WO2003073697A1 - Methode d'identification d'elements limitant les performances d'un reseau de communication - Google Patents
Methode d'identification d'elements limitant les performances d'un reseau de communication Download PDFInfo
- Publication number
- WO2003073697A1 WO2003073697A1 PCT/FR2002/000485 FR0200485W WO03073697A1 WO 2003073697 A1 WO2003073697 A1 WO 2003073697A1 FR 0200485 W FR0200485 W FR 0200485W WO 03073697 A1 WO03073697 A1 WO 03073697A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- variable
- local
- variables
- local variables
- distributed
- Prior art date
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L43/00—Arrangements for monitoring or testing data switching networks
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/16—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks using machine learning or artificial intelligence
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L43/00—Arrangements for monitoring or testing data switching networks
- H04L43/08—Monitoring or testing based on specific metrics, e.g. QoS, energy consumption or environmental parameters
- H04L43/091—Measuring contribution of individual network components to actual service level
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L43/00—Arrangements for monitoring or testing data switching networks
- H04L43/08—Monitoring or testing based on specific metrics, e.g. QoS, energy consumption or environmental parameters
- H04L43/0852—Delays
Definitions
- the present invention relates to a method of identifying elements 5 limiting the performance of a communication network.
- Such a network connects a plurality of elements each connected to at least one other element.
- the terminal elements accessible to network users, the intermediate elements such as switches or routers which are arranged between the different terminals, and the physical links which also belong to the network.
- the network makes it possible to convey information between two terminals or even between two intermediate elements, the term information having here to be taken in its widest meaning.
- IP Internet Protocol
- the network therefore interconnects a plurality of elements and it must be ensured that it has the required performance. These performances consist
- the various elements, intermediate or terminal, are provided with equipment which can limit the transmission performance.
- the transit time between two terminals depends on the speed of the output interface of the first terminal, the load of a switch and the speed on a physical link.
- Transit time is a distributed variable because it involves several network elements, while the interface speed, the switch load and the link speed are local variables because they only concern one element of the network. network.
- a distributed variable which therefore depends on several local variables 35 may fall below a required performance threshold.
- the present invention thus relates to a method of identifying the local variable or variables which are responsible for this drop in performance.
- this method for identifying at least one limiting element comprises the following steps:
- the method further comprises a step for calculating a multidimensional correlation function of the distributed variable with the local variables, the limiting element being that of which the local variable has the greatest contribution in this function.
- the correlation function is a multilinear regression.
- the method comprises a step for removing from this set a local variable whose correlation coefficient with another local variable exceeds a predetermined threshold.
- the method comprises a step for removing from this set a local variable whose correlation coefficient with the distributed variable is less than a predetermined threshold.
- the method includes a step to remove from this set the local variables whose correlation coefficients with the distributed variable are the lowest, a predetermined number of local variables being retained in the set.
- the method comprises a step of preselecting the local variables making up this set.
- this preselection stage rejects from the set the local variables intrinsically correlated to the variable distributed by the constitution of the network. Likewise, this preselection stage rejects from the set the local variables which have a number of measurements lower than a predetermined level.
- this set comprising n local variables
- the method comprises a step for producing derived models obtained by retaining at least some of the combinations comprising at most d local variables, d being less than n, and it further comprises a step for substituting one of these derived models for the initial model.
- the method comprises, for each derived model, a step for calculating an influence estimator for each of the local variables.
- the method includes a step of selecting these derived models.
- this selection step retains the derived models for which all the influence estimators exceed a predetermined validation threshold or those for which the ratio between each influence estimator and the highest influence estimator exceeds a validation ratio predetermined.
- the method comprises a step for calculating an error estimator for each of the derived models which estimates the error between the latter and the distributed variable.
- the selection step keeps the derived model which is associated with the error estimator having the lowest value.
- the NET network in particular connects a first T1, a second T2 and a third terminal T3 respectively via a first 11, a second 12 and a third 13 interface.
- the network has three switches SW1, SW2, SW3, and a ROUT router.
- the link from the first T1 to the third terminal T3 therefore involves the first interface 11, the first switch SW1, the router ROUT, the second switch SW2 and the third interface 13. This link makes it possible to provide a communication service.
- the distributed variable that constitutes the transit time between the first T1 and the third terminal T3 therefore depends on a plurality of local variables, including: - the speed of the first interface 11,
- the different variables, local or distributed, are mean values of instantaneous values sampled in synchronism.
- Various techniques can be used to acquire the value of these variables and we will therefore mention intrusive measurement and the collection of local indicators.
- the intrusive measurement is done by means of a program which performs a prescribed action during which different variables are measured such as the total execution time.
- Collecting local indicators consists of consulting the lists of indicators kept by most equipment, these indicators reflecting their states.
- M C.I + ⁇ Wj.Vj, where:
- J - Vj is the time series of the local variable j
- - I is the identity vector (1,1,., 1).
- Multilinear regression leads to the weights Wj following the minimization of the merit function ⁇ 2 which represents the sum of the squares of the differences between the values of the model and the corresponding values of the distributed variable:
- Vji is the ith value of the local variable j
- the model is of the form:
- X 2 ⁇ [Ri- (C + W1.V1i + W2.V2i + W3.V3i)] 2 i
- a table is therefore provided in which all the variables appear, whether local or distributed.
- a first way is to find out whether two local variables are highly correlated and, if so, to remove one of these variables from the model.
- the set of p local variables appears in an entry list and we consider the variable Vj, the value j appearing in a first counter initialized to 1.
- the correlation coefficient k between the variables Vj and VI, is then calculated.
- the variable Vj is introduced into an output list and if the correlation coefficient k exceeds a predetermined threshold, 0.9 for example, the variable VI is removed from the input list.
- the previous operation is iterated by incrementing the second counter by one unit until I reaches the value p.
- the aggregation operation is thus continued until the first and second counters are respectively assigned the values (p - 1) and p.
- a second way to reduce the number of local variables to take into account in the model is to calculate the correlation coefficient of each local variable with the distributed variable.
- a first option only the local variables are retained leading to a coefficient greater than a predetermined threshold, 0.5 for example.
- a predetermined threshold 0.5 for example.
- only the local variables leading to the highest coefficients are kept. In this case, a predetermined number of local variables is retained, 5 for example.
- the second means can be implemented before the first means.
- a third way to reduce the number of local variables is to pre-select these variables.
- This third means can naturally be used before the first or second means. It is preferably used at the start of the method according to the invention.
- an influence estimator Elj which specifies the influence of the local variable Vj in the derived model in question.
- - CAW is the sum of the absolute values of the weights Wj, - Q the number of local variables involved in the model, and
- the - Ri is the ith value of the distributed variable R
- Vji is the ith value of the local variable j
- - Rmax is the maximum value of the distributed variable R
- - u is a weighting exponent.
- the value 0 is adopted for the weighting exponent u, the error then corresponding to the standard error as defined in the work cited above.
- Elj CCj. (Wj. Q. CAW) 2 / BSEj2
- the Elj influence estimators are therefore calculated relatively to at least some of the derived models for each local variable. Preferably, all derived models are taken into account.
- a derived model is selected only if all of the influence estimators relating to this model exceed a predetermined validation threshold.
- a derived model is selected only if the ratio between each influence estimator and the highest influence estimator exceeds a predetermined validation ratio.
- the first option, the second option or a combination of these two options can lead to a single derivative model selected. However, if not, it should be decided which of the selected models should be kept.
- an error estimator EE is defined which specifies the error between the derivative model MD concerned and the distributed variable. This estimator takes the following form, for example: - MDi is the ith value of the derivative model MD,
- - MD is the mean value of the derivative model MD
- - 1 is a correction exponent.
- t is the value 0 for the correction exponent t, which here also corresponds to the so-called standard error.
- the derived model which is kept is the one which leads to the error estimator having the lowest value.
- the application response time that is to say the time that elapses between the formulation of a request by a user and the response to this request
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Environmental & Geological Engineering (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Computer And Data Communications (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
Description
Claims
Priority Applications (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
EP02703679A EP1472818B1 (fr) | 2002-02-07 | 2002-02-07 | Methode d'identification d'elements limitant les performances d'un reseau de communication |
AU2002237365A AU2002237365A1 (en) | 2002-02-07 | 2002-02-07 | Method of identifying elements limiting the performances of a communication network |
DE60219078T DE60219078T2 (de) | 2002-02-07 | 2002-02-07 | Verfahren zur identifikation von den durchsatz begrenzenden elementen in einem kommunikationsnetz |
PCT/FR2002/000485 WO2003073697A1 (fr) | 2002-02-07 | 2002-02-07 | Methode d'identification d'elements limitant les performances d'un reseau de communication |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
PCT/FR2002/000485 WO2003073697A1 (fr) | 2002-02-07 | 2002-02-07 | Methode d'identification d'elements limitant les performances d'un reseau de communication |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2003073697A1 true WO2003073697A1 (fr) | 2003-09-04 |
Family
ID=27763521
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/FR2002/000485 WO2003073697A1 (fr) | 2002-02-07 | 2002-02-07 | Methode d'identification d'elements limitant les performances d'un reseau de communication |
Country Status (4)
Country | Link |
---|---|
EP (1) | EP1472818B1 (fr) |
AU (1) | AU2002237365A1 (fr) |
DE (1) | DE60219078T2 (fr) |
WO (1) | WO2003073697A1 (fr) |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5907810A (en) * | 1996-11-06 | 1999-05-25 | Northern Telecom Limited | Method for reducing paging load in a cellular communication system |
WO2000011629A1 (fr) * | 1998-08-07 | 2000-03-02 | Dinbis Ab | Procede et moyen de controle de grand itineraire |
-
2002
- 2002-02-07 DE DE60219078T patent/DE60219078T2/de not_active Expired - Lifetime
- 2002-02-07 AU AU2002237365A patent/AU2002237365A1/en not_active Abandoned
- 2002-02-07 EP EP02703679A patent/EP1472818B1/fr not_active Expired - Lifetime
- 2002-02-07 WO PCT/FR2002/000485 patent/WO2003073697A1/fr active IP Right Grant
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5907810A (en) * | 1996-11-06 | 1999-05-25 | Northern Telecom Limited | Method for reducing paging load in a cellular communication system |
WO2000011629A1 (fr) * | 1998-08-07 | 2000-03-02 | Dinbis Ab | Procede et moyen de controle de grand itineraire |
Non-Patent Citations (1)
Title |
---|
VOLOCHINE T ET AL: "NETWORK MANAGEMENT AND TRAFFIC ANALYSIS FOR CICNET", IEEE NETWORK,IEEE INC. NEW YORK,US, vol. 5, no. 5, 1 September 1991 (1991-09-01), pages 41 - 50, XP000248472, ISSN: 0890-8044 * |
Also Published As
Publication number | Publication date |
---|---|
EP1472818A1 (fr) | 2004-11-03 |
AU2002237365A1 (en) | 2003-09-09 |
DE60219078T2 (de) | 2007-12-13 |
EP1472818B1 (fr) | 2007-03-21 |
DE60219078D1 (de) | 2007-05-03 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2008056041A1 (fr) | Methode et appareil pour fournir un equilibrage de charge base sur le flux | |
EP1155531B1 (fr) | Systeme et procede de mesure des durees de transfert et des taux de pertes dans des reseaux de telecommunication haut-debit | |
CN106899448B (zh) | 适用于网络状态及性能测量的一体化动态赋权评估方法 | |
WO1996017297A1 (fr) | Outil d'aide a la repartition de la charge d'une application repartie | |
EP1743454A1 (fr) | Mesure de performance dans un reseau de transmission de paquets | |
WO2005122473A1 (fr) | Mesure de performance dans un reseau de transmission de paquets | |
FR2830094A1 (fr) | Procede et dispositif de simulation du comportement d'un reseau, permettant un dimensionnement a la demande | |
EP1401146A1 (fr) | Dispositif et procédé de planification de configuration d'un réseau de communications par prévision d'évolution | |
FR2843263A1 (fr) | Dispositif et procede de determination de chemins de routage dans un reseau de communications, en presence d'attributs de selection | |
EP1605632A2 (fr) | Corrélation de mesures passives de type bout-en-bout, au moyen de filtres hiérarchisés | |
WO2003073697A1 (fr) | Methode d'identification d'elements limitant les performances d'un reseau de communication | |
EP0874533B1 (fr) | Procédé d'ordonnancement de paquets à pertes équitables | |
FR2812734A1 (fr) | Methode d'identification d'elements limitant les performances d'un reseau de communication | |
FR2848044A1 (fr) | Procede et systeme informatique pour declencher une action sur des donnees de communication numerique | |
EP1469635B1 (fr) | Dispositif de gestion de mesures de paramètres de flux de type bout-en-bout pour un réseau de communications multi-domaines | |
EP3563558A1 (fr) | Reseau informatique de noeuds communiquant entre eux par messages en pair a pair et procede d'interconnexion entre noeuds associe | |
EP4301050B1 (fr) | Procédé de gestion de la connexion d'un équipement utilisateur à un réseau de communication sans fil distribué | |
EP1753170A1 (fr) | Dispositif de contrôle d'indicateurs de performances d'un réseau de commmunication, pour un outil de gestion et/ou d'optimisation de réseau | |
WO2003085887A1 (fr) | System de gestion de mesure de la qualite d'un service sur un reseau | |
EP1328084A1 (fr) | Détermination de la cause d'une dégradation de la qualité d'un service en fonction de l'évolution d'un ensemble de services | |
FR2846170A1 (fr) | Procede d'optimisation du trafic sur internet. | |
EP1398908A1 (fr) | Procédé et dispositif d'adaptation automatisée de SLA et/ou de services dans un réseau de communications | |
EP1667361B1 (fr) | Dispositif d'interfaçage entre des outils de diagnostic et des modules d'analyse d'un réseau de communication | |
EP1372295A1 (fr) | Dispositif et procédé de contrôle de profils, notamment de flux de données, dans un réseau de communications | |
FR2858735A1 (fr) | Equipement de reseau a filtrage adaptable de trafics en vue de mesures de parametres, et procede de mesure associe |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
AK | Designated states |
Kind code of ref document: A1 Designated state(s): AL AM AT AU AZ BA BB BG BR BY CA CH CN CU CZ DE DK EE ES FI GB GE GH GM HR HU ID IL IS JP KE KG KP KR KZ LC LK LR LS LT LU LV MD MG MK MN MW MX NO NZ PL PT RO RU SD SE SG SI SK SL TJ TM TR TT UA UG US UZ VN YU ZW |
|
AL | Designated countries for regional patents |
Kind code of ref document: A1 Designated state(s): GH GM KE LS MW MZ SD SL SZ TZ UG ZM ZW AM AZ BY KG KZ MD RU TJ TM AT BE CH CY DE DK ES FI FR GB GR IE IT LU MC NL PT SE TR BF BJ CF CG CI CM GA GN GQ GW ML MR NE SN TD TG |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application | ||
WWE | Wipo information: entry into national phase |
Ref document number: 2002703679 Country of ref document: EP |
|
WWP | Wipo information: published in national office |
Ref document number: 2002703679 Country of ref document: EP |
|
NENP | Non-entry into the national phase |
Ref country code: JP |
|
WWW | Wipo information: withdrawn in national office |
Country of ref document: JP |
|
WWG | Wipo information: grant in national office |
Ref document number: 2002703679 Country of ref document: EP |