CN105228196A - A kind of delay-tolerant network nodal cache management method being applicable to the Effect-based operation weight of many copies route - Google Patents

Abstract

The invention discloses a kind of delay-tolerant network nodal cache management method being applicable to the Effect-based operation weight of many copies route.For arbitrary node in network, if there is message to need to receive.Judge whether the vacant buffer memory of this node has sufficient space to receive this message, if had: receive this message also carries out message weight again division to the message in node.If do not had: judge that whether this node is the destination node of this message, if destination node, whether the cache size of decision node current residual and all high weight message buffering size sums are greater than the size of this message, if be greater than the message selected in high weight message and need to delete, until there is sufficient space to receive this message.If be not more than, high weight message is all deleted, select in low weight message the message needing to delete, until there is sufficient space to receive this message.The present invention can effectively improve the nodal cache efficiency of management.

Description

A kind of delay-tolerant network nodal cache management method being applicable to the Effect-based operation weight of many copies route

Technical field

The invention belongs to delay-tolerant network field, particularly relate to a kind of delay-tolerant network nodal cache management method being applicable to the Effect-based operation weight of many copies route.

Background technology

Delay-tolerant network (DTN) is a hot issue of sensor network research field in recent years.It has a wide range of applications, and comprises planetary networks, rural network, Military Network, AdHoc network etc.Delay-tolerant network is the new network that a kind of and legacy network has significantly difference, has the features such as unreliable link, high latency, dynamic topology, node resource be limited.Message can experience longer time delay and can not ensure necessarily there is an end-to-end fullpath in transmittance process, in order to overcome the restriction of these networks, usually many copies Routing Protocol is adopted in delay-tolerant network, many copies Routing Protocol can improve the arrival rate of transmission of messages and reduce transmission of news time delay, but the memory space that many copies Routing Protocol makes node limited has very easily been consumed, cause cache overflow phenomenon.In delay-tolerant network, the main task of cache management is exactly when cache overflow, determines which message will be dropped.Therefore, a kind of efficient cache management strategy seems particularly important for the performance improving network.

The domestic and international achievement in research for delay-tolerant network interior joint cache management is relatively less at present, and some basic buffer queue management strategies mainly contain: (1) Drop-Front (DF): first abandon the message that in nodal cache district, queuing time is the longest; (2) Drop-Oldest (DO): first abandon in nodal cache district and remain lifetime minimum message; (3) Drop-Last (DL): first abandon the up-to-date message be received in nodal cache district; (4) Drop-Youngest (DY): abandon the message having the longest residue life span in nodal cache district.

Several buffer queue management strategy is fairly simple above, in document ReactiveWeightBasedBufferManagementPolicyforDTNRoutingPr otocols, the people such as RashidS propose a kind of delay-tolerant network nodal cache management strategy of Effect-based operation weight.Its main thought is: according to the weight of message in node, message in node is divided into high weight message and low weight message, when nodal cache space is overflowed, if present node is not the destination node of the new message received, then only by deleting the message release spatial cache of high weight, receive new message, throw away when enough spatial caches can not be had to receive this message if delete all high weight message, then abandon receiving this message.If present node is the destination node of the new message received, nodal cache remaining space and the spatial cache sum shared by high weight message throw away the size being less than this message, discharge spatial cache until there is sufficient space to receive this message by deleting the message that in low weight message, TTL is minimum.The method can effectively manage nodal cache; protect the message that in node, some weight are lower, but when deleting message, only using the foundation of the size of TTL as deletion message; easily the message being about to arrive destination node is deleted, there is certain limitation in this way.

In document Bufferschemeoptimizationofepidemicroutingindelaytolerant networks, ShenJ, a kind of cache management strategy being applicable to epidemic Routing Protocol is proposed Deng people, its main thought is: when nodal cache overflows, reach maximum to meet the average delivery rate of message in network, consider from multiple attributes of message, ask the method be worth most to decide to delete which message by Lagrange.But the size of message is not considered when determining to delete message, this just makes to delete multiple little message to receive a very large message, is easy to the performance having influence on network.

Summary of the invention

The object of this invention is to provide a kind of nodal cache efficiency of management that can effectively improve, be applicable to the delay-tolerant network nodal cache management method of the Effect-based operation weight of many copies route.

Be applicable to a delay-tolerant network nodal cache management method for the Effect-based operation weight of many copies route, comprise the following steps,

Step one: by DTN weight division methods, weight division is carried out to the message in node, the message after division is divided into high weight message and low weight message;

Step 2: arbitrary node in network, when needing when there being message to receive, judges whether the vacant buffer memory of this node has enough free spaces to receive this message, if so, then performs step 7; Otherwise, perform next step;

Step 3: judge that whether present node is the destination node of this message, if so, perform next step, otherwise perform step 8; Step 4: whether the free buffer size of decision node current residual and all high weight message buffering size sums are greater than the size of this message, if so, then perform next step, otherwise perform step 6;

Step 5: selected in high weight message the message needing to delete by the system of selection of DTN message, until there is sufficient space to receive this message, turn to step 7;

Step 6: high weight message all deleted, is selected in low weight message the message needing to delete, until there is sufficient space to receive this message, turns to step 7 by the system of selection of DTN message;

Step 7: receive this message and return step one;

Step 8: whether the free buffer size of node current residual and all high weight message buffering size sums are greater than the size of this message, if so, then perform step 5, otherwise perform step 9;

Step 9: abandon receiving, and ignore this message.

A kind of delay-tolerant network nodal cache management method being applicable to the Effect-based operation weight of many copies route of the present invention, can also comprise:

1, DTN weight division methods is: in node, weight is greater than the message of average weight is high weight message, and in node, weight is less than the message of average weight is low weight message;

Wherein, the average weight of message in each node:

$MAoVWeight=\frac{1}{I\left(t\right)}\underset{i=1}{\overset{I\left(t\right)}{\Σ}}MWeight\left(i\right)$

The total weight of all message in node:

$MToTWeight=\underset{i=1}{\overset{I\left(t\right)}{\Σ}}MWeight\left(i\right)$

The wherein sum of message in I (t) representation node;

In node each message weight be:

$MWeight\left(i\right)=\frac{1}{R_i}+\frac{1}{{\mathrm{MSTQ}}_i}+{\mathrm{MS}}_i+n_i\left(T_i\right)+m_i\left(T_i\right)$

Wherein T _irepresent the time that message i has experienced in a network, R _irepresent the surplus value of the life cycle of message i, MSTQ _irepresent the time that message i stops at message queue, MS _irepresent message i, n _i(T _i) represent at experience T _ithe number of copies that in time network, current message i produces, m _i(T _i) represent at experience T _ithe node number that time message i meets.

2, the system of selection of DTN message is, the message i selected in node meets the following conditions: message i's be worth the minimum value in all message in node for this reason;

Wherein L represents the node number in network, the phase encountering rate between λ nodes.

Beneficial effect:

For the merits and demerits had in existing nodal cache management strategy, the present invention proposes a kind of delay-tolerant network nodal cache management strategy being applicable to the Effect-based operation weight of many copies route.Its main thought: first according to the weight of message in node, the message in node is divided into high weight message and low weight message, and wherein the weight of message i refers to:

$MWeight\left(i\right)=\frac{1}{R_i}+\frac{1}{{\mathrm{MSTQ}}_i}+{\mathrm{MS}}_i+n_i\left(T_i\right)+m_i\left(T_i\right)$

Wherein, T _irepresent the size of the time that message i has experienced in a network, R _irepresent the size of the surplus value of the life cycle of message i, MSTQ _irepresent the size of message i in the time that message queue stops, MS _irepresent the size of message i, n _i(T _i) represent at experience T _ithe size of the number of copies that current message i produces in time network, m _i(T _i) represent at experience T _ithe node number that time message i meets.Be greater than the message of the average weight of message in node in high weight message dactylus point, be less than the message of the average weight of message in node in low weight message dactylus point, wherein average weight refers to that the weight summation of all message in node is in the number divided by message.When nodal cache space is overflowed, if present node is not the destination node of the new message received, reach maximum to meet the average delivery rate of message in network, then multiple attributes of roundup, ask the method be worth most to decide which message of deleting in high weight message by Lagrange and discharge spatial cache, until have sufficient space to receive new message, if all high weight message and node residue free buffer space are less than the size of current message, then abandon receiving this message.If the destination node of the new message received is present node, and the spatial cache residue size of free buffer and the spatial cache sum shared by high weight message are still less than the size of this message, then delete all high weight message, and ask the method be worth most to consider the attribute of multiple message by Lagrange, decide the message of deleting which low weight, straight way has enough release spatial caches to receive this message.

The advantage that the present invention has: (1), by message in node is divided into high weight message and low weight message, when nodal cache overflows, can effectively protect low weight message blindly not deleted.(2) reaching maximum to meet the average delivery rate of message in network, by multiple attributes of roundup, and determining to delete which message by the method for drawing erlang day to minimize, effectively can improve the efficiency of nodal cache management.

Accompanying drawing explanation

A kind of flow chart being applicable to the delay-tolerant network nodal cache management strategy of the Effect-based operation weight of many copies route of Fig. 1;

A kind of frame diagram being applicable to the delay-tolerant network nodal cache management strategy of the Effect-based operation weight of many copies route of Fig. 2.

Embodiment

Below in conjunction with accompanying drawing, the present invention is described in further details.

Main purpose of the present invention proposes a kind of delay-tolerant network nodal cache strategy being applicable to the Effect-based operation weight of many copies route.The main thought of this strategy is: first according to the weight of message in node, and the message in node is divided into high weight message and low weight message, and wherein the weight of message i refers to:

$MWeight\left(i\right)=\frac{1}{R_i}+\frac{1}{{\mathrm{MSTQ}}_i}+{\mathrm{MS}}_i+n_i\left(T_i\right)+m_i\left(T_i\right)$

Wherein, T _irepresent the size of the time that message i has experienced in a network, R _irepresent the size of the surplus value of the life cycle of message i, MSTQ _irepresent the size of message i in the time that message queue stops, MS _irepresent the size of message i, n _i(T _i) represent at experience T _ithe size of the number of copies that current message i produces in time network, m _i(T _i) represent at experience T _ithe node number that time message i meets.Be greater than the message of the average weight of message in node in high weight message dactylus point, be less than the message of the average weight of message in node in low weight message dactylus point, wherein average weight refers to that the weight summation of all message in node is in the number divided by message.When nodal cache space is overflowed, if present node is not the destination node of the new message received, then multiple attributes of roundup, ask the method be worth most to decide which message of deleting in high weight message by Lagrange and discharge spatial cache, until have sufficient space to receive new message, if all high weight message and node residue free buffer space are less than the size of current message, then abandon receiving this message.If the destination node of the new message received is present node, and the spatial cache residue size of free buffer and the spatial cache sum shared by high weight message still can not meet the size of this message, then delete all high weight message, and ask the method be worth most to consider the attribute of multiple message by Lagrange, decide the message of deleting which low weight, straight way has enough release spatial caches to receive this message.

The present invention is achieved in that as shown in Figure 1:

Step one: by DTN weight division methods, carry out weight division to the message in node, the message after division is divided into high weight message and low weight message.

Step 2: for arbitrary node in network, when needing when there being message to receive.

Step 3: judge whether the vacant buffer memory of this node has enough free spaces to receive this message, if so, then performs step 8; Otherwise, perform step 4.

Step 4: judge that whether present node is the destination node of this message, if so, perform step 5.Otherwise execution step 9.

Step 5: whether the free buffer size of node current residual and all high weight message buffering size sums are greater than the size of this message, if so, then perform step 6, otherwise perform step 7.

Step 6: selected in high weight message the message needing to delete by the system of selection of DTN message, until there is sufficient space to receive this message, turn to step 8.

Step 7: high weight message all deleted, is selected in low weight message the message needing to delete, until there is sufficient space to receive this message, turns to step 8 by the system of selection of DTN message.

Step 8: receive this message and return step one.

Step 9: whether the free buffer size of node current residual and all high weight message buffering size sums are greater than the size of this message, if so, then perform step 6, otherwise perform step 10.

Step 10: abandon receiving, and ignore this message.

The DTN weight division methods mentioned in step one also comprises: for any one the message i in node, the weight of message i: $MWeight\left(i\right)=\frac{1}{R_i}+\frac{1}{{\mathrm{MSTQ}}_i}+{\mathrm{MS}}_i+n_i\left(T_i\right)+m_i\left(T_i\right),$ Wherein T _irepresent the size of the time that message i has experienced in a network, R _irepresent the size of the surplus value of the life cycle of message i, MSTQ _irepresent the size of message i in the time that message queue stops, MS _irepresent the size of message i, n _i(T _i) represent at experience T _ithe size of the number of copies that current message i produces in time network, m _i(T _i) represent at experience T _ithe node number that time message i meets.Determine each message in node weight after, obtain the total weight of all message in node: the wherein sum of message in I (t) representation node.The average weight of message in each node is obtained after determining the total weight of message: the message making weight in node be greater than average weight is high weight message, and the message that weight is less than average weight is low weight message.

The DTN message system of selection mentioned in step 6 and step 7 also comprises: in order to the average delivery rate meeting message reaches maximum, and message i selected in node should meet following condition: message i's be worth the minimum value in all message in node for this reason, will be designated as formula (1), n _i(T _i)≤m _i(T _i)+1, n _i(T _i)≤L, m _i(T _i)≤L-1, T _i≤ TTL _i, wherein: T _irepresent the size of the time that message i has experienced in a network, L represents the size of the node number in network, m _i(T _i) represent at experience T _ithe node number that time message i meets, the phase encountering rate between λ nodes, R _irepresent the size of the surplus value of the TTL of message i, TTL _irepresent the TTL of message i, n _i(T _i) represent at experience T _ithe size of the number of copies that current message i produces in time network.In order to obtain message i, we can be constructed as follows and draw erlang day function:

$\begin{array}{c}L(T_i,\mathrm{\ϵ})=\[(1-\frac{m_i\left(T_i\right)}{L-1}){\mathrm{\λR}}_i{e}^{-{\mathrm{\λn}}_i\left(T_i\right)R_i}\]\\ +{\mathrm{\ϵ}}_1\[n_i\left(T_i\right)-m_i\left(T_i\right)-1\]\\ +{\mathrm{\ϵ}}_2\[n_i\left(T_i\right)-L\]\\ +{\mathrm{\ϵ}}_3\[m_i\left(T_i\right)-L+1\]\\ +{\mathrm{\ϵ}}_4\[T_i-{\mathrm{TTL}}_i\]\end{array}---\left(2\right)$

Wherein, ε _xlagrange coefficient, R _i=TTL _i-T _i, wherein TTL _irepresent the TTL of message i, Lagrange duality function can be drawn by formula (2):

$\begin{array}{c}D\left(\mathrm{\ϵ}\right)=\inf T_i\[L(T_i,\mathrm{\ϵ})\]\\ =\inf T_i\{\[(1-\frac{m_i\left(T_i\right)}{L-1}){\mathrm{\λR}}_i{e}^{-{\mathrm{\λn}}_i\left(T_i\right)R_i}\]\\ +{\mathrm{\ϵ}}_1\[n_i\left(T_i\right)-m_i\left(T_i\right)-1\]\\ +{\mathrm{\ϵ}}_2\[n_i\left(T_i\right)-L\]\\ +{\mathrm{\ϵ}}_3\[m_i\left(T_i\right)-L+1\]\\ +{\mathrm{\ϵ}}_4\[T_i-{\mathrm{TTL}}_i\]\}\end{array}---\left(3\right)$

Next to variable T in formula (2) _idifferentiate, can obtain formula (4) as follows:

$\begin{array}{c}\▿L(T_i,\mathrm{\ϵ})=\[-\frac{\▿m_i\left(T_i\right)}{L-1}{\mathrm{\λR}}_i{e}^{-{\mathrm{\λn}}_i\left(T_i\right)R_i}+(1-\frac{m_i\left(T_i\right)}{L-1})\·\\ (\mathrm{\λ}\▿R_i{e}^{-{\mathrm{\λn}}_i\left(T_i\right)R_i}-{\mathrm{\λ}}^2{R_i}^2{e}^{-{\mathrm{\λn}}_i\left(T_i\right)R_i}\▿n_i(T_i)-\\ {\mathrm{\λ}}^2{R}_i{e}^{-{\mathrm{\λn}}_i\left(T_i\right)R_i}{n}_i\left(T_i\right)\▿R_i)\]+\\ {\mathrm{\ϵ}}_1\[\▿n_i\left(T_i\right)-\▿m_i\left(T_i\right)\]+\\ {\mathrm{\ϵ}}_2\▿n_i\left(T_i\right)+{\mathrm{\ϵ}}_3\▿m_i\left(T_i\right)+{\mathrm{\ϵ}}_4\end{array}---\left(4\right)$

Another formula (4) equals 0; T can be drawn _iand the relation between ε, suppose T _i=g (ε), then uses the T that g (ε) replaces in formula (3) _i, we can obtain the following formula about D (ε):

$\begin{array}{c}D\left(\mathrm{\ϵ}\right)=\[(1-\frac{m_i\left(g\right(\mathrm{\ϵ}\left)\right)}{L-1})\·\mathrm{\λ}({\mathrm{TTL}}_i-g(\mathrm{\ϵ}\left)\right)e^{-{\mathrm{\λn}}_i\left(g\right(\mathrm{\ϵ}\left)\right)({\mathrm{TTL}}_i-g(\mathrm{\ϵ}\left)\right)}\]\\ +{\mathrm{\ϵ}}_1\[n_i\left(g\right(\mathrm{\ϵ}\left)\right)-m_i\left(g\right(\mathrm{\ϵ}\left)\right)-1\]\\ +{\mathrm{\ϵ}}_2\[n_i\left(g\right(\mathrm{\ϵ}\left)\right)-L\]\\ +{\mathrm{\ϵ}}_3\[m_i\left(g\right(\mathrm{\ϵ}\left)\right)-L+1\]\\ +{\mathrm{\ϵ}}_4\[g\left(\mathrm{\ϵ}\right)-{\mathrm{TTL}}_i\]\end{array}---\left(5\right)$

Therefore we can draw the dual problem of equal value with formula (1): ask max:D (ε), wherein ε 1 >=0, ε 2 >=0, ε 3 >=0, ε 4 >=0, first to formula (5) differentiate, draw following formula (6):

$\begin{array}{c}\▿D\left(\mathrm{\ϵ}\right)=\[-\frac{\▿m_i\left(g\right(\mathrm{\ϵ}\left)\right)\▿g\left(\mathrm{\ϵ}\right)}{L-1}\·\mathrm{\λ}({\mathrm{TTL}}_i-g(\mathrm{\ϵ}\left)\right)e^{-{\mathrm{\λn}}_i\left(g\right(\mathrm{\ϵ}\left)\right)({\mathrm{TTL}}_i-g(\mathrm{\ϵ}\left)\right)}\\ +(1-\frac{m_i\left(g\right(\mathrm{\ϵ}\left)\right)}{L-1})\\ \·(-\mathrm{\λ}\▿g(\mathrm{\ϵ}\left)e^{-{\mathrm{\λn}}_i\left(g\right(\mathrm{\ϵ}\left)\right)({\mathrm{TTL}}_i-g(\mathrm{\ϵ}\left)\right)}\right)\\ -{\mathrm{\λ}}^2{({\mathrm{TTL}}_i-g(\mathrm{\ϵ}\left)\right)}^2{e}^{-{\mathrm{\λn}}_i\left(g\right(\mathrm{\ϵ}\left)\right)({\mathrm{TTL}}_i-g(\mathrm{\ϵ}\left)\right)}\▿n_i\left(g\right(\mathrm{\ϵ}\left)\right)\▿g\left(\mathrm{\ϵ}\right)\\ +{\mathrm{\λ}}^2({\mathrm{TTL}}_i-g(\mathrm{\ϵ}\left)\right)e^{-{\mathrm{\λn}}_i\left(g\right(\mathrm{\ϵ}\left)\right)({\mathrm{TTL}}_i-g(\mathrm{\ϵ}\left)\right)}{n}_i\left(g\right(\mathrm{\ϵ}\left)\right)\▿g\left(\mathrm{\ϵ}\right))\]\\ +{\mathrm{\ϵ}}_1\[\▿n_i\left(g\right(\mathrm{\ϵ}\left)\right)\▿g\left(\mathrm{\ϵ}\right)-\▿m_i\left(g\right(\mathrm{\ϵ}\left)\right)\▿g\left(\mathrm{\ϵ}\right)\]\\ +{\mathrm{\ϵ}}_2\▿n_i\left(g\right(\mathrm{\ϵ}\left)\right)+{\mathrm{\ϵ}}_3\▿m_i\left(g\right(\mathrm{\ϵ}\left)\right)\▿g\left(\mathrm{\ϵ}\right)+{\mathrm{\ϵ}}_4\▿g\left(\mathrm{\ϵ}\right)\end{array}---\left(6\right)$

Another formula (6) equal 0 we can draw and the optimal solution of dual problem be designated as ε *, due to T _i=g (ε), so T _i ^*=g (ε ^*), therefore we can draw to delete and meet message i.

Figure 2 shows that example, Fig. 2 is the exemplary plot of nodes buffer memory, wherein in nodal cache, black lines part represents the spatial cache shared by low weight message, the spatial cache of white blank parts shared by high weight message, the unappropriated spatial cache of black mesh portion representation node, black lines arrow LWL represents low weight message queue, and white blank arrowhead HWL represents high weight message queue.

First by DTN weight division methods, carry out weight division to the message in node, the message after division is divided into high weight message and low weight message.For arbitrary node in network, if there is message to need to receive.

Then judge whether the vacant buffer memory of this node has sufficient space to receive this message, if this node has enough un-occupied space to receive this message, then receive this message also carries out message weight again division to the message in node.

If the vacant slow of this node does not have this message of enough space-reception, then judge that whether this node is the destination node of this message, if this node is the destination node of this message, then whether the cache size of decision node current residual and all high weight message buffering size sums are greater than the size of this message, if the cache size of current residual and all high weight message buffering size sums are greater than the size of this message, the message needing to delete is selected in high weight message, until there is sufficient space to receive this message by the system of selection of DTN message.

If the cache size of current residual and all high weight message buffering size sums are not more than the size of this message, high weight message is all deleted, the message needing to delete is selected in low weight message, until there is sufficient space to receive this message by the system of selection of DTN message.

If the vacant slow of this node does not have this message of enough space-reception, and this node is not the destination node of this message, the size sum of the high weight message buffering of vacant mitigation of this node is less than this message, then ignore this message.Otherwise selected in high weight message the message needing to delete by the system of selection of DTN message, until there is sufficient space to receive this message.

Claims (3)

1. be applicable to a delay-tolerant network nodal cache management method for the Effect-based operation weight of many copies route, it is characterized in that: comprise the following steps,

Step one: by DTN weight division methods, weight division is carried out to the message in node, the message after division is divided into high weight message and low weight message;

Step 2: arbitrary node in network, when needing when there being message to receive, judges whether the vacant buffer memory of this node has enough free spaces to receive this message, if so, then performs step 7; Otherwise, perform next step;

Step 3: judge that whether present node is the destination node of this message, if so, perform next step, otherwise perform step 8; Step 4: whether the free buffer size of decision node current residual and all high weight message buffering size sums are greater than the size of this message, if so, then perform next step, otherwise perform step 6;

Step 5: selected in high weight message the message needing to delete by the system of selection of DTN message, until there is sufficient space to receive this message, turn to step 7;

Step 6: high weight message all deleted, is selected in low weight message the message needing to delete, until there is sufficient space to receive this message, turns to step 7 by the system of selection of DTN message;

Step 7: receive this message and return step one;

Step 8: whether the free buffer size of node current residual and all high weight message buffering size sums are greater than the size of this message, if so, then perform step 5, otherwise perform step 9;

Step 9: abandon receiving, and ignore this message.

2. a kind of delay-tolerant network nodal cache management method being applicable to the Effect-based operation weight of many copies route according to claim 1, it is characterized in that: described DTN weight division methods is: in node, weight is greater than the message of average weight is high weight message, in node, weight is less than the message of average weight is low weight message;

Wherein, the average weight of message in each node:

$MAoVWeight=\frac{1}{I\left(t\right)}\underset{i=1}{\overset{I\left(t\right)}{\Σ}}MWeight\left(i\right)$

The total weight of all message in node:

$MToTWeight=\underset{i=1}{\overset{I\left(t\right)}{\Σ}}MWeight\left(i\right)$

The wherein sum of message in I (t) representation node;

In node each message weight be:

$MWeight\left(i\right)=\frac{1}{R_i}+\frac{1}{{\mathrm{MSTQ}}_i}+{\mathrm{MS}}_i+n_i\left(T_i\right)+m_i\left(T_i\right)$

Wherein T _irepresent the time that message i has experienced in a network, R _irepresent the surplus value of the life cycle of message i, MSTQ _irepresent the time that message i stops at message queue, MS _irepresent message i, n _i(T _i) represent at experience T _ithe number of copies that in time network, current message i produces, m _i(T _i) represent at experience T _ithe node number that time message i meets.

3. a kind of delay-tolerant network nodal cache management method being applicable to the Effect-based operation weight of many copies route according to claim 1, it is characterized in that: described DTN message system of selection is that the message i selected in node meets the following conditions: message i's be worth the minimum value in all message in node for this reason;

Wherein L represents the node number in network, the phase encountering rate between λ nodes.