CN102790804A - Intelligent mobile agent-based unstructured P2P network load balancing method and system - Google Patents

Abstract

The invention provides an intelligent mobile agent-based unstructured P2P (peer-to-peer) network load balancing method and corresponding system, which aim to avoid network node congestion and realize load balance, the technical scheme adopted by the invention is that the method comprises the following steps: S1, each node in the network regularly initiates a topology adjustment mechanism and is to be connected with another node with a high processing capacity, so as to form a center type topological structure; and S2, the congestion state of the node is judged according to the congestion degree of the node in a certain moment, so as to further optimize the network topology.

Description

Non-structure peer-to-peer network load-balancing method and system based on intelligent mobile agent

Technical field

The present invention relates to the peer-to-peer network technical field, relate in particular to a kind of non-structure peer-to-peer network load-balancing method and system based on intelligent mobile agent.

Background technology

In recent years, (Peer-to-Peer, P2P) network develops into a kind of important network schemer to equity gradually.This be because with traditional client/server (Client/Server, C/S) pattern is compared, peer-to-peer network have zmodem, extensibility strong, with low cost, can make full use of advantage such as distributed resource.Peer-to-peer network is in the existing application widely of fermentation such as file-sharing, Distributed Calculation, streaming media service.

Peer-to-peer network is divided into two kinds: have structure with structureless, the be connected with strict restriction of the peer-to-peer network of structure to node arranged, and structureless peer network node be connected with the very big degree of freedom.Because the distributed nature of non-structure peer network node; Node is only known the resource of its neighbor node; But and do not know in the network distribution of resource situation on other node, so a great difficult problem is the resource orientation problem in the network in the non-structure peer-to-peer network.Proposed many searching methods for this reason and realized resources effective location on the non-structure peer-to-peer network, like inundation method, random walks, APS (Adaptive Probability Search) etc.Yet above method in the resource lookup process because the forwarding of query messages; Cause the message of part of nodes needs processing in search procedure in the network message congestion too much to occur; And then have a strong impact on proceeding of search; Cause the decline of search performance, thereby how to realize in peer-to-peer network that load balancing just seems particularly important.

So-called load balancing is meant shares offered load on each node according to the disposal ability of node.Existing load-balancing method mainly contains two types; One type is research realizes network through dynamic change network topology structure load balancing strategy; The method that changes network topology structure can make the strong node of disposal ability have more neighbours; Share more offered load, make full use of the disposal ability of self, reach the purpose of load balancing.Adjust network structure like Gia based on the satisfaction level of node; The DANTE that Luis etc. propose, node regularly initiate to reconnect, and utilize the attraction of node to confirm to reconnect those nodes.The speed that node initiates to inquire about in these methods is all fixed, if the query load in the network of the basicly stable back of topological structure increases suddenly is easy to cause congested at Centroid.Another kind of is that research realizes load-balancing method on the basis that does not change topology of networks; These class methods realize load balancing through the node of on purpose transferring to the unnecessary query messages of congested node unnecessary disposal ability; In this type research the problem of most critical be how to find congested; Because characteristics such as the high flexibility of intelligent mobile agent, high efficiency, low load, low communication delay, high asynchronism have been applied in the congested cognitive method it gradually.Shekhar etc. have proposed a kind of intelligent mobile agent Routing Protocol based on congested perception, and intelligent mobile agent is selected next access node randomly from neighbor node; Li etc. have proposed a kind of peer-to-peer network load-balancing method based on intelligent mobile agent, and intelligent mobile agent is sought congested node through all nodes in the rotation visit node.But existing load-balancing method based on intelligent mobile agent all is on the basis that does not change topology of networks, to carry out.

In view of this, a kind of new non-structure peer-to-peer network load-balancing method and system have been necessary to propose based on intelligent mobile agent.

Summary of the invention

The object of the present invention is to provide a kind of non-structure peer-to-peer network load-balancing method and system, avoided network node congested, realize load balancing based on intelligent mobile agent.

A kind of non-structure peer-to-peer network load-balancing method based on intelligent mobile agent of the present invention said method comprising the steps of:

Each node is regularly initiated topology adjustment mechanism in S1, the network, and the node strong to disposal ability connects, and forms the center type topological structure;

S2, the congestion state through computing node congestion level decision node at a time further carry out topology optimization.

As further improvement of the present invention, said step S1 is specially:

S11, node P _iSend and collect the both candidate nodes collection that message is collected this node topology adjustment of participation;

S12, node is collected the neighbor node that forwards to is selected at random, and add this node to S set _iTime-to-live TTL subtracts 1 simultaneously; If time-to-live TTL is not equal to 0, then repeating step S12 continues to collect node;

S13, both candidate nodes collection Si is returned to message initiate node P _i

S14, from node P _iThe set of current neighbor node in select the most weak and connection degree of disposal ability greater than 1 node P _{Min (Ni)}

S15, from both candidate nodes collection S _iThe middle the strongest node of disposal ability that takes out is designated as P _{Max (Si)}, and with it from S _iIf middle deletion is node P _{Max (Si)}Disposal ability be not more than P _{Min (Ni)}, execution in step S17; If node P _{Max (Si)}Disposal ability greater than P _{Min (Ni)}, then to node P _{Max (Si)}Send connection request, if this node P _{Max (Si)}The refusal connection request, then repeating step S15 continues to select the next one to reconnect node, if node P _{Max (Si)}Acceptance reconnects request, then execution in step S16;

S16, node P _iBreak off and the neighbor node P that selects _{Min (Ni)}Connection and with node P _{Max (Si)}Connect.If reaching the upper limit that predefined node reconnects number, the node number that connects again counts n, then execution in step S17; If not, then execution in step S14 selects to participate in the node that reconnects next time;

S17, this topology adjustment finish.

As further improvement of the present invention, said step S2 is specially:

S21, the intelligent mobile agent computing node P that gathers information _jCongestion level at current time t

And utilize

Value come the congestion state of decision node, comprise each node set overload, normal load and three kinds of load conditions of overload, and set the load condition threshold value, with what calculate

Thereby value and the load condition threshold of setting are confirmed the present load state of node; If node P _jLoad condition is overload, then execution in step S22; Otherwise execution in step S23 carries out the optimization of topology, selects next access node;

S22, said node P _jSelect a part of node to break off at random and connect, and derive a sub-intelligent mobile agent; Said sub-intelligent mobile agent is a father node with its generation node; In network, seek the node of unnecessary disposal ability with the mode of breadth First traversal; Every load condition that reaches node elder generation decision node, if node be underloading just select a node to be reconnected to be attached thereto at random, continue to seek next node then; Reconnect node up to all waiting and all reconnect and finish, sub-intelligent mobile agent initiatively exits network;

S23, intelligent mobile agent are according to node P _jNeighbor node P _oTo the attraction table of this intelligent mobile agent and the node listing of visiting, select next access node, and move to this node; The node listing of said visit is used to write down the node that intelligent body was visited; Said neighbor node attraction table is the table to the intelligent mobile agent attraction of its neighbor node that any node is used to write down in the network.

As further improvement of the present invention, among the said step S21, said node P _jCongestion level at current time t

Computing formula be:

${\mathrm{CL}}_{P_j}\left(t\right)=\frac{1+Q_{P_j}\left(t\right)}{C_{P_j}}---\left(1\right)$

Wherein,

Expression node P _jHandle the time that query messages is required; If node just all is placed in the message buffering formation of this node in the message that processing messages arrives so this moment,

Expression is t node P constantly _jBuffer queue in pending message count.

As a further improvement on the present invention, further comprise among the said step S21:

The said node P of step S210 _jWhether accept the connection request of other node according to the load condition decision of oneself: if node P _jCurrent is overload, node P _jDo not accept other node connection request; If node P _jCurrent is normal load, and if this moment node P _jLast status be overload, node P then _jDo not accept other node connection request; If node P _jLast status is not an overload, and node P _jCurrent is normal condition, then node P _jAccept the connection request of other node; If node P _jCurrent state is then node P of underloading _jAccept the connection request of other node.

As further improvement of the present invention, among the said step S21, said node P _jCongestion level at current time t

Computing formula be:

${\mathrm{CL}}_{P_j}\left(t\right)=\frac{1+Q_{P_j}\left(t\right)}{C_{P_j}}---\left(1\right)$

Wherein, Expression node P _jHandle the time that query messages is required; If node just all is placed in the message buffering formation of this node in the message that processing messages arrives so this moment,

Expression is t node P constantly _jBuffer queue in pending message count.

As further improvement of the present invention, " will calculate among the said step S21 Thereby value and the load condition threshold of setting are confirmed the present load state of node " be specially: two threshold value U of setting _Thred, B _Thred, if

Node P _jState transship, note variable flag=1, if Node P _jBe normally loaded, note variable flag=0.If

Node P _jState be underloading, note variable flag=-1.

As further improvement of the present invention, said step S210 specifically comprises: said node P _jCurrent variable state flag=1, node P _jDo not accept other node connection request; If node P _jCurrent state flag=0, and if this moment node P _jLast variable state flag=1, then a node P _jDo not accept other node connection request; If node P _jA last variable state flag=0 or-1, and node P _jCurrent variable state flag=0 is node P then _jCan accept the connection request of other node; If node P _jCurrent variable state flag=-1, then node P _jCan accept the connection request of other node.

As further improvement of the present invention, among the said step S23, said node P _jBe provided with the neighbor node P of this node _oThe attraction table, this table is used for writing down node P _jGo up the disposal ability of all neighbor nodes

And degree of communication χ (P _o, k); Neighbor node P _oAttraction to intelligent mobile agent is calculated as follows:

$A_{P_o}=C_{P_o}\×\mathrm{\χ}(P_o,k)---\left(2\right)$

Be neighbor node P _oDisposal ability, represent this neighbor node P _oHandle the time that query messages is required; χ (P _o, k) be node P _oDegree of communication; Node P _oAttraction

Big more, then intelligent mobile agent is from P _jMove to node P _oPossibility high more;

Said neighbor node P _oThe degree of communication computing formula following:

$\mathrm{\χ}(P_o,k)=\underset{h=1}{\overset{h=k}{\mathrm{\Σ}}}\frac{N(P_o,h)}{h^{\mathrm{\δ}}}---\left(3\right)$

Wherein, N (P _o, h) expression and neighbor node P _oThe number of the node of h jumping apart, k representes to calculate the radius of degree of communication, and δ is a control coefrficient, 1/h ^δBe used for controlling the weight of different distance node, χ (P to the degree of communication influence _o, k) big more node P _oThe query messages that possibly receive is many more.

Correspondingly, a kind of non-structure peer-to-peer network SiteServer LBS based on intelligent mobile agent, said system comprises:

The topology adjustment unit is used for each node of network and regularly initiates topology adjustment mechanism, and the node strong to disposal ability connects, and forms the unit of center type topological structure;

The topological optimization unit is used for the congestion state through computing node congestion level decision node at a time, further carries out the unit of topology optimization;

Wherein, said topological adjustment unit comprises:

Node P _iSend and collect the unit that message is collected the both candidate nodes collection of participating in this node topology adjustment;

Node is collected forwards to a neighbor node of selecting at random, and add this node to S set _i

With both candidate nodes collection S _iReturn to message and initiate node P _iThe unit;

From node P _iThe set of current neighbor node in select the most weak and connection degree of disposal ability greater than 1 node P _{Min (Ni)}The unit;

From both candidate nodes collection S _iThe middle the strongest node P of disposal ability that takes out _{Max (Si)}, comparison node P _{Max (Si)}With P _{Min (Ni)}Disposal ability, then to node P _{Max (Si)}Send the unit of connection request;

Node P _iBreak off and the neighbor node P that selects _{Min (Ni)}Connection and with node P _{Max (Si)}The unit that connects;

Said topological optimization unit comprises:

The intelligent mobile agent computing node P that gathers information _jCongestion level at current time t

Each node is set overload, normal load and three kinds of load conditions of overload, and set the load condition threshold value, with what calculate

Thereby value and the load condition threshold of setting are confirmed the unit of the present load state of node;

Node P _jSelect a part of node to break off at random and connect, and derive the unit of a sub-intelligent mobile agent;

Sub-intelligent mobile agent is a father node with its generation node; In network, seek the node of unnecessary disposal ability with the mode of breadth First traversal; Every load condition that reaches a node elder generation decision node; If node be underloading just select a node to be reconnected to be attached thereto at random; Continue to seek next node then; Reconnecting node up to all waiting all reconnects and finishes the unit that sub-intelligent mobile agent initiatively exits network;

Intelligent mobile agent is according to node P _jNeighbor node P _oThe attraction of intelligent mobile agent and the node listing of visiting are selected next access node, and move to the unit of this node.

As the further improvement of system of the present invention, said node P _jCongestion level at current time t Computing formula be:

${\mathrm{CL}}_{P_j}\left(t\right)=\frac{1+Q_{P_j}\left(t\right)}{C_{P_j}}---\left(1\right)$

Wherein, Expression node P _jHandle the time that query messages is required; If node just all is placed in the message buffering formation of this node in the message that processing messages arrives so this moment,

Expression is t node P constantly _jBuffer queue in pending message count;

Said neighbor node P _oAttraction to intelligent mobile agent is calculated as follows:

$A_{P_o}=C_{P_o}\×\mathrm{\χ}(P_o,k)---\left(2\right)$

Be neighbor node P _oDisposal ability, represent this neighbor node P _oHandle the time that query messages is required; χ (P _o, k) be node P _oDegree of communication; Node P _oAttraction

Big more, then intelligent mobile agent is from P _jMove to node P _oPossibility high more;

Said neighbor node P _oThe degree of communication computing formula following:

$\mathrm{\χ}(P_o,k)=\underset{h=1}{\overset{h=k}{\mathrm{\Σ}}}\frac{N(P_o,h)}{h^{\mathrm{\δ}}}---\left(3\right)$

Wherein, N (P _o, h) expression and neighbor node P _oThe number of the node of h jumping apart, k representes to calculate the radius of degree of communication, and δ is a control coefrficient, 1/h ^δBe used for controlling the weight of different distance node, χ (P to the degree of communication influence _o, k) big more node P _oThe query messages that possibly receive is many more.

The invention has the beneficial effects as follows: each node regularly starts topology and reconnects mechanism in the network of the present invention, to form the topological structure of centralization.Simultaneously the node information such as disposal ability, degree of communication of constantly collecting neighbor node have purpose constantly migration and the constantly loading condition of important node in the monitor network in network to instruct intelligent mobile agent; Topology of networks is optimized adjustment; Avoid congested generation, thereby realize the load balancing on the non-structure peer-to-peer network, can also adjust network topology structure timely according to the load state of network; Can avoid network node congested effectively, improve resource lookup efficient.

Description of drawings

Fig. 1 is the flow chart of a kind of non-structure peer-to-peer network load-balancing method based on intelligent mobile agent of the present invention.

Fig. 2 is the particular flow sheet that step S1 network node topology is regularly adjusted among Fig. 1.

Fig. 3 is the particular flow sheet of the topological optimization of step S2 intelligent mobile agent among Fig. 1.

Fig. 4 is a network topology simulation drawing in an embodiment of the present invention.

Fig. 5 reconnects the aggregation extent figure of back network node for process topology in an embodiment of the present invention.

Fig. 6 reconnects the average connection degree variation diagram of different capable nodes in the network of back for process topology in an embodiment of the present invention.

Fig. 7 counts variation diagram for the average nodal of the process that inquiry after the process topology reconnects in an embodiment of the present invention needs.

Fig. 8 is a kind of non-structure peer-to-peer network SiteServer LBS structural representation based on intelligent mobile agent of the present invention.

Embodiment

Below will combine each execution mode shown in the drawings to describe the present invention.But these execution modes do not limit the present invention, and the conversion on the structure that those of ordinary skill in the art makes according to these execution modes, method or the function all is included in protection scope of the present invention.

Join the flow chart for a kind of non-structure peer-to-peer network load-balancing method based on intelligent mobile agent of the present invention shown in Figure 1, may further comprise the steps:

Each node is regularly initiated topology adjustment mechanism in S1, the network, and the node strong to disposal ability connects, and forms the center type topological structure;

S2, the congestion state through computing node congestion level decision node at a time further carry out topology optimization.

Joining shown in Figure 2 is the particular flow sheet that step S1 network node topology is regularly adjusted in the present embodiment, may further comprise the steps:

S11, node P _iSend and collect the both candidate nodes collection that message is collected this node topology adjustment of participation.Suppose that this both candidate nodes set is S _i, and initial S _iBe sky, (Time to Live TTL) is K to the time-to-live TTL of this message;

S12, node is collected the neighbor node that forwards to is selected at random, and add this node to S set _iThe time-to-live TTL that collects message subtracts 1, judges whether time-to-live TTL equals 0, if node collection process finishes, and execution in step S13, if not, repeating step S12 continues to collect this node;

S13, with both candidate nodes collection S _iReturn to message and initiate node P _i

S14, from node P _iThe set of current neighbor node in select the most weak and connection degree of disposal ability greater than 1 node P _{Min (Ni)}, N _iBe node P _iThe set of neighbor node;

S15, from both candidate nodes collection S _iThe middle the strongest node P of disposal ability that takes out _{Max (Si)}, and with node P _{Max (Si)}It is from S _iMiddle deletion.Comparison node P _{Max (Si)}With P _{Min (Ni)}Disposal ability, if node P _{Max (Si)}Disposal ability be not more than P _{Min (Ni)}, execution in step S17 then; If node P _{Max (Si)}Disposal ability greater than P _{Min (Ni)}, then to node P _{Max (Si)}Send connection request.If node P _{Max (Si)}The refusal connection request, then repeating step S15 continues to select the next one to reconnect node, if node P _{Max (Si)}Accept connection request, then execution in step S16;

S16, node P _iBreak off and the neighbor node P that selects _{Min (Ni)}Connection and with node P _{Max (Si)}Connect.If reaching the upper limit that predefined node reconnects number, the node number that connects again counts n, then execution in step S17; If not, then execution in step S14 selects to participate in the node that reconnects next time;

S17, this topology adjustment finish.

The simple regular topology of node self that relies on reconnects; Finally can form one is the centralized network at center with the strongest several nodes of disposal ability; Internet resources search efficiency there not being this structure under the congested situation is the highest; It is congested that but the Centroid that the resource lookup request of nearly all node all can be sent, Centroid possibly be easy to take place, and this can have a strong impact on the efficient of resource lookup; For fear of congested generation, introduced Topology Optimization Method on this basis again based on intelligent mobile agent.

Joining shown in Figure 3 is the particular flow sheet of the topological optimization of step S2 intelligent mobile agent in this execution mode, may further comprise the steps:

S21, the intelligent mobile agent computing node P that gathers information _jCongestion level at current time t

And utilize

Value come the congestion state of decision node, confirm the present load state of node, if node P _jLoad condition transships, then execution in step S22; Otherwise execution in step S23 carries out the optimization of topology, selects next access node;

S22, node P _jSelect a part of node to break off at random and connect, and derive a sub-intelligent mobile agent; Sub-intelligent mobile agent is a father node with its generation node; In network, seek the node of unnecessary disposal ability with the mode of breadth First traversal; Every load condition that reaches node elder generation decision node, if node be underloading just select a node to be reconnected to be attached thereto at random, continue to seek next node then; Reconnect node up to all waiting and all reconnect and finish, sub-intelligent mobile agent initiatively exits network;

S23, intelligent mobile agent are according to node P _jNeighbours' node listing that its attraction tabulation and intelligent mobile agent were visited select next access node, and move to this node.

Wherein among the step S21, node P _jCongestion level at current time t

Computing formula be:

${\mathrm{CL}}_{P_j}\left(t\right)=\frac{1+Q_{P_j}\left(t\right)}{C_{P_j}}---\left(1\right)$

Expression node P _jHandle the time that query messages is required; If node just all is placed in the message buffering formation of this node in the message that processing messages arrives so this moment,

Expression is t node P constantly _jBuffer queue in pending message count.This formulate is forwarded to node P at query messages of moment t _jNeed the time of cost to handling to finish.

" confirming the present load state of node " among the step S21 is specially:

Each node all is set with three load conditions, according to

Value and two threshold value U of setting _Thred, B _ThredConfirm the present load state of node thereby compare:

If

Node P _jState transship, note variable flag=1, think that this node is congested this moment;

If

Node P _jBe normally loaded, note variable flag=0;

If

Node P _jState be underloading, note variable flag=-1.

Further, node P _jWhether accept the connection request of other node according to the load condition decision of oneself.Be node P _jCurrent variable state flag=1, node P _jDo not accept other node connection request; If node P _jCurrent state flag=0, and if this moment node P _jLast variable state flag=1, then a node P _jDo not accept other node connection request; If node P _jA last variable state flag=0 or-1, and node P _jCurrent variable state flag=0 is node P then _jCan accept the connection request of other node; If node P _jCurrent variable state flag=-1, then node P _jCan accept the connection request of other node.

In step S23, for confirming node P _jGo up the attraction power of all neighbor nodes, node P _jMaintain the neighbor node attraction table of this node, this table is used for writing down node P _jGo up the disposal ability and the degree of communication of all neighbor nodes.Suppose node P _jA certain neighbor node be P _o, P _oThe degree of communication computing formula following:

$\mathrm{\χ}(P_o,k)=\underset{h=1}{\overset{h=k}{\mathrm{\Σ}}}\frac{N(P_o,h)}{h^{\mathrm{\δ}}}---\left(2\right)$

N (P wherein _o, h) expression and node P _oThe number of the node of h jumping apart, k representes to calculate the radius of degree of communication, and δ is a control coefrficient.1/h ^δBe used for controlling the weight of different distance node, χ (P to the degree of communication influence _o, k) big more node P _oThe query messages that possibly receive is many more.

Node P _oAttraction to intelligent mobile agent is calculated as follows:

$A_{P_o}=C_{P_o}\×\mathrm{\χ}(P_o,k)---\left(3\right)$

Be node P _oDisposal ability, χ (P _o, k) be node P _oDegree of communication.Node P _oAttraction

Big more, then intelligent mobile agent is from P _jMove to node P _oPossibility high more.

Provide the example of a practical implementation process of the present invention below, comprise 10000 network nodes in the instance, 50 different data are arranged on each node, each data has 100 copies in network.First node by adding network generates intelligent mobile agent; The disposal ability of node divides five levels to be respectively: 0.1,1,10,100,1000; The percentage of shared network node corresponds to: 20,45,30,4.9,0.1; Adopt the mode of random walks to carry out resource searching, utilize following two modes will carry out topology and reconnect, to realize load balancing.A topology that is node self regularly carries out reconnects, and purpose is to make the strong node of disposal ability can have more connection, thereby can handle more query messages, makes full use of their disposal ability; Another is based on the topological optimization of intelligent mobile agent, and purpose is to prevent that the strong node of disposal ability from having too much neighbours and causing congestedly, influences the performance of network on the contrary.

Each node whenever initiated once at a distance from 30 seconds that topology reconnects, and made self to the strong node connection of disposal ability, and Fig. 4 is a simulation drawing of network topology, is node P in the simulation drawing below ₁Once topological reconnect procedure:

S11, node P ₁Send and collect the both candidate nodes collection that message is collected this node topology adjustment of participation.Suppose that this both candidate nodes set is S ₁, and initial S ₁Be sky, time-to-live TTL is 5.

S12, node is collected the neighbor node that forwards to is selected at random, and add this node to S set ₁, time-to-live TTL subtracts 1 simultaneously.If time-to-live TTL is not equal to 0, then repeating step S12 continues to collect node; Otherwise node collection process finishes, execution in step S13.

S13, with both candidate nodes collection S ₁Return to message and initiate node P ₁, suppose S ₁Be { P, P ₄, P ₉, P ₁₂, P ₁₃, follow execution in step S14.

S14, from node P ₁The set of current neighbor node, according to Fig. 4 P then ₁The current neighbor node set of node is { P, P ₃, select the most weak and connection degree of disposal ability greater than 1 node, be designated as P _{Min (N1)}, N ₁Be node P ₁The neighbor node collection; Follow execution in step S15.P for example ₁Neighbor node in the most weak node of current disposal ability be P ₃, P _{Min (N1)}Be P ₃Node, execution in step S15 then.

S15, from both candidate nodes collection S ₁The middle the strongest node of disposal ability that takes out is designated as P _{Max (S1)}, with P _{Max (S1)}From S ₁If middle deletion is node P _{Max (S1)}Disposal ability be not more than P _{Min (N1)}, the disposal ability that all residue both candidate nodes then are described is all than node P ₁Existing neighbor node a little less than, execution in step S17; If node P _{Max (S1)}Disposal ability greater than P _{Min (N1)}, then to node P _{Max (S1)}Send connection request, if this node P _{Max (S1)}The refusal connection request, then repeating step S15 continues to select the next one to reconnect node, if node P _{Max (S1)}Acceptance reconnects request, then execution in step S16.For example from both candidate nodes collection S ₁Selecting the strongest node of current disposal ability is P ₄, and P ₄Disposal ability greater than the P that selects among the step S14 ₃Node, then node P ₁To P ₄Send connection request, and P ₄Accepted connection request, then execution in step S16.

S16, node P ₁Break off and the neighbor node P that selects _{Min (N1)}Connection and with node P _{Max (S1)}Connect, if there be n both candidate nodes to participate in reconnecting, n selects 3 in this instance, then execution in step S17; Otherwise execution in step S14 begins to select to participate in the node that reconnects next time.Node P for example ₁With the P that selects through step S14 ₃Node breaks off and connecting and and P ₄Node connects, owing to there are not 3 both candidate nodes to participate in reconnecting, then execution in step S14 begins to select to participate in the node that reconnects next time.

S17, this topology adjustment finish.

Join shown in Figure 4ly, suppose that the current place of intelligent mobile agent node is P, intelligent mobile agent is following in the process of topology optimization of node P:

S21, the intelligent mobile agent congestion level CL of computing node P that gather information at current time T _PAnd utilize CL (T), _P(T) value is come the congestion state of decision node, and each node all is set with three load conditions simultaneously, and we are with the CL that calculates _P(T) value and two threshold value U that set _Thred, B _ThredConfirm the present load state of node thereby compare.Two threshold value U that set _Thred, B _ThredBe respectively 0.1 and 0.05.If CL _P() > T; U _Thred, the state of node transships, the current variable state flag=1 of note node P when moment T, and think that this node is congested this moment.If B _Thred≤CL _P(T)≤U _Thred, node P is a normal load, note flag=0.If CL _P(T)<b _Thred, the state of node P is a underloading, note flag=-1.Node P determines next execution in step according to its load condition, and even the state of node P transships, and then execution in step S22 carries out the optimization of topology; Otherwise execution in step S23 selects next access node.

Whether this exterior node P accepts the connection request of other node according to the load condition decision of oneself.Be the current variable state of node P flag=1, node P does not accept other node connection request; If node P current state flag=0, and if this moment node P a last variable state flag=1, then node P does not accept other node connection request; If the last variable state flag=0 of node P or-1, the current variable state of node P flag=0 then node P can accept the connection request of other node; If the current variable state of node P flag=-1, then node P can accept the connection request of other node;

S22, node P select a part of node to break off connection with it at random, once break off with two nodes in the present embodiment being connected.Suppose that node P is connected with two nodes disconnections, and derive a sub-intelligent mobile agent.

Sub-intelligent mobile agent mainly is responsible for reconnecting the connection that is disconnected; Sub-intelligent mobile agent is a father node with its generation node, in network, seeks the node of unnecessary disposal ability, every load condition that reaches the first decision node of a node with the mode of breadth First traversal; If node be underloading just select a node to be reconnected to be attached thereto at random; Continue to seek next node then, reconnect node up to all waiting and all reconnect and finish, this moment, sub-intelligent mobile agent initiatively exitted network;

S23, intelligent mobile agent are selected next access node according to the neighbours of node P to its attraction and the node listing of visiting thereof, and move to this node, and wherein k and δ are respectively 2 and 1.The neighbor node of supposing node P among Fig. 4 sorts from big to small according to the attraction to intelligent mobile agent and is: P ₄, P ₅, P ₃, P ₁, P ₂, and the node listing that intelligent mobile agent was visited recently is: P ₁₂, P ₉, P ₄So the next access node of intelligent mobile agent is P ₅

Fig. 5 is aggregation extent CC (clustering coefficient) variation diagram of network node in the process of topology optimization, and the computing formula of CC is following:

$\mathrm{CC}=\frac{1}{\left|V\right|}\underset{p\&Element;V}{\mathrm{\&Sigma;}}{\mathrm{CC}}_p---\left(4\right)$

Wherein V is the set of all nodes in the network, CC _pBe the convergence factor of the neighbor node of node p, computing formula is following:

${\mathrm{CC}}_p=\frac{|(N_p\&times;N_p)\&cap;E|}{k_p(k_p-1)}---\left(5\right)$

N wherein _pBe the set of the neighbor node of node p, k _p=| N _p| be the number of the neighbor node of node p, E is the set of internodal connection in the network.0 ≤CC ≤1, the more little network node of CC value distribute more at random, the big more node of CC value is concentrated more.

Can be found out that by Fig. 5 along with the change of topological structure, the aggregation extent of node is increasingly high, and reach a basicly stable state, the topological structure of visible network is by completely random, and it is more and more concentrated to become.

Fig. 6 has provided the variation of same treatment capable nodes average nodal degree (number of neighbor node) in the topological reconnect procedure; Be the variation that node degree is observed by unit with 100 seconds in the experiment; Can find out to begin to have a large amount of nodes to reconnect the strongest node of processing ability, but reach a stable status soon, and the strong more node connectivity of disposal ability is high more; Explain that the strong more node of disposal ability can share more query load, can make full use of the disposal ability of network node.

Fig. 7 is the variation diagram of the average number of hops of successful inquiring needs, and the jumping figure that in the experiment per 1000 inquiries is needed is averaged and obtained the result like figure, along with the optimization to network topology structure; The average number of hops that inquiry needs reduces gradually; As can be seen from the figure, the jumping figure ratio that needs behind the topological optimization begins to have reduced a lot, and the jumping figure that inquiry needs is few more; The efficient of searching algorithm is just high more, the visible search performance that has improved network based on the Topology Optimization Method of intelligent mobile agent greatly.

Correspondingly, join shown in Figure 8ly, a kind of non-structure peer-to-peer network SiteServer LBS 100 based on intelligent mobile agent of the present invention comprises:

Topology adjustment unit 10 is used for each node of network and regularly initiates topology adjustment mechanism, and the node strong to disposal ability connects, and forms the unit of center type topological structure;

Topological optimization unit 20 is used for the congestion state through computing node congestion level decision node at a time, further carries out the unit of topology optimization;

Topology adjustment unit 10 comprises:

Node P _iSend and collect the unit that message is collected the both candidate nodes collection of participating in this node topology adjustment;

Node is collected forwards to a neighbor node of selecting at random, and add this node to S set _i

With both candidate nodes collection S _iReturn to message and initiate node P _iThe unit;

From node P _iThe set of current neighbor node in select the most weak and connection degree of disposal ability greater than 1 node P _{Min (Ni)}The unit;

From both candidate nodes collection S _iThe middle the strongest node P of disposal ability that takes out _{Max (Si)}, comparison node P _{Max (Si)}With P _{Min (Ni)}Disposal ability, then to node P _{Max (Si)}Send the unit of connection request;

Node P _iBreak off and the neighbor node P that selects _{Min (Ni)}Connection and with node P _{Max (Si)}The unit that connects;

Topological optimization unit 20 comprises:

The intelligent mobile agent computing node P that gathers information _jCongestion level at current time t Each node is set overload, normal load and three kinds of load conditions of overload, and set the load condition threshold value, with what calculate

Thereby value and the load condition threshold of setting are confirmed the unit of the present load state of node; Wherein, node P _jCongestion level at current time t

Computing formula be:

${\mathrm{CL}}_{P_j}\left(t\right)=\frac{1+Q_{P_j}\left(t\right)}{C_{P_j}}$

Wherein,

Expression node P _jHandle the time that query messages is required; If node just all is placed in the message buffering formation of this node in the message that processing messages arrives so this moment, Expression is t node P constantly _jBuffer queue in pending message count;

Node P _jSelect a part of node to break off at random and connect, and derive the unit of a sub-intelligent mobile agent;

Sub-intelligent mobile agent is a father node with its generation node; In network, seek the node of unnecessary disposal ability with the mode of breadth First traversal; Every load condition that reaches a node elder generation decision node; If node be underloading just select a node to be reconnected to be attached thereto at random; Continue to seek next node then; Reconnecting node up to all waiting all reconnects and finishes the unit that sub-intelligent mobile agent initiatively exits network;

Intelligent mobile agent is according to node P _jNeighbor node P _oThe attraction of intelligent mobile agent and the node listing of visiting are selected next access node, and move to the unit of this node; Wherein, neighbor node P _oAttraction to intelligent mobile agent is calculated as follows:

$A_{P_o}=C_{P_o}\&times;\mathrm{\&chi;}(P_o,k)$

Be neighbor node P _oDisposal ability, represent this neighbor node P _oHandle the time that query messages is required; X (P _o, k) be node P _oDegree of communication; Node P _oAttraction

Big more, then intelligent mobile agent is from P _jMove to node P _oPossibility high more;

Said neighbor node P _oThe degree of communication computing formula following:

$\mathrm{\&chi;}(P_o,k)=\underset{h=1}{\overset{h=k}{\mathrm{\&Sigma;}}}\frac{N(P_o,h)}{h^{\mathrm{\&delta;}}}$

Wherein, N (P _o, h) expression and neighbor node P _oThe number of the node of h jumping apart, k representes to calculate the radius of degree of communication, and δ is a control coefrficient, 1/h ^δBe used for controlling the weight of different distance node, χ (P to the degree of communication influence _o, k) big more node P _oThe query messages that possibly receive is many more.

Compared with prior art, each node regularly starts topology and reconnects mechanism in the network of the present invention, to form the topological structure of centralization.Simultaneously the node information such as disposal ability, degree of communication of constantly collecting neighbor node have purpose constantly migration and the constantly loading condition of important node in the monitor network in network to instruct intelligent mobile agent; Topology of networks is optimized adjustment; Avoid congested generation, thereby realize the load balancing on the non-structure peer-to-peer network, can also adjust network topology structure timely according to the load state of network; Can avoid network node congested effectively, improve resource lookup efficient.

For the convenience of describing, be divided into various unit with function when describing above the device and describe respectively.Certainly, when implementing the application, can in same or a plurality of softwares and/or hardware, realize the function of each unit.

Description through above execution mode can know, those skilled in the art can be well understood to the application and can realize by the mode that software adds essential general hardware platform.Based on such understanding; The part that the application's technical scheme contributes to prior art in essence in other words can be come out with the embodied of software product; This computer software product can be stored in the storage medium, like ROM/RAM, magnetic disc, CD etc., comprises that some instructions are with so that a computer equipment (can be a personal computer; Server, the perhaps network equipment etc.) carry out the described method of some part of each execution mode of the application or execution mode.

Device embodiments described above only is schematic; Wherein said unit as the separating component explanation can or can not be physically to separate also; The parts that show as the unit can be or can not be physical locations also; Promptly can be positioned at a place, perhaps also can be distributed on a plurality of NEs.Can select wherein some or all of module to realize the purpose of present embodiment scheme according to the actual needs.Those of ordinary skills promptly can understand and implement under the situation of not paying creative work.

The application can be used in numerous general or special purpose computingasystem environment or the configuration.For example: personal computer, server computer, handheld device or portable set, plate equipment, multicomputer system, the system based on microprocessor, set top box, programmable consumer-elcetronics devices, network PC, minicom, mainframe computer, comprise DCE of above any system or equipment or the like.

The application can describe in the general context of the computer executable instructions of being carried out by computer, for example program module.Usually, program module comprises the routine carrying out particular task or realize particular abstract, program, object, assembly, data structure or the like.Also can in DCE, put into practice the application, in these DCEs, by through communication network connected teleprocessing equipment execute the task.In DCE, program module can be arranged in this locality and the remote computer storage medium that comprises memory device.

Be to be understood that; Though this specification is described according to execution mode; But be not that each execution mode only comprises an independently technical scheme, this narrating mode of specification only is for clarity sake, and those skilled in the art should make specification as a whole; Technical scheme in each execution mode also can form other execution mode that it will be appreciated by those skilled in the art that through appropriate combination.

The listed a series of detailed description of preceding text only is specifying to feasibility execution mode of the present invention; They are not in order to restriction protection scope of the present invention, allly do not break away from equivalent execution mode or the change that skill of the present invention spirit done and all should be included within protection scope of the present invention.

Claims (10)

1. the non-structure peer-to-peer network load-balancing method based on intelligent mobile agent is characterized in that, said method comprising the steps of:

Each node is regularly initiated topology adjustment mechanism in S1, the network, and the node strong to disposal ability connects, and forms the center type topological structure;

S2, the congestion state through computing node congestion level decision node at a time further carry out topology optimization.

2. method according to claim 1 is characterized in that, said step S1 is specially:

S11, node P _iSend and collect the both candidate nodes collection that message is collected this node topology adjustment of participation;

S12, node is collected the neighbor node that forwards to is selected at random, and add this node to S set _iTime-to-live TTL subtracts 1 simultaneously; If time-to-live TTL is not equal to 0, then repeating step S12 continues to collect node;

S13, with both candidate nodes collection S _iReturn to message and initiate node P _i

S14, from node P _iThe set of current neighbor node in select the most weak and connection degree of disposal ability greater than 1 node P _{Min (Ni)}

S15, from both candidate nodes collection S _iThe middle the strongest node of disposal ability that takes out is designated as And with it from S _iIf middle deletion is node

Disposal ability be not more than

Execution in step S17; If node

Disposal ability greater than Then to node

Send connection request, if this node

The refusal connection request, then repeating step S15 continues to select the next one to reconnect node, if node

Acceptance reconnects request, then execution in step S16;

S16, node P _iBreak off and the neighbor node P that selects _{Min (Ni)}Connection and with node P _{Max (Si)}Connect; If reaching the upper limit that predefined node reconnects number, the node number that connects again counts n, then execution in step S17; If not, then execution in step S14 selects to participate in the node that reconnects next time;

S17, this topology adjustment finish.

3. the non-structure peer-to-peer network load-balancing method based on intelligent mobile agent according to claim 1 is characterized in that said step S2 is specially:

S21, the intelligent mobile agent computing node P that gathers information _jCongestion level at current time t

And utilize

Value come the congestion state of decision node, comprise each node set overload, normal load and three kinds of load conditions of overload, and set the load condition threshold value, with what calculate

Thereby value and the load condition threshold of setting are confirmed the present load state of node; If node P _jLoad condition is overload, then execution in step S22; Otherwise execution in step S23 carries out the optimization of topology, selects next access node;

S22, said node P _jSelect a part of node to break off at random and connect, and derive a sub-intelligent mobile agent; Said sub-intelligent mobile agent is a father node with its generation node; In network, seek the node of unnecessary disposal ability with the mode of breadth First traversal; Every load condition that reaches node elder generation decision node, if node be underloading just select a node to be reconnected to be attached thereto at random, continue to seek next node then; Reconnect node up to all waiting and all reconnect and finish, sub-intelligent mobile agent initiatively exits network;

S23, intelligent mobile agent are according to node P _jNeighbor node P _oTo the attraction table of this intelligent mobile agent and the node listing of visiting, select next access node, and move to this node; The node listing of said visit is used to write down the node that intelligent body was visited; Said neighbor node attraction table is the table to the intelligent mobile agent attraction of its neighbor node that any node is used to write down in the network.

4. the non-structure peer-to-peer network load-balancing method based on intelligent mobile agent according to claim 3 is characterized in that, further comprises among the said step S21:

The said node P of step S210 _jWhether accept the connection request of other node according to the load condition decision of oneself: if node P _jCurrent is overload, node P _jDo not accept other node connection request; If node P _jCurrent is normal load, and if this moment node P _jLast status be overload, node P then _jDo not accept other node connection request; If node P _jLast status is not an overload, and node P _jCurrent is normal condition, then node P _jAccept the connection request of other node; If node P _jCurrent state is then node P of underloading _jAccept the connection request of other node.

5. want 3 described methods according to right, it is characterized in that, among the said step S21, said node P _jCongestion level at current time t

Computing formula be:

Wherein,

Expression node P _jHandle the time that query messages is required; If node just all is placed in the message buffering formation of this node in the message that processing messages arrives so this moment,

Expression is t node P constantly _jBuffer queue in pending message count.

6. want 3 described non-structure peer-to-peer network load-balancing methods according to right, it is characterized in that based on intelligent mobile agent, among the said step S23, said node P _jBe provided with the neighbor node P of this node _oThe attraction table, this table is used for writing down node P _jGo up the disposal ability of all neighbor nodes And degree of communication χ (P _o, k);

Neighbor node P _oAttraction to intelligent mobile agent is calculated as follows:

$A_{P_o}=C_{P_o}\&times;\mathrm{\&chi;}(P_o,k)$

Be neighbor node P _oDisposal ability, represent this neighbor node P _oHandle the time that query messages is required;

χ (P _o, k) be node P _oDegree of communication; Node P _oAttraction

Big more, then intelligent mobile agent is from P _jMove to node P _oPossibility high more;

Said neighbor node P _oThe degree of communication computing formula following:

$\mathrm{\&chi;}(P_o,k)=\underset{h=1}{\overset{h=k}{\mathrm{\&Sigma;}}}\frac{N(P_o,h)}{h^{\mathrm{\&delta;}}},$

Wherein, N (P _o, h) expression and neighbor node P _oThe number of the node of h jumping apart, k representes to calculate the radius of degree of communication, and δ is a control coefrficient, 1/h ^δBe used for controlling the weight of different distance node, χ (P to the degree of communication influence _o, k) big more node P _oThe query messages that possibly receive is many more.

7. want 4 described non-structure peer-to-peer network load-balancing methods according to right, it is characterized in that, " will calculate among the said step S21 based on intelligent mobile agent

Thereby value and the load condition threshold of setting are confirmed the present load state of node " be specially: two threshold value U of setting _Thred, B _Thred, if Node P _jState transship, note variable flag=1, if

Node P _jBe normally loaded, note variable flag=0.If

Node P _jState be underloading, note variable flag=-1.

8. want 7 described non-structure peer-to-peer network load-balancing methods according to right, it is characterized in that said step S210 specifically comprises: said node P based on intelligent mobile agent _jCurrent variable state flag=1, node P _jDo not accept other node connection request; If node P _jCurrent state flag=0, and if this moment node P _jLast variable state flag=1, then a node P _jDo not accept other node connection request; If node P _jA last variable state flag=0 or-1, and node P _jCurrent variable state flag=0 is node P then _jCan accept the connection request of other node; If node P _jCurrent variable state flag=-1, then node P _jCan accept the connection request of other node.

9. non-structure peer-to-peer network SiteServer LBS based on intelligent mobile agent is characterized in that said system comprises:

The topology adjustment unit is used for each node of network and regularly initiates topology adjustment mechanism, and the node strong to disposal ability connects, and forms the unit of center type topological structure;

The topological optimization unit is used for the congestion state through computing node congestion level decision node at a time, further carries out the unit of topology optimization;

Wherein, said topological adjustment unit comprises:

Node P _iSend and collect the unit that message is collected the both candidate nodes collection of participating in this node topology adjustment; Node is collected forwards to a neighbor node of selecting at random, and add this node to S set i; With both candidate nodes collection S _iReturn to message and initiate node P _iThe unit;

From node P _iThe set of current neighbor node in select the most weak and connection degree of disposal ability greater than 1 node P _{Min (Ni)}The unit;

From both candidate nodes collection S _iThe middle the strongest node P of disposal ability that takes out _{Max (Si)}, comparison node P _{Max (Si)}With P _{Min (Ni)}Disposal ability, then to node P _{Max (Si)}Send the unit of connection request;

Node P _iBreak off and the neighbor node P that selects _{Min (Ni)}Connection and with node P _{Max (Si)}The unit that connects;

Said topological optimization unit comprises:

The intelligent mobile agent computing node P that gathers information _jCongestion level at current time t

Each node is set overload, normal load and three kinds of load conditions of overload, and set the load condition threshold value, with what calculate Thereby value and the load condition threshold of setting are confirmed the unit of the present load state of node;

Node P _jSelect a part of node to break off at random and connect, and derive the unit of a sub-intelligent mobile agent;

Sub-intelligent mobile agent is a father node with its generation node; In network, seek the node of unnecessary disposal ability with the mode of breadth First traversal; Every load condition that reaches a node elder generation decision node; If node be underloading just select a node to be reconnected to be attached thereto at random; Continue to seek next node then; Reconnecting node up to all waiting all reconnects and finishes the unit that sub-intelligent mobile agent initiatively exits network;

Intelligent mobile agent is according to node P _jNeighbor node P _oThe attraction of intelligent mobile agent and the node listing of visiting are selected next access node, and move to the unit of this node.

10. the non-structure peer-to-peer network SiteServer LBS based on intelligent mobile agent according to claim 9 is characterized in that said node P _jCongestion level at current time t Computing formula be: Wherein, Expression node P _jHandle the time that query messages is required; If node just all is placed in the message buffering formation of this node in the message that processing messages arrives so this moment,

Expression is t node P constantly _jBuffer queue in pending message count;

Said neighbor node P _oAttraction to intelligent mobile agent is calculated as follows:

$A_{P_o}=C_{P_o}\&times;\mathrm{\&chi;}(P_o,k),$

Be neighbor node P _oDisposal ability, represent this neighbor node P _oHandle the time that query messages is required; χ (P _o, k) be node P _oDegree of communication; Node P _oAttraction

Big more, then intelligent mobile agent is from P _jMove to node P _oPossibility high more;

Said neighbor node P _oThe degree of communication computing formula following:

$\mathrm{\&chi;}(P_o,k)=\underset{h=1}{\overset{h=k}{\mathrm{\&Sigma;}}}\frac{N(P_o,h)}{h^{\mathrm{\&delta;}}},$

Wherein, N (P _o, h) expression and neighbor node P _oThe number of the node of h jumping apart, k representes to calculate the radius of degree of communication, and δ is a control coefrficient, 1/h ^δBe used for controlling the weight of different distance node, χ (P to the degree of communication influence _o, k) big more node P _oThe query messages that possibly receive is many more.

Images