefore, nt _ialso representation node X first time meets node Z _itime.Total _iup to the present, nodes X can to node Z in representative _ithe data volume that can transmit.Here can transmit, be not that representation node X is to node Z _ithe data volume transmitted, but representation node X can be transferred to node Z by directly contacting _idata volume.Such as, nodes X and node Z _ibetween the transmission rate that connects be v, link time is t, so this time link transmitted data amount can be v × t.

When nodes X just adds network, nodes X is known nothing for network.Therefore, be empty time the node capacity table of nodes X is initial.As time goes on, nodes X constantly meets other nodes, collecting network information, then uses these information to instruct message possible in the future to forward (may be the message that oneself creates, also may be other node forwarding messages of side).In the process of node collecting network information and collection network other nodes capacity information and upgrade the process of self capacity information.

When two nodes, when meeting, they need to exchange its respective capacity information, i.e. node capacity table, the then information of each self refresh own node capacities chart---for nodes X and node Y---.It is noted herein that after nodes X and node Y switching node capacities chart, nodes X upgrades NCT _x, so NCT of receiving of node Y _xit is out-of-date just to become.But now need not exchange capacity again, because its renewal just increased after node Y upgraded.Therefore, time the exchange of node capacity table only occurs in link foundation.After nodes X and node Y have exchanged node capacity table, nodes X needs the node capacity table upgrading self, but its renewal does not relate to whole node capacity table, only relates to reached at the node of all node Y.

Except node meet time, when connecting disconnection between node, also need to recalculate capacity, but the capacity now calculated only relates to the node disconnected.After nodes X and node Y disconnect, nodes X needs first to recalculate C (X, Y, Y), and then recalculates C (X, Y).The former account form is as follows:

$C(X,Z_i,Z_i)=\frac{{\mathrm{total}}_i+sp\×duration}{ct-{\mathrm{bt}}_i}$

Wherein, ct represents current time, the transmission rate of sp representative this time link, the duration that duration representative this time connects.In addition, also need to carry out periodic refreshing to node capacity table.Namely, each a period of time refreshes node capacity table, thus reduces the property delayed of node capacity table.The method for refreshing of node capacity table is fairly simple, and its computing formula is as follows:

$C(X,Z_i,Z_i)=\frac{{\mathrm{total}}_i}{ct-{\mathrm{bt}}_i}$

In order to sort to message and the forwarding of guide message, we define the forwarding capacity (relaycapacity, rc) of message on the basis of capacity.The forwarding capacity of a message is defined as " count from this message creation time, this message has been forwarded to the data volume of destination node ".Certainly, we can not the forwarding capacity of Obtaining Accurate message, and therefore forwarding capacity is here an estimated value, is determined by the forwarding situation of message.

Consult Fig. 3.Fig. 3 message forwarding capacity store data structure---MessageInfo (MI).The MI of message is made up of three parts, comprising: forwarding capacity (rc), capacity acceleration (acceleration) and message forwarding capacity recent update time (ut) that message is current.When message establishing time, message establishing MI for this reason, and the forwarding capacity of this message and capacity acceleration are initialized as 0, the last update set of time of this message forwarding capacity is current time.Once message forwards, just need to recalculate the capacity acceleration of message, computational methods are as follows:

MI _m.acceleration＝MI _m.accleration+C(Y，des _m)

Capacity acceleration by message m increases the capacity between node Y to m destination node.And the forwarding capacity of message is undertaken calculating by the capacity acceleration of message, computational methods are as follows:

MI _m.rc＝MI _m.rc+MI _m.acceleration*(ct-MI _m.ut)

Wherein, ct represents current time.Intuitively understand, namely the increase of the forwarding capacity of message was updated to the capacity of current time increase from last time.

On the basis of message forwarding capacity, the priority of message can be defined, comprising: loss priority and forwarding priority.Loss priority defines when node messages buffering area is full, and which message should be selected to abandon.The loss priority of message is directly determined by the forwarding capacity of message, and a message forwarding capacity is larger, then describing message, to be transferred to the data volume of destination node more.Therefore, when buffering area is full, should preferentially abandon this message.Use p _mrepresent the loss priority of message m, its computational methods are:

p _m＝1/MI _m.rc

The p of a message _mbe worth less, represent its priority higher, be also more likely dropped with regard to this message.

The forwarding priority of message determines when connecting, and should preferentially forward which message.In DTN network, a node, at synchronization, may set up multiple connection with multiple node.Therefore, when forwarding messages, not only to consider message, also will consider to connect.That is, repeating process is not only the process selecting message, is also the process selecting to connect.In order to address this problem, define the priority of (message, node) two tuples, then select a transmission of messages to specific node according to the priority of this two tuple.We use p _{t (m, Y)}represent two tuple (m, Y) priority, its computational methods are as follows:

$p_{t\left(mY\right)}=\frac{C(Y,{\mathrm{des}}_m)}{{\mathrm{MI}}_m.acceleration}$

Namely two tuples (m, Y) are determined divided by the capacity acceleration of message to the capacity of message m destination node by node Y, understand the capacity acceleration ratio namely m being forwarded to node Y and increasing intuitively.The p of two tuples _{t (m, Y)}be worth larger, then this two tuple of prioritizing selection forwards.In addition, MI is worked as _mwhen being .acceleration 0, p _{t (m, Y)}be set to C (Y, des _m).

On the basis of message priority, the forwarding strategy of message can be defined, comprise the following steps:

(1) MI of all message in buffering area is calculated.

(2) all two tuples (message connects) are set up.

(3) priority of all two tuples is calculated, and according to priority size sequence.

(4) if there is more message to transmit, to select and two tuples (m, Y) removing priority the highest are transmitted.Otherwise, go to 7.

(5) attempt message m to be forwarded to node Y, if node Y accepts message, then start this transmission; Otherwise, go to 4.

(6) when message m is transmitted, MI is upgraded _mand by MI _msend to node Y, go to 4.If disconnect with node Y and cause Transmission, then directly go to 4.

(7) do not have message to transmit, wait for new connection or new message.

As step 5 describes, when message m is transferred to node Y, node Y needs to determine whether accept message m.Node Y determines whether accepting this message according to many factors.Node Y first first determines that oneself is not at message transfer, then needs to determine do not have this message in buffering area.Time condition when above all meets, next node Y just only needs to do last preparation for accepting this message---application space.If node Y has enough free spaces to carry out storing message m, so then accept this message.Otherwise, node Y needs to determine that whether carrying out message abandons, thus vacating space carrys out storing message m.In time having sufficient space to accept message m, then accept this message; Otherwise refusal accepts this message.

Claims (7)

1., based on a DTN network route method for link transmission ability, comprise the definition of DTN network link transmission ability and message priority forwarding strategy, it is characterized in that:

1) concept defining " virtual link " represents and distinguishes the link between DTN nodes, and virtual link is made up of three nodes: source node, next-hop node and destination node, has asymmetric feature;

2) use " capacity " to define the transmittability of " virtual link ", and define the calculating of capacity, storage, propagation and update method, there is asymmetric feature;

3) use " forwarding capacity " to define the delivery quantity of message, and devise the data structure of message forwarding capacity;

4) on forwarding capacity basis, define the loss priority of message and the forwarding priority of (message, node) two tuples, the loss priority of message is larger, is more likely dropped; (message, node) two tuple forwarding priority define and preferably select which message to be forwarded to which node, the forwarding priority of message is larger, more understands preferentially being forwarded;

5) Effect-based operation loss priority and (message, node) two routing forwarding strategy of tuple forwarding priority is devised.

2. method according to claim 1, is characterized in that: nodes X is as follows to the calculation of capacity formula of node Y:

$C(X,Y)=\frac{{\mathrm{\Σ}}_{i=1}^{n}C(X,Z_i,Z_i)}{n+{\mathrm{\Σ}}_{i=1}^n\frac{C(X,Z_{i,},Z_i)}{C(Z_i,Y)}}$

3. method according to claim 1, is characterized in that: capacity adopts node capacity table and virtual link capacities chart to store.Two fields that node capacity table comprises (can reach node, capacity).Five fields that virtual link capacities chart comprises (can go directly node, capacity, records settling time, total amount of data, node capacity table).

4. method according to claim 1, is characterized in that: virtual link vl (X, Z _i, Z _i) calculation of capacity formula as follows:

$C(X,Z_i,Z_i)=\frac{{\mathrm{total}}_i+sp\×duration}{ct-{\mathrm{bt}}_i}$

Virtual link vl (X, Z _i, Z _i) capacity refresh formula as follows:

$C(X,Z_i,Z_i)=\frac{{\mathrm{total}}_i}{ct-{\mathrm{bt}}_i}$

5. method according to claim 1, is characterized in that: the forwarding capacity of message defines to be counted from message creation time, and this message has been forwarded to the data volume of destination node.Its store data structure comprises three fields: forwarding capacity (rc), capacity acceleration (acceleration) and message forwarding capacity recent update time (ut) that message is current.It is as follows that it calculates update method:

MI _m.acceleration＝MI _m.accleration+C(Y,des _m)

MI _m.rc＝MI _m.rc+MI _m.acceleration*(ct-MI _m.ut)

6. method according to claim 1, is characterized in that: the computational methods of the loss priority of message are as follows:

p _m＝1/MI _m.rc

7. method according to claim 1, is characterized in that: the computational methods of (message, node) forwarding priority are as follows:

$p_{t(m,Y)}=\frac{C(Y,{\mathrm{des}}_m)}{{\mathrm{MI}}_m.acceleration}$