CN101291255B - Heuristic fault positioning method for network of next generation - Google Patents

Abstract

The invention relates to a heuristic fault positioning method for a next generation network. The method comprises the following steps that: firstly, observable signs in a network system are assembled together; one or a plurality of faults which are most related to the observable signs are found out; secondly, according to the following phenomenon that the faults just meet to lead and generate all the signs of an observable sign assembly, the faults are combined to a fault positioning assembly; finally, the faults are judged and positioned by the fault positioning assembly. The method can consider the difficulty and application demand brought by 3G, the next generation network and other new networks, meet the need of real-time fault positioning in a large-scale business network and have high fault detection accuracy, low fault error detection rate, short detection time, rapid fault positioning time and low computational complexity.

Description

A kind of heuristic fault positioning method that is used for next generation network

Technical field

The present invention relates to a kind of heuristic fault positioning method that is used for next generation network, particularly a kind of technology of next generation network detection failure belongs to communication network detection technique field.

Background technology

The complexity of communication network, real-time and uncertainty are even make the multiple advanced technology means of integrated use also can't avoid the generation of fault fully.In order to reduce and to avoid network failure as far as possible to communication network and professional influence thereof, guarantee communication service high availability and high reliability, need efficient, the practical Fault Locating Method of development.The difficult point of communication service fault location is: (1) has not only reduced the efficient of fault location if the communication network fault produces the caused alarm windstorm of a large amount of alarm informations, also can flood some urgent, special alarms.(2) time of origin of each fault and position and other information all have sudden and uncertain, the simulation modeling difficulty, in addition, the communication network fault warning may exist lose, false, out of order, repetition, multiple phenomenon such as imperfect, may exist in addition and dodge situation disconnected and that periodically reappear, thereby make that the pests occurrence rule of grasping fault is extremely difficult.(3) network failure has transitivity usually, and single fault may cause other faults or many places fault, has strengthened the difficulty of fault location root.(4) the external sign of a lot of faults is fewer, and can not observe at the place of breaking down.(5) may there be a plurality of faults in communication network simultaneously at synchronization, and its sign information is weave in often.(6) the alarm form and the complete degree of isomery equipment and heterogeneous network have nothing in common with each other, and skew also usually appears in global clock, and these cause very difficulty of fault location.(7) computation complexity of fault location is very high, especially accomplishes very difficulty of accurate fault location.

Now, traditional Fault-Locating Test has obtained good effect at aspects such as SDH/WDM, procotol diagnosis, but still have following limitation: (1) lays particular emphasis on the fault location of communication network protocol layer, especially for the fault of network connections such as location link layer, network layer; (2) lay particular emphasis on management, the main discovery and the fixer network resource problem to the Internet resources layer; (3) adopt deterministic network failure propagation model mostly; (4) condition of supposing is too desirable, for example a plurality of faults can not take place simultaneously, false alarms etc. can not occur.Obviously, these conditions and judgment technology can not adapt to and satisfy the demand for development of current communication service.As everyone knows, along with the development of 3G and next generation network NGN new network technology such as (Next GenerationNetwork), following factors must be considered and satisfy to Fault-Locating Test: (1) will support determinate fault location and non-determinant fault location simultaneously; (2) according to concrete network and business, take suitable fault model and propagation model thereof; (3) consider real network and professional complex environment, the alarm sign may exist loses with false situation etc.; (4) have higher fault detect rate and lower false drop rate; (5) computation complexity of measuring method is lower, can satisfy the fault location requirement of extensive communication service.In a word, how to develop as early as possible a kind of can be quick, convenient, detect and the technology of fault location easily, become the very focus of urgent task and concern of scientific and technical personnel in the industry.

Summary of the invention

In view of this, the purpose of this invention is to provide a kind of heuristic fault positioning method that is used for next generation network, this method can be considered difficult point and the application demand that above-mentioned new network brings, satisfy the needs of extensive business network real time fail location, and fault detect accuracy rate height, the fault false drop rate is lower, and, detection time is short, and fault location time is fast, and computation complexity is also lower.

In order to achieve the above object, the invention provides a kind of heuristic fault positioning method that is used for next generation network, it is characterized in that: earlier the observable sign of network system is gathered together, therefrom find out the one or more faults that are associated the most with these observable signs; Again according to following phenomenon: when these faults satisfied all indications that causes producing in the set of may observe sign just, these faults will be formed " fault location set ", then by should " fault location set " judging and fault location; This method comprises following operating procedure:

(1) fault location set G is set, and it is carried out initialization, even this set G is an empty set: the fault that wherein also is not positioned;

(2) set up may observe sign set T, so that according to each the sign s among this may observe sign set T _i, find out its pairing possible fault, reconstruct corresponding failure collection F to be selected with this may observe sign set T, natural number subscript i is the sequence number of sign factor in the formula;

(3) to each the fault f among the failure collection F to be selected _i, natural number subscript i is the sequence number of fault in the formula; Calculate this fault f _iAnd the relevance degree bi between the may observe sign set T joins the degree of correlation set B that is used for preserving each relevance degree bi with this relevance degree bi again;

(4) a calculating sign set R who is used to preserve the intermediate variable of relevant sign is set, and should calculates sign and gather R and be initialized as empty set:

(5) each relevance degree bi in the degree of correlation set B is ranked according to order from high to low, and take out relevance degree bi according to the order of sequence successively, following compare operation is carried out in recirculation:

(51) find out pairing certain fault f with current relevance degree bi _i

(52) find out and this fault f _iPairing certain may observe sign subclass S _i

(53) judge this may observe sign subclass S _iWhether include and calculate the element that does not have among the sign set R, if i.e. S _iHelp to calculate the growth that sign is gathered the element number of R, then this may observe sign subclass S _iR carries out union operation with the set of calculating sign, promptly increases calculating sign set R: R=R ∪ S _i, carry out subsequent operation again; Otherwise, the operation of redirect execution in step (55);

(54) fault f _iAdd among the fault location set G, i.e. G=G ∪ f _i

(55) judge whether calculated sign set R equates with may observe sign set T, if then carry out subsequent step (6); Otherwise, return execution in step (51);

Judge whether described calculating sign collection R equates with may observe sign set T; If then carry out subsequent operation; Otherwise, return and re-execute this step operation;

(6) output fault location set G.

Described may observe sign set T is system or network when breaking down, set that can observed various signs;

Described may observe sign subclass S _iBe certain fault f _iThe set of all signs that can observe that produced;

Described fault location set G adopts that this method can be confirmed and locate and satisfies just to cause producing the set that the may observe sign is gathered whole faults of all indications among the T;

Described failure collection F to be selected is the set of pairing, various contingent all faults of all indications among the may observe sign set T.

In the described step (3), certain fault f _iAnd the computing formula of the relevance degree bi between the may observe sign set T is:

$\mathrm{bi}==\frac{{\mathrm{\Σ}}_{s\∈T}p\left(f_i\right|s)}{{\mathrm{\Σ}}_{s\∈A_i}p\left(f_i\right|s)};$ In the formula, p (f _i| s) expression is when producing certain sign s, and this sign s is by certain fault f _iThe probability that is caused; Ai represents when producing certain fault f _iThe time, the set of all indications that causes by this fault; Wherein, calculate p (f _i| the formula of value s) is: $p\left(f_i\right|s)=\frac{p\left(f_i\right)p\left(s\right|f_i)}{{\mathrm{\Σ}}_{g\∈N}p\left(g\right)p\left(s\right|g)};$ In the formula, f _iRepresent certain fault among the failure collection F to be selected; P (f _i) expression certain fault f among the failure collection F to be selected _iThe probability that produces; P (s|f _i) expression certain fault f in failure collection F to be selected _iDuring generation, this fault f _iThe probability that causes certain sign s to produce; N represents all set of fault in the system; G represents certain fault among whole failure collection N of system; The probability that fault g produces among whole failure collection N of p (g) expression system; P (s|g) expression is when fault g among whole failure collection N of system produces, and the probability that this fault g causes sign s to produce is p (s|g).

Compare operation in the described step (5) further comprises following content of operation:

(51) find out pairing certain fault f with current relevance degree bi _i

(52) find out and this fault f _iPairing certain may observe sign subclass S _i

(53) judge this may observe sign subclass S _iWhether include and calculate the element that does not have among the sign set R, if i.e. S _iHelp to calculate the growth that sign is gathered the element number of R, then this may observe sign subclass S _iR carries out union operation with the set of calculating sign, promptly increases calculating sign set R: R=R ∪ S _i, carry out subsequent operation again; Otherwise, the operation of redirect execution in step (55);

(54) fault f _iAdd among the fault location set G, i.e. G=G ∪ f _i

(55) judge whether calculated sign set R equates with may observe sign set T, if then carry out subsequent step (6); Otherwise, return execution in step (51).

Advantage of the present invention and effect: compare with additive method, the present invention is used for the fault detect accuracy rate height of the heuristic fault positioning method of next generation network, the fault false drop rate is lower, and, detection time is short, and fault location time is fast, and computation complexity is also lower, can satisfy the needs of extensive business network real time fail location, therefore good application prospects is arranged in next generation network.

Description of drawings

Fig. 1 is the heuristic fault positioning method operating procedure flow chart that the present invention is used for next generation network.

Fig. 2 (A), (B), (C), (D) are respectively that BSD method of the present invention is carried out four experimental result curves of emulation embodiment: fault detect rate, fault detect rate variance, fault false drop rate and the comparison diagram of failure detection time under the scene of determinate fault location.

Fig. 3 (A), (B), (C), (D) are respectively that BSD method of the present invention is carried out four experimental result curves of emulation embodiment: fault detect rate, fault detect rate variance, fault false drop rate and the comparison diagram of failure detection time under the scene of non-determinant fault location.

Embodiment

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

Introduce several noun of the present invention earlier:

(1) may observe sign set T is meant when system or network break down, set that can observed sign.

(2) fault location set G is meant by method of the present invention, the set that causes producing whole faults of all indications among the may observe sign set T that can confirm and locate and satisfied just.

(3) failure collection F to be selected is meant that all indications among the may observe sign set T is pairing, the set of various contingent all faults.

(4) calculate sign set R and be meant in the inventive method step, for preserving the intermediate variable that sign is set.

(5) degree of correlation set B is meant the set that is used to preserve relevance degree bi.

(6) may observe sign subclass S _iBe meant certain fault f _iThe set of all signs that can observe that produced.

The present invention is a kind of heuristic fault positioning method that is used for next generation network, its main thought is: the various signs that earlier network system be can observe gather together, therefrom find out the one or more faults the most relevant with these observable signs, again according to following phenomenon: when these faults satisfy all indications that causes producing in the set of may observe sign just, these faults will be formed " fault location set ", then by should " fault location set " judging and fault location.

Referring to Fig. 1, introduce the operating procedure of the inventive method:

(1) fault location set G is set, and it is carried out initialization, even this set G is an empty set: the fault that wherein also is not positioned.

(2) set up may observe sign set T, so that according to each the sign s among this may observe sign set T _i, find out its pairing possible fault, reconstruct corresponding failure collection F to be selected with this may observe sign set T, natural number subscript i is the sequence number of sign factor in the formula.

(3) to each the fault f among the failure collection F to be selected _i, natural number subscript i is the sequence number of fault in the formula; Calculate this fault f _iAnd the relevance degree bi between the may observe sign set T joins the degree of correlation set B that is used for preserving each relevance degree bi with this relevance degree bi again.In this step (3), certain fault f _iAnd the computing formula of the relevance degree bi between the may observe sign set T is: $\mathrm{bi}==\frac{{\mathrm{\Σ}}_{s\∈T}p\left(f_i\right|s)}{{\mathrm{\Σ}}_{s\∈A_i}p\left(f_i\right|s)};$ In the formula, p (f _i| s) expression is when producing certain sign s, and this sign s is by certain fault f _iThe probability that is caused; Ai represents when producing certain fault f _iThe time, the set of all indications that causes by this fault; Wherein, calculate p (f _i| the formula of value s) is: $p\left(f_i\right|s)=\frac{p\left(f_i\right)p\left(s\right|f_i)}{{\mathrm{\Σ}}_{g\∈N}p\left(g\right)p\left(s\right|g)};$ In the formula, f _iRepresent certain fault among the failure collection F to be selected; P (f _i) expression certain fault f among the failure collection F to be selected _iThe probability that produces; P (s|f _i) expression certain fault f in failure collection F to be selected _iDuring generation, this fault f _iThe probability that causes certain sign s to produce; N represents all set of fault in the system; G represents certain fault among whole failure collection N of system; The probability that fault g produces among whole failure collection N of p (g) expression system; P (s|g) expression is when fault g among whole failure collection N of system produces, and the probability that this fault g causes sign s to produce is p (s|g).

(4) a calculating sign set R who is used to preserve the intermediate variable of relevant sign is set, and should calculates sign and gather R and be initialized as empty set.

(5) each relevance degree bi in the degree of correlation set B is ranked according to order from high to low, and take out relevance degree bi according to the order of sequence successively, compare operation is carried out in recirculation, judges whether described calculating sign collection R equates with may observe sign set T; If then carry out subsequent step (6) operation; Otherwise, return the startup operation that re-executes this step.Compare operation in this step comprises following content of operation:

(51) find out pairing certain fault f with current relevance degree bi _i

(52) find out and this fault f _iPairing certain may observe sign subclass S _i

(53) judge this may observe sign subclass S _iWhether include and calculate the element that does not have among the sign set R, if i.e. S _iHelp to calculate the growth that sign is gathered the element number of R, then this may observe sign subclass S _iR carries out union operation with the set of calculating sign, promptly increases calculating sign set R: R=R ∪ S _i, carry out subsequent operation again; Otherwise, the operation of redirect execution in step (55);

(54) fault f _iAdd among the fault location set G, i.e. G=G ∪ f _i

(55) judge whether calculated sign set R equates with may observe sign set T, if then carry out subsequent step (6); Otherwise, return execution in step (51).

(6) output fault location set G.

Method of the present invention has been carried out the emulation experiment of embodiment, simulating scenes is divided into determinate fault location scene and non-determinant fault location scene two classes, and be provided with 4 modules in the simulated environment: network generation module, failure symptom concern generation module, fault use-case generation module, root fault location module.

Embodiment carries out emulation experiment with multimedia ring service system, and this test situation brief introduction is as follows:

Set up may observe sign set T earlier, the sign among this set T is: and CRBT is inconsistent with customization, and the CRBT call loss is bigger, the CRBT call interruption, system journal reports an error in a large number, can't hear CRBT }.Set up failure collection F to be selected according to actual conditions then, wherein fault comprises: and database process is unusual, the hard drive space deficiency, and the office data mistake, part signaling card damages }.Calculate the relevance degree between each fault and the may observe sign set T more respectively, and it is added in the degree of correlation set B, the degree of correlation set B of formation is: and 0.308107,0.292739,0.557341,1.000000}.After re-using the inventive method and calculating, the actual output fault location set G that obtains is: { part signaling card damages, the office data mistake }.

The MCA/MCA+ method that proposes in result of the test of the present invention and the list of references " Fault Management for Internet Service:Modeling and Algorithms " (publishing in IEEE ICC 2006) contrasts, and the index of contrast comprises fault detect rate, fault detect rate variance, failure detection time, fault false drop rate etc.Experimental result is as follows:

Referring to Fig. 2, be presented under the scene of determinate fault location, BSD method of the present invention is better than MCA algorithm and MCA+ algorithm, has higher fault detect rate, less fault false drop rate, and less fault location time.

Referring to Fig. 3, be presented under the scene of non-determinant fault location, BSD method of the present invention also is better than MCA algorithm and MCA+ algorithm, has higher fault detect rate, less fault false drop rate, and less fault location time.

Claims (3)

1. a heuristic fault positioning method that is used for next generation network is characterized in that: earlier the observable sign of network system is gathered together, therefrom find out the one or more faults that are associated the most with these observable signs; Again according to following phenomenon: when these faults satisfied all indications that causes producing in the set of may observe sign just, these faults will be formed " fault location set ", then by should " fault location set " judging and fault location; This method comprises following operating procedure:

(1) fault location set G is set, and it is carried out initialization, make that promptly this set G is an empty set: the fault that wherein also is not positioned;

(2) set up may observe sign set T, so that can be according to each the sign S among this may observe sign set T _i, find out its pairing possible fault, reconstruct corresponding failure collection F to be selected with this may observe sign set T, natural number subscript i is the sequence number of sign element in the formula;

(3) to each the fault f among the failure collection F to be selected _i, natural number subscript i is the sequence number of fault in the formula, calculates this fault f _iAnd the relevance degree bi between the may observe sign set T joins the degree of correlation set B that is used for preserving each relevance degree bi with this relevance degree bi again;

(4) a calculating sign set R who is used to preserve the intermediate variable of relevant sign is set, and should calculates sign and gather R and be initialized as empty set:

(5) each relevance degree bi in the degree of correlation set B is ranked according to order from high to low, and take out relevance degree bi according to the order of sequence successively, following compare operation is carried out in recirculation:

(51) find out pairing certain fault f with current relevance degree bi _i

(52) find out and this fault f _iPairing certain may observe sign subclass S _i

(53) judge this may observe sign subclass S _iWhether include and calculate the element that does not have among the sign set R, if i.e. S _iHelp to calculate the growth that sign is gathered the element number of R, then this may observe sign subclass S _iR carries out union operation with the set of calculating sign, promptly increases calculating sign set R: R=R ∪ S _i, carry out subsequent operation again; Otherwise, the operation of redirect execution in step (55);

(54) fault f _iAdd among the fault location set G, i.e. G=G ∪ f _i

(55) judge whether calculated sign set R equates with may observe sign set T, if then carry out subsequent step (6); Otherwise, return execution in step (51);

Judge whether described calculating sign set R equates with may observe sign set T; If then carry out subsequent operation; Otherwise, return and continue to carry out this step operation;

(6) output fault location set G.

2. heuristic fault positioning method according to claim 1 is characterized in that: described may observe sign set T is system or network when breaking down, set that can observed various signs;

Described may observe sign subclass S _iBe certain fault f _iThe set of all signs that can observe that produced;

Described fault location set G adopts that this method can be confirmed and locate and satisfies just to cause producing the set that the may observe sign is gathered whole faults of all indications among the T;

Described failure collection F to be selected is the set of pairing, various contingent all faults of all indications among the may observe sign set T.

3. heuristic fault positioning method according to claim 1 is characterized in that: in the described step (3), and certain fault f _iAnd the computing formula of the relevance degree bi between the may observe sign set T is:

In the formula, p (f _i| s) expression is when producing certain sign s, and this sign s is by certain fault f _iThe probability that is caused; Ai represents when producing certain fault f _iThe time, the set of all indications that causes by this fault; Wherein, calculate p (f _i| the formula of value s) is:

In the formula, f _iRepresent certain fault among the failure collection F to be selected; P (f _i) expression certain fault f among the failure collection F to be selected _iThe probability that produces; P (s|f _i) expression certain fault f in failure collection F to be selected _iDuring generation, this fault f _iThe probability that causes certain sign s to produce; N represents all set of fault in the system; G represents certain fault among whole failure collection N of system; The probability that fault g produces among whole failure collection N of p (g) expression system; P (s|g) expression is when fault g among whole failure collection N of system produces, and the probability that this fault g causes sign s to produce is p (s|g).

Images