CN101753450A - Method for joint optimization of resources in three-layered network - Google Patents

Abstract

The present invention is a kind of method for joint optimization of resources in three-layered network, be divided into two parts: resource optimization and interlayer resource optimization in the layer, each layer utilizes the ant agent of control plane to carry out routing, definition path quality critical parameter θ is comprehensive relevant with communication all multifactor, with optimal path renewal routing table and ant pheromones table; Lower network blocking probability, when transmission node lost efficacy, date back to a node, communicate by letter according to ant pheromones table real-time recovery; Interlayer connects by the interlayer link, and the service request that the upper strata is got clogged utilizes lower floor's idling-resource to transmit.The transmission of the interior service request of the routing of ant agent and network is synchronous process in the layer.Utilized the idling-resource of each layer fully, reduce dynamic service request reject rate greatly, improve network resource utilization, solved the difficult problem of three-layer network resource combined optimization, the response of realization real-time and effective network condition is communicated by letter with recovering, and the present invention can be directly used in the multitiered network combined optimization more than three layers.

Description

Method for joint optimization of resources in three-layered network

Technical field

The invention belongs to communication technical field, relate generally to the resource combined optimization in the multilayer optical-fiber network, specifically is a kind of method for joint optimization of resources in three-layered network.

Background technology

Along with the fast development of telecommunications network, computer network and broadcasting and television network, network configuration generation great variety develops and transition to network N GN of future generation (Next Generation Network)." network integration " is the main thought of future network development, realizes integrated service and infeasible by a kind of network standard.On network configuration, multiple network and the present situation of depositing have experienced IP/ATM, IP/SDH and IP/WDM several stages.Though IP/WDM is considered to the ideal solution of Next Generation Internet network, but for telecom operators, when introducing new network technology, will keep the telecommunication service that can bring income with and corresponding network technology, therefore, in a very long time, the situation of the such three-layer network coexistence of IP/SDH/WDM will appear in network configuration.Present domestic two WDM Experimental Networks that built up: Chinese Natural Science Foundation's net (NSFCnet) and Chinese high speed information demonstration network (CAINONET) all are to use the domestic IP/SDH/WDM device build that is mature on the whole.

Mostly traditional network route method is the method based on Dijkstra, is aided with various strategies again.In middle-size and small-size centralized network, the Dijkstra method has good effect, but along with the increase of network size and this method of popularizing of distributed network little by little are not suitable for the new-type network structure.Existing multitiered network is the IP/WDM two-tier network of considering mostly simultaneously, and existing network infrastructure is the IP//SDH/WDM three-decker, and the method for joint optimization of resources of therefore studying three layers of optical-fiber network has important use to be worth and practical significance.Be necessary to merge and problem such as share is carried out deep discussion and planning with regard to concrete optional path searching, route and interlayer resource.

Open on April 15th, 2009 by China national Department of Intellectual Property, publication number is the patent application " a kind of wavelength route optical network planing method of dynamic service " of CN 101409596A, discloses a kind of wavelength route optical network planing method of dynamic service.This method utilizes static optimization method to obtain the transceiver number of each optical switch node in the optical-fiber network of static optimization and the reuse wavelengths number of every optical fiber, with its input as the dynamic service simulation, obtains network blocking probability.Adjust the intensity of professional randomized jitter according to network blocking probability, repeatedly call static optimization method and finish optical network plan.This method has guaranteed the global optimization characteristic of program results based on static optimization method, by the blocking rate checking, plans that resulting wavelength route optical network can adapt to the service dynamic characteristic simultaneously, makes dynamic service lower network blocking rate be lower than the blocking rate upper limit.There is following defective in this method: 1) can obtain certain effect though utilize the method for static optimization that dynamic service is optimized, can not be applicable to dynamic service fully, can't utilize Internet resources to greatest extent; 2) Route Selection for business is too simple in this method, just selects shortest path, and in the situation of reality, needs to consider some other factor, as link utilization and time delay etc.

And for example, China national Department of Intellectual Property on August 9th, 2006 is open, and publication number is the patent application " heavy route method behind the lamina optical network route selection failure " of CN 1816007A, discloses the heavy route method behind a kind of lamina optical network route selection failure.In the method, when a source originating end routing failure, two nodes that route is different carry out heavy-route as an eliminating information to be delivered to previous source originating end by signaling message, up to the heavy-route success or date back to the first node heavy-route failure that route connects, wherein, when carrying out heavy-route, can only select in two nodes in any one two tuple of getting rid of information centering at most.This invention can solve the validity in level route Route Selection or the heavy-route failure trace-back process effectively, prevents the situation that has link but do not have route to walk.There is following defective in this patent application: 1) though the mechanism of recalling searching that heavy-route adopts can find new route, and spended time is long, if node is too much in the route, time delay will increase greatly, can't handle a lot of real time business; 2) in the process of recalling, if the node of recalling also lost efficacy, this method will lose efficacy so, not have universality.

Project team of the present invention is retrieved to domestic and international patent documentation, the journal article of publishing and from the network channel, finds report or the document closely related and the same with the present invention more as yet.

Summary of the invention

The objective of the invention is to overcome the shortcoming of above-mentioned technology or method existence, proposed a kind ofly to be applicable to the IP/SDH/WDM method for joint optimization of resources in three-layered network based on ant agent, the present invention separates the routing and the transmission of dynamic service, be beneficial to business is carried out real-time transmission, the real-time recovery in path, the resource utilization of raising network; Pathfinding at control plane is undertaken by ant agent, not only take all factors into consideration network such as many-sided situations such as network congestion situation, the grades of service, also adapt to large-scale distributed network, can network condition be responded real-time and effective, reduce the reject rate of dynamic service request; By the cooperative cooperating between heterogeneous networks layer and the layer, make full use of the idling-resource of each layer.

Below the present invention is described in detail

The present invention is a kind of method for joint optimization of resources in three-layered network, it is characterized in that: this method is divided into two parts: i.e. resource optimization and interlayer resource optimization in the layer, in each layer, the routing and the service request transmission of dynamic service are separated, employing utilizes and carries out routing in the ant agent of control plane based on the comprehensive method for routing AIR (Ant agent-based integrated routingalgorithm) of ant agent; Take combined optimization between layer and layer, the service request that the upper strata is got clogged can utilize the resource of lower floor's free time to transmit, and reaches the interlayer combined optimization, and the routing of ant agent and the arrival of service request are synchronous processes, and concrete grammar includes:

The AIR algorithm includes in the routing step of control plane:

One. network being carried out initialization, in three-layer network IP/SDH/WDM network is carried out initialization, is that the WDM layer is that the IP layer is followed successively by the 1st layer to the upper strata from lower floor, the 2nd layer, the 3rd layer, initialization path quality assessment parameter θ=θ ₀=0, θ ₀It is the value that is stored in the current best path quality critical parameter in the ant pheromones table.

Two. set up two tables among each node of each layer of three-layer network: a kind of is routing table, another kind is an ant pheromones table, ant agent is searched optimal path according to ant pheromones table in network, simultaneously optimal path is recorded in the routing table, because of the ant that keeping some in the network is also promptly searched carrying out routing always, so that the arrival of the routing of ant agent and service request is a synchronous process.Routing table is used to write down and connects each to the next-hop node information of the right path of sourcesink node at this node place; Ant pheromones table is used to write down the pheromones that ant agent discharges at this node place, and ant agent is to come search path according to ant pheromones table.By in network, remaining the ant of some, information according to the real-time renewal routing table of network state, because ant agent is at control plane, therefore can not influence the business transmission of datum plane, guarantee that the routing of ant agent and the arrival of service request are synchronous processes;

Search optimal path, random choose pair of source destination node is right in network, sends ant agent to search optimal path between it, and according to the quality of estimating the path with undefined path quality critical parameter:

0≤ζ≤1 wherein, σ is a Control Parameter, the path that has big more θ value is good more path.

(i, j): be a link in the network;

B _Ij: be link (i, bandwidth capacity j);

Br _Ij: be link (i, remaining bandwidth j);

R _Ij: R _Ij=Br _Ij/ B _IjBe this link (i, remaining bandwidth rate j);

R ^Min: the R that goes up all links for path L _IjIn minimum value;

H: be a jumping figure from source node s to the path L the destination node d.

The quality assessment parameter θ in the path of being searched and formerly be stored in path values θ in the above-mentioned ant pheromones table ₀Compare, if θ is greater than θ ₀, illustrate that then the path that is searched is better than the path in the routing table, this path is recorded in the original path of replacement in the routing table, and makes θ just become new θ ₀Be stored in the above-mentioned ant pheromones table, ant agent continues to search; If θ is less than or equal to θ ₀, then the path that searched of explanation is because the path in the routing table, abandons the path and the θ that are searched so, and ant agent continues to search, and these ant agent begin to advance from netinit the time, bring in constant renewal in, and stop during to all professional terminations.

The path quality critical parameter θ of definition has taken all factors into consideration many correlative factors, is the shortest path not only, also relates to link utilization.The present invention in other words is to be undertaken by ant agent in the pathfinding of control plane, not only take all factors into consideration many-sided situation of network, as network congestion situation, the grade of service etc., can also adapt to the situation of large-scale distributed network, can implement effectively network condition to be responded.

Three. decision node to the ant quantity of process whether reach the upper limit of ant population, be then to discharge ant agent; , then do not continue to send ant agent that routing information is upgraded.

Four. in the process that business transmits,, then date back in the node, utilize the information of the ant pheromones table among this node, reselect new next and jump path node, real-time recovery communication if node failure is arranged.If meet node failure, will date back to original point and re-route in prior art, in time, prolong, can not real time reaction, and the present invention only need turn back to and re-routes in the node, has shortened time delay, the communication of realization real-time recovery.

The step of interlayer combined optimization includes:

One). the l layer at three-layer network dynamically arrives a service request, l=1,2,3;

Two). whether at l layer have enough resource transmit this business, if enough resources are arranged, carry out step 5 if judging), if there are not enough resources, carry out step 3);

Three). judge whether the l layer is the 1st layer, if the 1st layer, refuse this business, if not the 1st layer, carry out step 4);

Four). should business be sent to the l-1 layer and transmit, and judge whether have enough resources to transmit this business, if enough resources are arranged at the l-1 layer by the interlayer link, carry out step 5), if there are not enough resources, refuse this business, transmission has reduced network blocking probability effectively like this;

Five). transmit this business according to the path of being write down in each node route list of this layer;

Six). business is sent to next-hop node, and judges whether this node is destination node, if destination node, then task termination is if not destination node carries out step 5).

In brief.Whether when the present invention arrives in the dynamic service request, at first judging at this layer has enough resource transmission should business, if enough resources are arranged, then according to the routing table in this node layer business to be transmitted; If do not have enough resources at this layer, then forwarding lower floor's network according to adjacent node annexation matrix between the different layers to by the interlayer link transmits, be sent to the destination node of former transport layer again by the interlayer link, if still do not have enough resource transmission should business at following one deck, then refuse this business.

The present invention separates the routing and the transmission of dynamic service, and routing is to carry out by ant agent at control plane, and transmitting then is to carry out at datum plane, and the processing that separates like this helps business is carried out real-time transmission, the real-time recovery in path etc.

Utilization of the present invention is carried out routing in the ant agent of control plane, and routing and professional transmission are separated.Take the strategy of combined optimization between layer and layer, the business that makes the upper strata get clogged can utilize the resource of lower floor's free time to transmit, thereby reaches the purpose of interlayer combined optimization.

The present invention can make full use of the idling-resource of each layer, thereby reach the purpose of network-wide optimization by the cooperative cooperating between heterogeneous networks layer and the layer.

Realization of the present invention also is: ant agent is searched optimal path in the routing step 2 of control plane, and concrete steps comprise:

1 in network random choose pair of source destination node to (s d), sends ant agent to search optimal path between it;

2 at each node place, and ant agent is specifically carried out according to following formula according to selecting next-hop node with information in the ant pheromones table and Path selection probability function:

$p_{\mathrm{ij}}^k=\frac{{\left[{\mathrm{\τ}}_{\mathrm{ij}}\right]}^{\mathrm{\α}}{\left[R_{\mathrm{ij}}\right]}^{\mathrm{\β}}}{\underset{{l\∈\mathrm{neighbor}}_i^k}{\mathrm{\Σ}}{\left[{\mathrm{\τ}}_{\mathrm{il}}\right]}^{\mathrm{\α}}{\left[R_{\mathrm{il}}\right]}^{\mathrm{\β}}}---\left(2\right)$

3. choose after the next-hop node, according to following formula

τ _ij←(1-δ)τ _ij+δτ ₀ (3)

Renewal local message element, ant agent is gone to hop node then: wherein, δ is the plain factor of evaporation of local message, satisfies 0＜δ＜1, τ ₀It is the initial value of pheromones;

4. judge whether present node is destination node,,, carry out step 2 if present node is not a destination node if carrying out next step is step 5;

5. be destination node as if present node, calculating path quality assessment parameter θ, and compare θ and θ ₀If θ is greater than θ ₀, make θ ₀=θ also carry out step 6.; If θ is less than or equal to θ ₀, carry out step 7.

6. send reverse ant agent to source node from destination node, the routing table of upgrading the node that pass through on the way with also upgrade ant pheromones table as follows:

${\mathrm{\τ}}_{\mathrm{ij}}\←(1-\mathrm{\ρ}){\mathrm{\τ}}_{\mathrm{ij}}+\mathrm{\ρ\Δ}{\mathrm{\τ}}_{\mathrm{ij}}^{\mathrm{bs}},$ $\∀(i,j)\∈T^{\mathrm{bs}}$

Wherein, T wherein ^BsThe expression optimal path;

Be the pheromones increment, H ^BsIt is the jumping figure of optimal path; Parameter ρ is the plain factor of evaporation of global information.

7. judge whether the number of the ant agent of being sent reaches the upper limit of this ant population ant agent number, if reach, carrying out next step is step 8, if do not reach, then carry out step 1;

8. stop to discharge ant agent, search the cycle up to the next one.

The present invention is to be undertaken by ant agent in the pathfinding of control plane, not only take all factors into consideration many-sided situation of network, as network congestion situation, the grade of service etc., can also adapt to the situation of large-scale distributed network, can implement effectively network condition to be responded.

Though can obtain certain effect because the method for the static optimization in the prior art is optimized dynamic service, can not be applicable to dynamic service fully, can't utilize Internet resources to greatest extent; When carrying out heavy-route, can only select in two nodes in any one two tuple of getting rid of information centering at most; Though and the mechanism of recalling searching that heavy-route adopts can find new route, spended time is long, if node is too much in the route, time delay will increase greatly, can't handle problems such as a lot of real time business.The present invention is directed to above problem, three-layer network is carried out joint optimization of resources, whole optimization is divided into two parts: i.e. resource optimization and interlayer resource optimization in the layer, in each layer in three-layer network, employing is based on the comprehensive method for routing AIR of ant agent, carry out this layer at interlayer whether the judgement that enough resource transmission should business is arranged, and connect by link between the interlayer node, the business that the realization upper strata gets clogged can utilize the resource of lower floor's free time to transmit, cooperative cooperating between layer and the layer, make full use of the idling-resource of each layer, reached the combined optimization of the whole network.

The present invention separates routing and professional transmission.Ant agent is at control plane, does not influence the business transmission of datum plane, can be grasped obtaining optimal path the network integral status, and can carry out real-time routing to the business of each arrival, improves the resource utilization of network.

The present invention is in the process that business transmits, if node failure is arranged, then date back in the node, utilize the information of the ant pheromones table among this node, reselect new next and jump path node, with respect to the initial point that dates back to of prior art, step and time have been saved, real-time high-efficiency network condition is responded, real-time recovery communication reduces the reject rate of dynamic service request greatly.

Explore good innovation because the present invention constantly sums up in the process of research joint optimization of resources in three-layered network, summarize the path quality critical parameter

The expression formula of probability of expression, ant agent access node:

And adjacent node annexation matrix between the different layers, successfully solve routing, transmitted, dated back to concrete technical problems such as a node, real-time recovery communication, realized resource-sharing in the net, the idling-resource of each layer makes full use of, the purpose of network-wide optimization, need not improve, the present invention can be directly used in the multitiered network combined optimization more than three layers.

Description of drawings:

Fig. 1 is that AIR algorithm among the present invention is at the flow chart of the routing of control plane;

Fig. 2 is the flow chart of the interlayer combined optimization among the present invention;

Fig. 3 is a network hierarchy structural representation of the present invention;

Fig. 4 is the topological diagram of the three-layer network IP/SDH/WDM that emulation is adopted among the present invention;

Fig. 5 is multilayer optical-fiber network method for joint optimization of resources of the present invention and additive method service blocking rate comparative graph relatively.

Embodiment:

For making the purpose, technical solutions and advantages of the present invention clearer, embodiment of the present invention is described further in detail below in conjunction with accompanying drawing.

Embodiment 1:

Referring to Fig. 1, the present invention is a kind of method for joint optimization of resources in three-layered network, is to realize on three-layer network, in the 1st layer of optical-fiber network, be with optical transmission device, optical cross-connection equipment (OXCs) as network node, connect some network node with optical fiber.In the 2nd layer of circuit network, be with circuit transmission equipment, digital cross connect equipment (DXCs) as network node, connect some network node with transmission circuit.In the 3rd layer of packet network, be with network equipments such as router, switches as network node, connect some network node with transmission link.

This method for joint optimization of resources in three-layered network is divided into two parts: resource optimization and interlayer resource optimization in the layer, in each layer, employing utilizes and carries out routing in the ant agent of control plane based on the comprehensive method for routing AIR (Ant agent-based integrated routingalgorithm) of ant agent; Take combined optimization between layer and layer, the service request that the upper strata is got clogged can utilize the resource of lower floor's free time to transmit, and reaches the interlayer combined optimization.The present invention separates the routing and the service request transmission of dynamic service, and the routing of ant agent and the arrival of service request are synchronous processes.Concrete grammar includes:

The AIR algorithm includes in the routing step of control plane:

Network being carried out initialization, in three-layer network IP//SDH/WDM network is carried out initialization, among the three-layer network IP//SDH/WDM, is that the WDM layer is that the IP layer is followed successively by the 1st layer to the upper strata from lower floor, the 2nd layer, and the 3rd layer.Initialization path quality assessment parameter θ=θ ₀=0, θ ₀It is the value that is stored in the current best path quality critical parameter in the ant pheromones table.

Two. set up two tables among each node of each layer of three-layer network: a kind of is routing table, is used to write down connect each to the next-hop node information of the right path of sourcesink node at this node place; Another kind is an ant pheromones table, is used to write down the pheromones that ant agent discharges at this node place.Begin to search optimal path simultaneously, random choose pair of source destination node is right in network, sends ant agent to search optimal path between it, and according to the quality of estimating the path with undefined path quality critical parameter:

Wherein 0≤ζ≤1 in the present embodiment, makes ζ=0.6; σ is a Control Parameter, is set at σ=1/15 here.And think that in the method the path that has big more θ value is good more path.

For above-mentioned formula is made an explanation, provide to give a definition:

(i, j): be a link in the network;

B _Ij: be link (i, bandwidth capacity j);

Br _Ij: be link (i, remaining bandwidth j);

R _Ij: R _Ij=Br _Ij/ B _IjBe this link (i, remaining bandwidth rate j);

R ^Min: the R that goes up all links for path L _IjIn minimum value;

H: be a jumping figure from source node s to the path L the destination node d.

The quality assessment parameter θ in the path of being searched and formerly be stored in path values θ in the above-mentioned ant pheromones table ₀Compare, if θ is greater than θ ₀, illustrate that then the path that is searched is better than the path in the routing table, this path is recorded in the original path of replacement in the routing table, and makes θ just become new θ ₀Be stored in the above-mentioned ant pheromones table, ant agent continues to search; If θ is less than or equal to θ ₀Illustrate that then the path that is searched is not because the path in the routing table, abandon the path and the θ that are searched so, ant agent continues to search, these ant agent begin to advance from netinit the time, constantly routing table and ant pheromones table are upgraded, stopped during to all professional terminations.

Three. decision node to the ant quantity of process whether reach the upper limit of ant population, be, then stop to discharge ant agent, not, then continue to send ant agent that routing information is upgraded.

Four. in the process that business transmits,, then date back in the node, utilize the information of the ant pheromones table among this node, reselect new next and jump path node, real-time recovery communication if node failure is arranged;

Referring to Fig. 2, key point of the present invention also is:

Carry out combined optimization between the heterogeneous networks layer, its step includes:

One). the l layer at three-layer network dynamically arrives a service request, l=2 in the present embodiment;

Two). whether at l layer have enough resource transmit this business, if enough resources are arranged, carry out step 5 if judging), if there are not enough resources, carry out step 3);

Three). judge whether the l layer is the 1st layer, if the 1st layer, then illustrating does not have lower floor's network under this layer network, refuse this business, if not the 1st layer promptly the 2nd layer, then illustrating has lower floor's network under this layer network, carry out step 4);

Four). should business be sent to the l-1 layer by the interlayer link, promptly the 1st layer is transmitted, and participates in Fig. 3, and judge that whether be the 1st layer at the l-1 layer has enough resources to transmit this business, if enough resources are arranged, carries out step 5), if there are not enough resources, refuse this business;

Five). transmit this business according to the path of being write down in each node route list of this layer;

Six). business is sent to next-hop node, and judges whether this node is destination node, if destination node, then task termination is if not destination node carries out step 5).

The present invention certain to node between in dynamic service request when transmission, arranged, whether at first judge at this layer has enough resource transmission should business, if enough resources are arranged, then according to the routing table in this node layer business to be transmitted; If do not have enough resources at this layer, then forwarding lower floor's network according to adjacent node annexation matrix between the different layers to by the interlayer link transmits, be sent to the destination node of former transport layer again by the interlayer link, if still do not have enough resource transmission should business at lower floor's network, then refuse this business.By the interlayer combined optimization, can share the Internet resources of heterogeneous networks layer, thereby reduce the blocking rate of whole three-layer network.The resource-sharing of cooperative cooperating reaches the combined optimization of the multitiered network of real-time recovery communication between the search binder course of the interior optimal path of layer.

Embodiment 2:

Method for joint optimization of resources in three-layered network is with embodiment 1, and referring to Fig. 1 and Fig. 2, ant agent is searched optimal path in the routing step 2 of control plane, and concrete steps comprise:

1 in network random choose pair of source destination node to (s d), sends ant agent to search optimal path between it;

2 at each node place, and ant agent is selected next-hop node according to information in the ant pheromones table and path routing probability function, specifically carries out according to following formula:

$p_{\mathrm{ij}}^k=\frac{{\left[{\mathrm{\τ}}_{\mathrm{ij}}\right]}^{\mathrm{\α}}{\left[R_{\mathrm{ij}}\right]}^{\mathrm{\β}}}{\underset{{l\∈\mathrm{neighbor}}_i^k}{\mathrm{\Σ}}{\left[{\mathrm{\τ}}_{\mathrm{il}}\right]}^{\mathrm{\α}}{\left[R_{\mathrm{il}}\right]}^{\mathrm{\β}}}---\left(2\right)$

3. choose after the next-hop node j, according to following formula

τ _ij←(1-δ)τ _ij+δτ ₀ (3)

Renewal local message element, ant agent k goes to hop node then: wherein, δ is the plain factor of evaporation of local message, satisfies 0＜δ＜1, gets δ=0.8 here, τ ₀Be the initial value of pheromones, value is τ ₀=1;

4. judge whether present node is destination node d,,, carry out step 2 if present node i is not a destination node if carrying out next step is step 5;

5. judge whether present node is destination node, if present node i is a destination node, calculating path quality assessment parameter θ, and compare θ and θ ₀If θ is greater than θ ₀, make θ ₀=θ also carry out step 6.; If θ is less than or equal to θ ₀, carry out step 7.

6. send reverse ant agent k to source node from destination node, the routing table of upgrading the node that pass through on the way with also upgrade ant pheromones table as follows:

${\mathrm{\τ}}_{\mathrm{ij}}\←(1-\mathrm{\ρ}){\mathrm{\τ}}_{\mathrm{ij}}+\mathrm{\ρ\Δ}{\mathrm{\τ}}_{\mathrm{ij}}^{\mathrm{bs}},$ $\∀(i,j)\∈T^{\mathrm{bs}}---\left(4\right)$

T in the formula ^BsThe expression optimal path;

Be the pheromones increment, H ^BsIt is the jumping figure of optimal path; Parameter ρ is the plain factor of evaporation of global information, and value is ρ=0.7.

7. judge whether the number of the ant agent k that is sent reaches the upper limit of this ant population ant agent number, and in the present embodiment, ant population size is 100.If reach, carrying out next step is step 8., if do not reach, then carry out step 1;

8. stop to discharge ant agent k, up to the next search cycle.

The present invention is by the setting of path quality critical parameter, can search the optimal path in the layer efficiently, and, can network condition be responded real-time and effective according to ant group method self-organizing and distributed characteristics, realize professional real-time transmission, the real-time recovery communication in path.

Embodiment 3:

Method for joint optimization of resources in three-layered network is with embodiment 2, referring to Fig. 3,

Represent the set of network nodes of each layer respectively with following mathematic(al) representation, with the adjacency matrix of adjacent node in one deck, adjacent node annexation matrix between the different layers.

The set of network nodes of l layer is: G ^l=(V ^l, E ^l), V wherein ^lRepresent l layer network node set, E ^lRepresent the internodal adjacent link set of l layer network, l=1,2,3.For example, l=1 represents physics optical-fiber network layer, l=2 indication circuit layer, and l=3 represents packet-based IP network layer.

The adjacency matrix of adjacent node is N in the l layer ^l, same, l=1,2,3.In Fig. 3 with N ²Be that l=2 is an example:

In the superincumbent matrix, N ²Represent that this is in layer 2 network, A2～E2 is a network node, is matrix element in the braces.Element is to have link to exist between 1 these two nodes of expression in the matrix, and element is not have link to exist between 0 these two nodes of expression.

Adjacent node annexation matrix is N between the different layers ¹²And N ²³N ¹²Adjacent node annexation matrix between the expression layers 1 and 2, N ²³Adjacent node annexation matrix between the expression layers 2 and 3.In Fig. 3 with N ¹²Be example:

In the matrix element be 1 the expression these two different layers node between the interlayer link is arranged, element be 0 the expression these two different layers node between do not have the interlayer link.

Need to prove, in the 1st layer of optical-fiber network, be with optical transmission device, optical cross-connection equipment (OXCs) as network node, connect some network node with optical fiber.In the 2nd layer of circuit network, be with circuit transmission equipment, digital cross connect equipment (DXCs) as network node, connect some network node with transmission circuit.In the 3rd layer of packet network, be with network equipments such as router, switches as network node, connect some network node with transmission equipment.

In the 1st step, network is carried out initialization, and set up two kinds of tables in each node of each layer: a kind of is routing table, is used to write down connect each to the next-hop node information of the right path of sourcesink node at this node place; Another kind is an ant pheromones table, is used to write down the pheromones that ant agent discharges at this node place.

These ant agent are according to AIR algorithm operation, search different nodes between optimal path, and this path is recorded in the routing table of each node.Begin to advance when it should be noted that these ant agent, stop during to all professional terminations from netinit.The pathfinding of ant agent and the transmission of service request are synchronous processes, and because ant agent is at control plane, therefore can not influence the business transmission of datum plane.So this algorithm can be grasped obtaining optimal path the network integral status, and can the business of each arrival be transmitted in real time.

The 2nd step on each layer network, dynamically arrived service request, and the arrival rate of this service request is obeyed Poisson distribution, the duration obeys index distribution of service connection request.Referring to Fig. 3, be example with layer 2 network among Fig. 3.The adjacency matrix of this layer is N ², the 2nd layer of adjacent node annexation matrix with the 1st layer is N ²³If at node to (C2 has service request between E2), and whether then at first judge at this layer has enough resource transmission should business, if enough resources are arranged, then according to the routing table in this node layer business to be transmitted; If do not have enough resources, be N then according to adjacent node annexation matrix between the different layers at this layer ¹²(C2 as can be known, E2) be (A1 at the 1st layer mapping node, F1), so should business from the 2nd layer by interlayer link C2～A1 forward to the 1st layer (A1 transmits between F1), is sent to destination node E2 by interlayer link F1～E2 again, if still do not have enough resource transmission should business at the 1st layer, then refuse this business.

The 3rd step, in the process that business transmits,, then date back in the node if node failure is arranged, utilize the information of the ant pheromones table among this node, reselect new next according to formula (2) and jump path node, real-time recovery communication.

The present invention has fully taken all factors into consideration the jumping figure in link utilization and path in layer, and utilize ant agent to carry out real-time routing, thereby can avoid the link that those are frequently used, make congested that these crucial links can be not too early, there is the jumping figure of having controlled the path to make this path can not take too many Internet resources simultaneously, more consider, thereby reduce network blocking probability for follow-up business.Between layer and layer, utilized the idling-resource of each layer of network fully, the business that makes upper layer network get clogged can be utilized the idling-resource of lower floor's network, reach the interlayer resource-sharing, thereby reduce the blocking rate of whole three-layer network, improve the resource utilization of network, make Internet resources be able to combined optimization.

Embodiment 4:

The validity based on the comprehensive method for routing of multilayer of ant agent for check the present invention proposes adopts Computer Simulation to verify, as Fig. 4 and Fig. 5.Emulation topological diagram 4 is three-layer network IP/SDH/WDM of U.S. MCI net, and imposing a condition of emulation experiment of the present invention is as follows:

1) the maximum configured bandwidth of network IP layer (the 3rd layer) link is 622Mbit/s, the maximum configured bandwidth of SDH layer (the 2nd layer) link is 2.5Gbit/s, 8 wavelength of WDM layer (the 1st layer) configuration, the bandwidth of each wavelength is 2.5Gbit/s, and each node of WDM layer all has the long transfer capability of all-wave.Being configured to of interlayer link bandwidth: the 3rd layer to the 2nd layer interlayer link bandwidth is 622Mbit/s, and the 2nd layer to the 1st layer interlayer link bandwidth is 2.5Gbit/s.Wherein the IP layer has 19 nodes and 32 in abutting connection with link, and the SDH layer has 22 nodes and 47 in abutting connection with link, and the WDM layer has 27 nodes and 56 in abutting connection with link.

2) each dynamic service is equiprobable produces in three layers at random, and its sourcesink node produces with equiprobability in this layer equally at random, all sourcesink nodes of each network layer to traffic intensity all identical.

3) arrival of service connection request is Poisson process, and it is the Poisson distribution of λ that arrival rate is obeyed average, and it is the exponential distribution of 1/ μ that the duration of service connection request is obeyed average, and offered load is λ/μ Erlangs.

4) system does not have waiting list, in case connection request is rejected, then abandons immediately.

5) each emulation produces 100000 service connection request, and result data is the assembly average of simulation run 100 times.

Fig. 5 is multilayer optical-fiber network method for joint optimization of resources of the present invention and additive method service blocking rate comparative graph relatively, and the resulting experimental data of Fig. 5 is an emulation gained under Fig. 4 emulation topology, and simulated conditions as mentioned above.Referring to Fig. 5, the improvement for three layers of joint route algorithm performance relatively the present invention is based on ant agent compares the present invention and other several modes.Wherein Mode1 is illustrated in the shortest path routing algorithm that adopts in each layer, does not have the strategy of combined optimization between network layer and layer.It is the AIR routing algorithm that Mode2 is illustrated in each layer what adopt, network layer with layer between do not adopt the strategy of combined optimization.What Mode3 adopted in each layer is the AIR routing algorithm, adopts the strategy of combined optimization simultaneously between network layer and layer, method for joint optimization of resources in three-layered network promptly proposed by the invention.As can be seen from Figure 5, increase along with traffic intensity, Mode2 has the blocking rate lower than Mode1, this be because, in each layer, the AIR routing policy has more performance than shortest path algorithm, because it has considered the resource distribution situation of network, make professional transmission always on those have the link of lower link utilization, transmit, thereby can utilize the Internet resources of each layer more fully.Relatively the curve of Mode2 and Mode1 illustrates that the AIR algorithm is effectively in individual layer, can more effective reduction network blocking probability with respect to other algorithms.Simultaneously, can also see, Mode3 is that whole technical scheme of the present invention has minimum blocking rate, when traffic intensity is 90 Erlangs (Ireland), the blocking rate of Mode3 is 0.17, be far smaller than the blocking rate 0.83 of Mode1 under the identical services intensity and the blocking rate 0.68 of Mode2, this is because used the AIR algorithm in layer, adopted interlayer joint optimization of resources strategy at interlayer, make the business of upper layer network can use the idling-resource of lower floor's network, thereby reach the purpose that Internet resources are shared between layer and the layer, the present invention can be directly used in the multitiered network combined optimization more than three layers.

Claims (2)

1. method for joint optimization of resources in three-layered network, it is characterized in that: this method is divided into two parts i.e. interior resource optimization of layer and interlayer resource optimization, in each layer, the routing and the service request transmission of dynamic service are separated, employing utilizes and carries out routing in the ant agent of control plane based on the comprehensive method for routing AIR of ant agent; Take combined optimization between layer and layer, the service request that the upper strata is got clogged can utilize the resource of lower floor's free time to transmit, and reaches the interlayer combined optimization, and the routing of ant agent and the transmission of service request are synchronous processes, and concrete grammar includes:

The AIR algorithm includes in the routing step of control plane:

(1). network is carried out initialization, in three-layer network IP/SDH/WDM, network is carried out initialization, initialization path quality assessment parameter θ=θ ₀=0;

(2). set up two tables among each node of each layer of multitiered network: a kind of is routing table, another kind is an ant pheromones table, routing table be used for storing the sourcesink node between routing information, ant pheromones table, be used to write down the pheromones that ant agent discharges at this node place, ant agent is to come search path according to ant pheromones table, by in network, remaining the ant of some, according to the information of the real-time renewal routing table of network state, guarantee that the routing of ant agent and the arrival of service request are synchronous processes;

The search of optimal path is that random choose pair of source destination node is right in network, sends ant agent to search optimal path between it, and according to the quality of estimating the path with undefined path quality critical parameter:

$\mathrm{\θ}={\mathrm{\ζe}}^{R^{\mathrm{mn}}-1}+(1-\mathrm{\ζ})e^{-\mathrm{\σH}}$

The quality assessment parameter θ in the path of being searched and formerly be stored in θ in the above-mentioned ant pheromones table ₀Compare, if θ is greater than θ ₀, make θ become new θ ₀Be stored in above-mentioned two tables, if θ is less than or equal to θ ₀, abandon the path and the θ that are searched;

(3). decision node to the ant quantity of process whether reach the upper limit of ant population, be, then stop to discharge ant agent, not, then continue to send ant agent routing information is searched and to be upgraded;

(4). in the process that business transmits,, then date back in the node, utilize the information of the ant pheromones table among this node, reselect new next and jump path node, real-time recovery communication if node failure is arranged;

The step of interlayer combined optimization includes:

1). the l layer at three-layer network dynamically arrives a service request;

2). whether at l layer have enough resource transmit this business, if enough resources are arranged, carry out step 5) if judging, if there are not enough resources, carry out step 3);

3). judge whether the l layer is the 1st layer, if the 1st layer, refuse this business, if not the 1st layer, carry out step 4);

4). should business be sent to the l-1 layer and transmit, and judge whether have enough resources to transmit this business,, carry out step 5),, refuse this business if there are not enough resources if enough resources are arranged at the l-1 layer by the interlayer link;

5). transmit this business according to the path of being write down in each node route list of this layer;

6). business is sent to next-hop node, and judges whether this node is destination node, if destination node, then task termination is if not destination node carries out step 5).

2. multitiered network method for joint optimization of resources according to claim 1 is characterized in that: ant agent is searched optimal path in the routing step (2) of described control plane, and detailed process comprises:

(2a). random choose pair of source destination node is right in network, sends ant agent to search optimal path between it;

(2b). at each node place, ant agent is specifically carried out according to following formula according to selecting next-hop node with information in the ant pheromones table and Path selection probability function:

$p_{\mathrm{ij}}^k=\frac{{\left[{\mathrm{\τ}}_{\mathrm{ij}}\right]}^{\mathrm{\α}}{\left[R_{\mathrm{ij}}\right]}^{\mathrm{\β}}}{\underset{l\∈\mathrm{neighbo}{r}_i^k}{\mathrm{\Σ}}{\left[{\mathrm{\τ}}_{\mathrm{il}}\right]}^{\mathrm{\α}}{\left[R_{\mathrm{il}}\right]}^{\mathrm{\β}}}$

(2c). choose after the next-hop node, according to following formula

τ _ij←(1-δ)τ _ij+δτ ₀

Upgrade local ant pheromones, ant agent is gone to hop node then: wherein, δ is the plain factor of evaporation of local message, satisfies 0＜δ＜1, τ ₀It is the initial value of pheromones;

(2d). judge whether present node is destination node,,, carry out step (2b) if present node is not a destination node if carrying out next step is step (2e);

(2e). if present node is a destination node, calculating path quality assessment parameter θ, and compare θ and θ ₀If θ is greater than θ ₀, make θ ₀=θ also carries out step (2f); If θ is less than or equal to θ ₀, carry out step (2g);

(2f). send reverse ant agent to source node from destination node, the routing table of upgrading the node that pass through on the way with also upgrade ant pheromones table as follows:

${\mathrm{\τ}}_{\mathrm{ij}}\←(1-\mathrm{\ρ}){\mathrm{\τ}}_{\mathrm{ij}}+\mathrm{\ρ}{\mathrm{\Δ\τ}}_{\mathrm{ij}}^{\mathrm{bs}},\∀(i,j)\∈T^{\mathrm{bs}}$

Wherein, T wherein ^BsThe expression optimal path;

Be the pheromones increment, H ^BsIt is the jumping figure of optimal path; Parameter ρ is the plain factor of evaporation of global information;

(2g). judge whether the number of the ant agent of being sent reaches the upper limit of this ant population ant agent number, if reach, carrying out next step is step (2h), if do not reach, then carries out step (2a);

(2h). stop to discharge ant agent, search the cycle up to the next one.

Images