1、INTERNATIONAL TELECOMMUNICATION UNION CCITT THE INTERNATIONAL TELEGRAPH AND TELEPHONE CONSULTATIVE COMMITTEE TELEPHONE NETWORK AND ISDN QUALITY OF SERVICE, NETWORK MANAGEMENT AND TRAFFIC ENGINEERING E.711 (1 0/92) USER DEMAND MODELLING Recommendation E.711 FOREWORD The CCITT (the International Teleg
2、raph and Telephone Consultative Committee) is a permanent organ of the International Telecommunication Union (ITU). CCITT is responsible for studying technical, operating and tariff questions and issuing Recommendations 011 them with a view to standardizing telecommunications 011 a worldwide basis.
3、The Plenary Assembly of CCITT which meets every four years, establishes the topics for study and approves Recommendations prepared by its Study Groups. The approval of Recommendations by the members of CCIT between Plenary Assemblies is covered by the procedure laid down in CCITT Resolution No. 2 (M
4、elbourne, 1988). Recoinmetidatioti E.711 was revised by Study Group II and was approved under the Resolution No. 2 procedure on the 30th October 1992. CCIT NOTE In this Recommendation, the expression “Administration” is used for conciseness to indicate both a telecommunication administration and a r
5、ecognized private operating agency. O ITU 1993 All rights reserved. No part of this publication may be reproduced or utilized in any form or by any means, electronic or mechanical, including photocopying and microfilm, without permission in writing from the ITU. Recommendation E.711 USER DEMAND MODE
6、LLING (revised 1992) 1 Introduction ISDN users have vnrious needs for information transfer. The user demand, as manifested in the user/CPE interface see figure Ia)/E.711, is characterized by an arrival process and by a call duration andor a quantity of information attached to each arrivai. Through t
7、he mediation of the Customer Premises Equipments (CPEs), user demands are translated into call dcrnniids for specific telecommunication services in the user-network interface. These call demands are modelled in this Recommendation to ailow the characterization of the tr,affic offered to the layers 1
8、 to 3 of the ISDN. Bascd on the call demand modelling, users will be characterized by the generating process of their originating and terminating call demans. 2 Modelling of a call demand 2.1 Genrrcll The traffic offered to the layers I to 3 at the user-network interface is modelled by the call dema
9、nd see Figure lbYE.71 i. For traffic engineering purposes, a call demand is defined by a set of connection characteristics and by a cai1 pattern: - The connection characteristics are described by the values of certain aributes given by Recommenhtion 1.210. Only some of the low layer and general attr
10、ibutes are significant. The values of each attribute can be: - selected by the user; - - negotiated between them. This set of amibutes must be sufficient to identify precisely the resources needed by the cai1 demand in the network, both in the user and control planes. The call pattern of a call dema
11、nd is described in terms of sequences of events at the user-network interface and of the times between events. prescribed by the service provider; - The call pattern, associated with the connection characteristics, must be sufficient to quantify the traffic offered to the network resources by the ca
12、ll demand, in the user plane and in the control plane. 2.2 Connection chcrructeristics As mentioned in 5 2.1, the connection characteristics of a call demand are defined by a set of low layer and general auribute values, defined in Recommendation 1.210, which are significant for traffic engineering.
13、 The low layer atEibutes are: - information transfer mode; - information transfer rate; - information transfer capability; Recommendation E.711 (1 OB2) 1 CCITT RECMN*E-7LL 92 W 4862593 O576507 LTB W FIGURE 1E.711 User demand modelling by call demands - establishment of communication; - symmetry; - c
14、ommunication configuration; - access channel and rate; - - signalling access protocol layers 1 to 3; information access protocol layers 1 to 3. Among the general attributes, which are still under study, the attribute “Supplementary services is of particular significance for traffic engineering. The
15、list of other traffic significant general attributes is for further study. In practice, when the connection characterization is made for a specific traffic engineering task, only some of the above-mentiotied attributes are significant. On the other hand, some of the values of the attributes given in
16、 the I-Series Recommendations are not sufficient for traffic engineering purposes. As an example, the defined values of the communication configuration attributes are point-to-point, multipoint and broadcast. For traffic engineering purposes, the number and position of the points muid the identifica
17、tion of the originating and terminating points have to be specified as well. The definition of other connection characteristics significant from the traffic engineering point of view is for further study. Note - High layer attributes, such as high layer protocols, are not considered as connection ch
18、aracteristics since, from the viewpoint of traffic offered to layers 1 to 3 of the ISDN, high layer protocol messages should be considered as user information and thus modelled by means of the call pattern. 2.3 Cull pattern and trafic variables As mentioned iii 0 2.1, the call pattern of a call dema
19、nd is defined in terms of sequences of events at the user-network interface and of the times between events. 2 Recommendation E.711 (10/92) The call pattern is defined by a set of traffic variables. These traffic variables are expressed as statistical variables described by some parameters related t
20、o their distribution. This allows a large variety of call demands to be modelled by the same call pattern. Call demands with the same types of events, but different numbers of such events (e.g. different number of re-attempts), or different times between them (e.g. different holding times), could be
21、 modelled by the sNne call pattern. 2.3.1.1 Cull uttempt urrival process This is the process of manual or automatic attempts in which a call demand can result (see Figure 2E.711). These attempts are recognized by the system as layer 3 Set-up messages. This process may be characterized, in particular
22、, by: i Two kinds of traffic variables can be distinguished AA A4 AA - call variables, describing events occurring during the call Set-up and release phases and the times between these events; Tm - transaction variables, describing events occurring during the information transfer phase and the times
23、 between these events. In the following, these two kinds of traffic variables are described, as well as the parameters to characterize them. Nevertheless, it must be observed that, when the call demand modelling is made for a specific traffic engineering task, only some of the following traffic vari
24、ables are needed. 2.3.1 Cull variables - - A more detailed characterization of the call attempt arrivai process is for fder study. mean number of re-attempts in case of non-completion; mean time between the call attempts. E BCD Beginning of call demand (manifested by the first manual attempt) MA Man
25、uaiattempt AA Automaticauempt SI stanfirindication,indicatingdnompletion FIGW 2E.711 CaU atiempt sequence Recommendation E.711 (10192) 3 CCITT RECMN*E-733 92 M 4862593 0576509 T70 2.3.1.2 Cull attempt holding times For a call dematid, several call attempt holding times, delimited by layer 3 inessage
26、s, are of interest, e.g. dialling time, ringing time aid conversation time. The definition of the holding times and of the parameters of their distributions which characterize them is for further study. As a first step, it may be sufficieiit to consider the inem total holding time. 2.3.2 Trunsuction
27、 vuriuhles The informatiou contents at the user plane duriiig a call may be produced iii discrete trarisactioiis, i.e. intervals during which a user is contiiiuously producing iiifonnatioii (see Figure 3E.711). This subdivision may be significant from a traffic point of view iii packet switched serv
28、ices aiid in soine cases in circuit switched services, eng. when Digitiil Circuit Multiplication Equipment (DCME) is used. Note I - The workload as related to single traiisactioris within a specific call may uiidergo one or more segmentation stages. The entire subject of workload segmetihtioii is fo
29、r further study. Note 2 - In the definition of transaction variables, the high layer protocol messages should be coiisidered as user information. Transaction Transaction Transaction Call start start start Call connected release 1 1 1 1 1 b Time - - TO20362083 t Transaction length Nu/e - Iiormatioii
30、traiisfer oidy occurs duriiig the traiisactioiis. HGURE 3/E.711 Transactions of a call 2.3.2.1 Trunsactioti arrival process The ,mival process for traisactious within a call is for further study. Nevertheless, as a first step it may be sufficient to consider the mean number of transactions per call.
31、 2.3.2.2 runsciction length The traiisactioo length, expressed in bits, represents the workload offered by the transaction through the user- iietwork interface. The distribution of tr,uisactions lengths is for further study. The mean transaction length may be considered, as a first step. 4 Recommend
32、ation E.711 (10/92) - .- - - CCITT RECNN*E.731 92 m 48b2593 0576530 792 m 2.4 Illiistrcitiun of cull demutid modellirig Sets of connection characteristics cm be defined for each telecommunications service. For a given uaftic engineering task, :i limited number of cCd patterns cnn be associated with
33、each telecommunication service/coiinection characteristics combination. The range of possible call pattenis to consider could be assumed u pruri or derived from the user characterization. Table 1E.711 gives a structured format for listing connection characteristics atid cai1 pntttriis. Teleciiiiiiit
34、iiiicatic )ti services (ilallies) TABLE 1E.711 List of telecommunication services, connection characteristics and call patterns defined in the call demand modelling Call patterns CP 1,1,1 CP 1,1,2 = Criiiiiectioii characteristics (sets) I I xc 1.1 CP 1.13 TS 1 TS 2 xc 2,l CP 2,1,1 CP 2,2,1 xc 2,2 CP
35、 1,2,1 ce 1,2,2 XC 1,3 CP 1,3,1 CP 1,3,2 Call patterns listed in association with a telecommutiiatioiis service/conriection characteristics combination must be described iii tenns of traffic vnriables appropriate for that combination. Table 2E.711 gives exrunples of both connection characteristics a
36、nd call patteni descriptions. Recommendation E.711 (10/92) 5 CCITT RECflN*E.7LL 92 W 4862593 057b5LL 629 W TABLE 2E.711 Example of connection characteristics and call pattern descriptions Telecommunication services Telephony Data communication Connection characteristics Circuit switched 64 kbit/s Sp
37、eech On-demand Bidirectional symmetric Point-to-point Points in different districts of the same city Packet switched 64 kbit/s Unrestricted digital On-demand Unidirectional Point-to-two points Points in the same district Access channel: I) Access protocol: Recommendation Q.93 1 Call patterns Call va
38、riables Mean number of re-attempts Mean holding time Trclnsuction variables Mean number of messages for user-to-user signalling Mean message length Call variables Mean number of re-attempts Both connections established and released simultaneously Mean holding time Transaction variables Mean number o
39、f messages for user-to-user signalling Mean message length Mean number of transactions per call Mean transaction length 3 User characterization 3.1 General Froin a traffic engineering point of view users sharing the same CPE should not be characterized individually, but rather as an ensemble which i
40、s called the CPE user set. The CPE user set is taken to be the user, or set of users, accessing the network through a single CPE (regardless of the number of ISDN accesses from the CPE). A CPE user set is characterized by the generating process of its originating and terminating call demands. In the
41、 initial phase this process may be approximated by the mean rate of call dem,uids of each type during a reference period. A more detailed characterization of this process, which includes higher order statistical moments, is for further study. User characterization has to be done for each CPE user se
42、t, in order to dimension the user set ISDN accesses, and for each population of CPE user sets which could share network resources. 3.2 Churueterizution of u CPE user set The ch,uacterization of a CPE user set can be achieved by the following steps: - determine the call deinand rate, a(), for each te
43、lecommunication service i, under the reference conditions; - determine which sets of connection characteristics (id) are requested by the CPE user set, for each telecoinmunication service, i; 6 Recommendation E.711 (10/92) - - _ - - - CCITT RECMN*E-731 92 4862593 0576512 565 m - determine the propor
44、tion, p.x(ij), of each of these sets of connection characteristics; I Recommendation E.711 (10/92) 7 - derive the rates, rn(ij), of each set of connection characteristics, by the following formula: - determine the call patterns (ij,k) used by the CPE user set for each set of connection characteristi
45、cs (ij); - determine the proportion, pc(ij,k), of each of these call patterns (iJ,k); - finally, derive the call pattern rates, rc(j,k), by the following formula: 3.3 Chuacterizution of u population of CPE riser sets The demand of a population of CPE user sets is characterized by the mean value per
46、CPE user set in the population of the rates defined in 5 3.2. A practical procedure to obtain these values is to divide the population of the CPE user sets into CPE user set classes, each class consisting of CPE user sets for whom similar traffic behavior can be expected. This classification CN be b
47、ased on the nature of the users (e.g. residential, small business, etc.) andor their CPEs. Once the CPE user set classes have been specified, each ciass is characterized by its proportion in the population of CPE user sets, and by the variables (proportions and rates) defined in 5 3.2. Finally, the
48、mean value of each rate per CPE user set in the whole population is calculated as the weighted average of its me,m value in each class. An example of this procedure is provided in Annex A. 4 Recommendation history First published in 1988, entitled “User Demand“. Revised and republished in 1991. CCIT
49、T RECNN*E.733 72 4862573 0576513 4T3 I 920 8 18 25 210 ANNEX A (to Recommendation E.7 1 1) Example of the characterization of a population of CPE user sets according to their call demands Table A-IE.711 shows the break-down of the populatioii into classes of CPE user sets. The telecoinrnutiicatioti services used by each class are defined, as well as the corresponding call detnand rates. TAHLE A-1lE.711 Examples of classes of CPE user sets and their deiiiaiid rates for telecommiinication services CPE user set classes Residen tia1 Sinal1 busin
