traffic concept trafic engineering planning (1)

7/29/2019 Traffic Concept Trafic Engineering Planning (1)

1/44

CETTM MTNL

Traffic Concept

&Traffic Engineering Planning

(Mod Id:GSSFTRF005)


2/44

CETTM MTNL

Traffic Concepts Technological Developments

Services Developments

QoS - Quality of Service

Traffic engineering principles Traffic characterization

Traffic engg. Task

Traffic demand charecterisation

GOS objectives Traffic controls and dimensioning

Contents


3/44

CETTM MTNL

Telephone traffic is originated by the individual

needs of different subscribers and therefore is

beyond the control of administration

Telephone traffic for a particular exchange follows apattern of activity in that area

Normally there is a peak in morning and

afternoon and a dip during lunch period

Any of the subscriber ( or every subscriber ) canoriginate a call at any given moment.

The duration of calls are not known.

Telephone Traffic

Traffic Concepts


4/44

CETTM MTNL

2500

2000

1500

1000

500

0

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

Calls originated Time

per hour Variations in Calling Rate

On a network, load & traffic pattern varies during

the day with heavy traffic and low traffic durations

Traffic Pattern


5/44

CETTM MTNL

Busy Hour:The hour in which maximum trafficusually occurs in an exchange is known as busyHour

Busy Hour varies from day to day or over a

number of days Busy Hour Traffic is the average value

of maximum traffic in the busy hour

One hour period starting at the same

each day for which the Average TrafficVolume or Number of Call Attempts is

greatest over the days under

consideration


6/44

CETTM MTNL

One hour period starting at the same each day for

which the Average Traffic Volume or Number of CallAttempts is greatest over the days under

consideration

Busy Hour Call Attempt :No. of Call Attempts in a

busy hour Call Completion Ratio : Ratio of Number of Successful

Calls to Number of Offered Calls

Busy hour calling rate: No.of calls originated per

subscriber in the busy hour

Cost Constraints:Cost of the line and certain

individual equipment is independent of the volume of

traffic


7/44

CETTM MTNL

Exchange with 2000 Subscribers of averageBHCA 10,000 & CCR 60%. Calculate the BHCR.

Avg. BH Calls = BHCA x CCR

= 6000 CallsBHCR = Avg. BH Calls

Total No. of Subscribers = 3

Erlang:If one circuit is held continuously for onehour then the traffic carried by that circuit

amounts to one Erlang ( 1 Traffic Unit )


8/44

CETTM MTNL

Average traffic: Ratio of Sum of HoldingTimes to Period

A = S/T

If C = total number of calls during the period Tthen Avg. holding time t = S/C hrs per call

Therefore A = C * t/T

Traffic Intensity :Ratio of Period for which anequipment is occupied to total period ofobservation

Erlang traffic: Period of observation is

generally considered one hour


9/44

CETTM MTNL

A subscriber makes 3 calls of 3 min, 4 min and 2

min duration in 1 hour period.

Calculate subscriber traffic in Erlangs.

Solution:Traffic = Busy period /Observed period

= (3+4+2)/60

= 9/60

= 0.15 E

Example


10/44

CETTM MTNL

Traffic carried by network is usually lower than

traffic offered to the network The overload traffic is rejected and not carried by

network

The traffic rejected by network is the index of QoS

Grade of Service

GoS = Lost Traffic / Offered Traffic= (A-A0) / AA - Offered traffic, A0 - Carried Traffic

(A Ao) - Lost Traffic

Recommended GoS = 0.002

i.e. 2 out of 1000 calls allowed to be lost


11/44

CETTM MTNL

More general term than GoS includes other factors

like Quality of Speech, Error free transmission

capability, etc.

Loss system: Circuit switched networksWhen overloaded, a user call is blocked and userhas to make a retry

Delay System:Packet switched (store & forward)

networkDelay when extended beyond limits becomes a losssystem In a store & forward network if the queue

becomes full, then further requests have to be rejected

Quality of Service


12/44

CETTM MTNL

Parameters for Loss system :- Grade of Service

- Blocking Probability

Blocking Probability

Probability that all equipments in a system are

busy When all equipments are busy no further traffic

can be carried by the system and the further

arrival traffic is blocked

Blocking Model


13/44

CETTM MTNL

Queuing Model:Parameters for Delay System

Service Delays

Flow ControlTo prevent loss, queue of traffic is cleared to an

acceptable limit Flow Control technique used to

prevent loss of traffic


14/44

CETTM MTNL

Designing a cost effective network which

provides the required Quality of Service under

varied traffic conditions demands a formalscientific basis

Traffic Engineering provides a means to

determine the quantum of exchange equipmentsrequired to provide a particular level of service for

a given traffic pattern and volume.

Traffic Engineering


15/44

CETTM MTNL

Erlang-B formula

A: User traffic described by offered traffic AN: Network described by number of channels n

E: Quality-of-Service described by blocking

probability E

Robust to the traffic process


16/44

CETTM MTNL

Economy of scale


17/44

CETTM MTNL

Packet based transfer mode

Packetized voice

Wireless access networks Mixed core networks

Photonic backbone networks

Centralized & decentralized control

Networking development

Technological developments


18/44

CETTM MTNL

Differentiated services Narrowband & broadband

Real-time services:

Delay sensitive

Jitter (delay variation) sensitive Non-real-time services

Packet loss sensitive

Best effort services

Services development


19/44

CETTM MTNLQoS Quality of Service

User perceived QoS Operator perceived QoS

System perceived Qos

Differentiated QoS

Gold Silver Bronze in UMTS Other classifications in e.g. ATM

Service Level Agreements


20/44

CETTM MTNL

Traffic engineering principles

QoS can only be guaranteed by ressource reservation End

toend

1. Bandwidth based mechanism

Separation: Imply low utilization >= high cost

Minimum bandwidth guaranteed => worse case

guarantee Sharing : Imply high utilization >= low cost

Minimum guaranteed & Maximum bandwidth

We may get obtain both QoS and low cost

Virtual circuit switched networks (ATM, MPLS)Packet streams are characterized by their effective

bandwidth


21/44

CETTM MTNL

2. Priority mechanisms: split services into priority classes

High priority traffic:Preemptive-resume:

High QoS to limited amount of traffic

* Non-preemptive:

Lower QoS to limited amount of traffic

Low priority traffic: Best effort traffic

Requires Admission Control and Policing: specification

of traffic characteristics + control of these

Bandwidth based mechanism has built-in access

control and policing

Traffic engineering principles


22/44

CETTM MTNLPriority Queueing system

Type 1: Load 0.1 erlang, mean service time 0.1 sType 2: Load 0.8 erlang, Mean service time 1.6 s

No priority: W = 12.85 s (for everybody)

Non-preemptive: W1 = 1.43 s

W2 = 14.28 sPreemptive resume: W1 = 0.0056 s

W2 = 14.46 s

(twice as many type 1 jobs as of type 2)


23/44

CETTM MTNL

Processor sharing - Generalized

Processor sharing: all users share the

available capacityGeneralized Processor sharing: maximumcapacity for each user

Robust to the service time (file size)

Mean performance measures are the same as

for Erlangs waiting time system

This model is applicable for Best Effort traffic

(Web traffic)


24/44

CETTM MTNLTraffic and service characterization

A service type is characterized by

Qos parameters (discussed above) Traffic characteristics

Traffic characteristics are in general statistical

(random variables)

Examples are:

Bandwidth demand (simple):

Packetetized services(e.g. Web browsing): fluctuating

Streaming services: constant

VoIP: On/Off (two-level)

Packet arrival process (complex): Leaky bucket control


25/44

CETTM MTNL

Bundling (QoS point of view)

Different services should be kept separate logically. Connections with same characteristics should be

bundled

Grooming (ressource utilization point of view)To save multiplexing equipment and to increase

utilization.

This is important in core and backbone networks

Recent development in traffic modelling

Traffic and service characterization


26/44

CETTM MTNLTraffic Engineering Tasks

Traffic demand

characterization

Grade of service

objective

Traffic control anddimensioning

Performance

monitoring


27/44

CETTM MTNLTraffic Demand Characterisation

Traffic

Modelling

Traffic

Measurement

Traffic

Forcasting

Simplifying assumption

Relevant parameters

Validation of models

Estimation of parameters

value

Forcasting demands for the planning period


28/44

CETTM MTNL

Modelling of the user demand (E.711, E.716)

Modelling of the call demands

Call attributes: information transfer mode,

comunication configuration, etc.

Call pattern: call-level and packet-level traffic

variables Modelling of their arrival process

Modelling of the traffic offered to a group of resources

In the user plane (E.712) and in the control plane

(E.713) Modelling in mobile networks (E.760)

Estimate traffic demand in each coverage area

Estimate handover and location updating rates

Traffic Demand Charecteristics Modelling


29/44

CETTM MTNL

General and operational aspects

Short, medium and long term applications Operational requirements

Direct measurements vs. call detailed records.

Technical aspects

Measurements principles Criteria to choose the length of the read-out

period: statistical confidence, stationarity of the

process

Normal and high loads Estimation of offered traffic

Measurements requirements PSTN & N -ISDN , B-ISDN and SS No.7

TRAFFIC DEMAND CHARACTERISATION MEASUREMENTS


30/44

CETTM MTNL

Forecasting of traditional services

Principles and pre-requisites Base data: traffic, economic, social,

demographic data

Strategies for dealing with missing data

Mathematical techniques Models: curve-fitting, autoregressive,

ARIMA, Kalman filtering, etc

Methods for the choice and evaluation of

models implementation

TRAFFIC DEMAND CHARACTERISATION FORECASTING


31/44

CETTM MTNL

Forecasting of new servicesThere are not historical data

Techniques: market research, expert opinion

and sectorial econometrics

Combination of techniques, adjustments afterimplementation

TRAFFIC DEMAND CHARACTERISATION FORECASTING


32/44

CETTM MTNL

Finitetraffic handlingcapacity

Stochastictrafficdemands

Losses,delays..

GOSObjectives

QoS Requirements

User Oriented

Described in network

independent terms

GOS objective

Traffic related NP

objective

GOS Objectives:Principles


33/44

CETTM MTNL

Traffic routing Hierarchical or non-hierarchical

Fixed or dynamic (time-, state- or event-

dependent routing)

Network traffic management controls

Maintain the throughput under overload or

failure conditions

Protective or expansive controls

TRAFFIC CONTROLS & DIMENSIONING


34/44

CETTM MTNL

Service protection Discriminatory restriction of the access to

circuit groups with little idle capacity

For restricting overflow traffic and for

balancing or differentiating GOS

Packet-level traffic controls

Assure packet-level GOS objectives of the

accepted calls

Provide packet-level GOS differentiation

Signalling and IN controls



35/44

CETTM MTNL

Dimensioning of circuit groups

Only-path circuit groups and high-usage/

final circuit group arrangements

Single-rate and multi-rate connections

Service protection methods: comparison and

parameter optimisation

GOS objectives and cost optimisation criteria



36/44

CETTM MTNL

Dimensioning of cirtuit groups with DCME equipment Impact of special N-ISDN features

Attribute negotiation, service reservation,

multipoint connections

Network dimensioning using end-to-end GOS

objectives

Fixed and dynamic routing

Decomposition of the network into independent

blocks

Interactive procedures for network costoptimisation



37/44

CETTM MTNL


PACKET-SWITCHED NETWORKS (I)

New

connection

Packet-level

GOS evaluatedPacket-level

GOS satisfied Accepted

Accepted

Yes

No

Connection-oriented networks with Connection

Admission Control (CAC)


38/44

CETTM MTNL

Packet-level perspective

Packet-level independent from:

Connection-level offered traffic

Connection-level traffic controls

Network dimensioning

Connection-level perspective

Effective bandwidth summarises the packet level

Connection-level independent from:

Packet-level offered traffic


Similar to a multi-rate circuit-switched network



39/44

CETTM MTNL


PACKET-SWITCHED NETWORKS (I)

New

connection

Effective

bandwidth disignAvailable

bandwidth Accepted

Accepted

No

Connection-oriented networks with Connection Admission Control

(CAC)


40/44

CETTM MTNL

Packet-level perspective

Packet-level independent from:

Connection-level offered traffic

Connection-level traffic controls

Network dimensioning

Connection-level perspective

Effective bandwidth summarises the packet level

Connection-level independent from:

Packet-level offered traffic


Similar to a multi-rate circuit-switched network



41/44

CETTM MTNL


PACKET-SWITCHED NETWORKS (II)

Framework

Principles and definitions: CAC role Strategies for logical network configuration


Methods for CAC

Methods for GOS differentiation: loss and delay priority Methods for adaptive resource management: ABR and

ABT

Dimensioning

Circuit group dimensioning and network dimensioning

Traffic routing and service protection methods

Particular features of packet-switched networks


42/44

CETTM MTNL


SIGNALLING & INTELLINGENT NETWORKS

Dimensioning of signalling networks

Performance under failures and traffic overload

Maximum design link load max

Acceptable performance at load 2max

Allocation and dimensioning of intelligent networkresources

Particular features, e.g., mass calling situations

Fast implementation of new services with uncertain

forecast Quick and flexible procedures for allocation and

dimensioning


43/44

CETTM MTNL


SIGNALLING & INTELLINGENT NETWORKS

Traffic controls

Guidelines for the choice of control parameters

Requirements on node-level overload controls

Harmonisation principles: multivendor, multioperator


44/44

CETTM MTNLPERFOMANCE MONITORING

Detect performance degradations for taking

feedback actions

Degradation reason Feedback action

Short term Overload/Failure Management Control

Medium term

Long term Traffic growth / New Networkplanning

error,.Modelling

approx

Network reconfiguration,routing

changes,control adjustments

Common aspects with traffic measurements Traffic reference periods Monitored GOS must meet GOS objectives for

normal and high loads

Consistent with traffic intensity read-out periods End-to-end GOS monitoring Methods to approximate end-to-end delays by

means of local measurements

traffic concept trafic engineering planning (1)

Documents