CN102025621B - Routing selection method of cognitive network based on mapping mechanism - Google Patents

Abstract

The invention discloses a routing selection method of a cognitive network based on a mapping mechanism, which is divided into two parts, namely discovering network topology and defining link cost. The method is as follows: discovering a network topology structure by a wincap probe packet sniffer; on the basis of multi-plane parameter mapping, defining different link costs for different services of different users; and updating the link costs in real time according to the network states, thus realizing adaptive routing selection. In the routing selection method, the implementation is simple, the control can be flexible, the repeated links can be realized by adjusting the link cost instead of establishing a plurality of routing tables, and the work for a cognitive network router is reduced.

Description

Cognition network route selection method based on mapping mechanism

Technical field

The present invention is directed to the characteristics of cognition network heavy-route, the different business different user is carried out different Route Selection, simultaneously, at the real-time change of service condition in the network, the selection of cognitive route also will be carried out real-time update.Adopt Puli's nurse algorithm, on the basis of Topology Discovery, dynamic real-time is adjusted the link weight, thereby realizes cognitive routing policy.The invention belongs to the technical field of cognition network route selection method.

Background technology

How to realize in network that Route Selection is a focus of studying at present.Relatively agreement is such as RIP, OSPF in the Chang Yong territory, and inter-domain protocol all is extensive use of in network such as EIGRP etc.In many researchs, as the research basis, the route agreement various improvement have been carried out with graph theory.

Yet in the legacy network, the Route Selection of packet is selected next jumping according to routing table in the router, though this method can be delivered to the destination to packet, has some problems at present.At first, traditional routing table can not realize in the network packet of different user different business being adopted the policy selection of different routes, but adopts unified routing table to transmit; Secondly, when Network was busy, traditional routing table adopted unified routing policy, may cause some link busy, and the link that has is idle (though also can arrive the destination by this link forwarding) then; At last, when Link State changed, cognition network thought that network should find in real time that this changes, thereby adopts corresponding strategy to realize new Route Selection, and traditional route can not be accomplished this point obviously.Cognitive routing policy can be realized the routing policy selection of different user different business by the parameter mapping, adjusts Route Selection in real time by upgrading link cost, realizes redirection of router.

Summary of the invention

Technical problem: the objective of the invention is to propose a kind of cognition network route selection method based on mapping mechanism, by the Dynamic Selection link cost, realize being redirected of cognitive route, and then guarantee the QoS of Network.

Technical scheme:

The present invention is based on the cognition network route selection method of mapping mechanism, it is characterized in that on the basis of Topology Discovery, the weight of link is assessed to every day, and utilize Puli's nurse algorithm to set up new routing table and carry out routing forwarding, concrete grammar is as follows:

I) each domain server is selected detection network packet loss, time delay, bandwidth and throughput, when network packet loss rate＜d, time delay＜t, shake＜p, throughput＞m (d, t, p, m are the good threshold values of formulating when initial of expression network state), the expression network state is good, ignore weight this moment, and directly adopt routing table to transmit route, and do not need route is selected;

Ii) when detecting in the network parameter, do not satisfy when setting needs some domain servers, territory cognitive services device begins to start the cognitive Decision function, namely according to the weight of every the link of problem setting network that occurs in the network: when some domain server D0 in the network detect this territory network parameter and do not satisfy the qos parameter needs, distribute different link weights according to the different business different user;

Iii) survey data packet traffic and user type through this domain server D0, the weight of the bar link that domain server D0 links to each other with domain server D1, D2, D3 is respectively w01, w02, w03, and w01+w02+w03=100 is that all link weighted values that link to each other of any domain server are 100; Strengthen its link weight gradually for the link that does not satisfy qos requirement, reduce the weight with this domain server phase connected link simultaneously; For the different business different user, according to mapping mechanism, compare professional priority, the priority with them sorts from high to low successively, each when different business when not satisfying the domain server of QoS demand, next jumps direction weight increasing priority number in the present routing table of this packet;

Iv) regenerate minimum cost spanning tree according to Puli's nurse algorithm, seek next hop address and transmit packet;

V) continue to survey this domain server QoS network parameter, when satisfying the QoS needs again, stop using Puli's nurse algorithm to select route, ignore the network weight, select next jumping in the routing table;

The link QoS weight that links to each other with this domain server is distributed, and concrete allocation step is as follows:

A) domain server is collected packet loss d1, time delay t1, bandwidth p1 and the throughput m1 of every link being attached thereto at control plane, compares with the target qos parameter that has proposed, and obtains the health degree of every link:

$Q=\underset{T1}{\overset{T2}{\∫}}[(d1-d)+(t1-t)+(p-p1)+(m-m1)]\mathrm{dt};$

B) according to health degree Q1, Q2, the Q3...Qn of every link, can obtain the value of w01, w02, w03...w0n, concrete computational methods are as follows:

By that analogy,

Wherein n is number of links, and t represents the time.

Beneficial effect: the proposition of cognition network QoS Route Selection, we can realize:

1) the different business QoS demand of different user is classified, make cognition network can require to select different routes according to different levels user's difference;

2) definite link cost method based on the mapping of many plane parameters has been proposed, for further cognitive Route Selection lays the foundation;

3) propose real-time change according to network state, dynamically updated the method for link cost, realized the cognitive strategy of route.

Description of drawings

Fig. 1 is network portion domain server topological structure.Provided the continuous situation of part domain server link among the figure.

Fig. 2 is tactful plane Route Selection.Provide tactful plane among the figure and carried out the concrete steps of Route Selection according to user plane, service plane and control plane parameter.

Fig. 3 is domain server link weight redistribution method.Provided the link weight allocation step of domain server when the QoS demand does not satisfy among the figure.

Embodiment

Be elaborated below in conjunction with the technical scheme of accompanying drawing to invention:

The present invention proposes a kind of cognition network routing strategy based on mapping, cognitive agency mainly is divided into control plane, tactful plane, service plane and user plane.Based on the winpcap open source software, we have done a probe collection network parameter, construct the network topology full figure at control plane.User plane is determined the customer sla grade, and service plane determines to enter the class of business of network, and the present invention mainly is that the parameter to control plane, service plane and user plane is carried out information interaction on tactful plane, and then generates routing policy.

Source end and destination ip address that cognitive domain server is gathered according to the control plane probe make up network topology structure in the territory, carry out between each domain server then alternately, thereby obtain full mesh topology.Here hypothetical network part domain server topological structure as shown in Figure 1.

At user plane, when a business enters in the network a certain node, can know user gradation according to ip address, source.Can understand to belong to which kind of business respectively at service plane.The parameter that the strategy plane provides according to user plane, service plane and control plane, specifically select the route step as follows:

I) each domain server is selected detection network packet loss, time delay, bandwidth and throughput, as network packet loss rate＜d, time delay＜t, shake＜p is during throughput＞m (d, t, p, m are the good threshold values of formulating when initial of expression network state), the expression network state is good, ignore weight this moment, and directly adopt routing table to transmit route, and do not need route is selected, thereby the minimizing network delay alleviates the burden of cognitive domain server;

Ii) do not satisfy when setting needs when some domain servers detect in the network parameter, territory cognitive services device begins to start the cognitive Decision function, namely according to the weight of every the link of problem setting network that occurs in the network.When some domain server D0 in the network detect this territory network parameter and do not satisfy the qos parameter needs, distribute different link weights according to the different business different user;

Iii) survey data packet traffic and user type through this domain server D0, by topological structure as can be known, this domain server links to each other with domain server D1, D2, D3, article three, the weight of link is respectively w01, w02, w03, w01+w02+w03=100 (being that all link weighted values that link to each other of any domain server are 100), for the link that does not satisfy qos requirement, strengthen its link weight gradually, reduce other and the weight of this domain server phase connected link simultaneously.For the different business different user, according to mapping mechanism, we can compare professional priority, successively their priority is classified as from high to low: priority 0, priority 1 ... priority 11, at every turn when different business when not satisfying the domain server of QoS demand, just next is jumped direction weight and strengthens priority number (be 0 such as service priority, then weight strengthens 0, by that analogy) in the present routing table of this packet for we;

Iv) regenerate minimum cost spanning tree according to Puli's nurse algorithm, seek next hop address and transmit packet;

V) continue to survey this domain server QoS network parameter, when satisfying the QoS needs again, stop using Puli's nurse algorithm to select route, ignore the network weight, select next jumping of Route Selection in the routing table.

In step I ii, can see that initial link circuit weight w01, w02, w03 are certain, concrete determining step is as follows:

A) domain server is collected packet loss d1, time delay t1, bandwidth p1 and the throughput m1 of every link being attached thereto at control plane, compares with the target qos parameter that has proposed, according to formula

Can obtain the health degree of every link;

B) according to the value of Q1, Q2, Q3...Qn, can obtain the value of w01, w02, w03...w0n, concrete computing formula is as follows:

... by that analogy, $w0n=100*\frac{\mathrm{Qn}}{Q1+Q2+...+\mathrm{Qn}}.$

Claims (1)

1. the cognition network route selection method based on mapping mechanism is characterized in that on the basis of Topology Discovery the weight of every link being assessed, and utilizes Puli's nurse algorithm to set up new routing table and carry out routing forwarding, and concrete grammar is as follows:

I) each domain server is selected detection network packet loss, time delay, bandwidth and throughput, when network packet loss rate＜d, time delay＜a, shake＜p, throughput＞m, the expression network state is good, ignore weight this moment, and directly adopt routing table to transmit route, and do not need route is selected, d, a, p, m are the good threshold values of formulating when initial of expression network state;

Ii) when detecting in the network parameter, do not satisfy when setting needs some domain servers, territory cognitive services device begins to start the cognitive Decision function, namely according to the weight of every the link of problem setting network that occurs in the network: when some domain server D0 in the network detect this territory network parameter and do not satisfy the qos parameter needs, distribute different link weights according to the different business different user;

Iii) survey data packet traffic and user type through this domain server D0, the weight of the bar link that domain server D0 links to each other with domain server D1, D2, D3 is respectively w01, w02, w03, and w01+w02+w03=100 is that all link weighted values that link to each other of any domain server are 100; Strengthen its link weight, the weight that reduces to connect D0 and other servers simultaneously and satisfy the link of qos requirement gradually for the link that does not satisfy qos requirement; For the different business different user, according to mapping mechanism, compare professional priority, the priority with them sorts from high to low successively, each when different business when not satisfying the domain server of QoS demand, next jumps direction weight increasing priority number in the present routing table of this packet;

Iv) regenerate minimum cost spanning tree according to Puli's nurse algorithm, seek next hop address and transmit packet;

V) continue to survey this domain server QoS network parameter, when satisfying the QoS needs again, stop using Puli's nurse algorithm to select route, ignore the link weight in Puli's nurse algorithm, select next jumping in the routing table;

The link QoS weight that links to each other with this domain server is distributed, and concrete allocation step is as follows:

A) domain server is collected packet loss d1, time delay t1, bandwidth p1 and the throughput m1 of every link being attached thereto at control plane, is that described threshold value is compared with the target qos parameter that has proposed, and obtains the health degree of every link:

$Q=\underset{T1}{\overset{T2}{\∫}}[(d1-d)+(t1-a)+(p-p1)+(m-m1)]\mathrm{dt};$

B) according to health degree Q1, Q2, the Q3...Qn of every link, can obtain the value of w01, w02, w03...w0n, concrete computational methods are as follows: $w01=100*\frac{Q1}{Q1+Q2+...+\mathrm{Qn}},$ $w02=100*\frac{Q2}{Q1+Q2+...+\mathrm{Qn}},$ ... by that analogy,

Wherein n is number of links, and t represents the time, and T1 is the zero-time that threshold value is compared, and T2 is the concluding time that threshold value is compared.

Images