Professional Documents
Culture Documents
MTNL
CETTM
Contents
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
CETTM
Traffic Concepts
MTNL
Telephone Traffic
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 a pattern 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 ) can originate a call at any given moment. The duration of calls are not known.
CETTM
MTNL
Traffic Pattern
On a network, load & traffic pattern varies during the day with heavy traffic and low traffic durations
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 per hour
Time
CETTM
MTNL
CETTM
MTNL
One hour period starting at the same each day for which the Average Traffic Volume or Number of Call Attempts 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
CETTM
MTNL
Erlang:If one circuit is held continuously for one hour then the traffic carried by that circuit amounts to one Erlang ( 1 Traffic Unit ) Exchange with 2000 Subscribers of average BHCA 10,000 & CCR 60%. Calculate the BHCR. Avg. BH Calls = BHCA x CCR = 6000 Calls BHCR = Avg. BH Calls Total No. of Subscribers = 3
CETTM
MTNL
Times to Period A = S/T If C = total number of calls during the period T then Avg. holding time t = S/C hrs per call Therefore A = C * t/T
CETTM
MTNL
Example
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
CETTM
MTNL
Grade of Service
Recommended GoS = 0.002 i.e. 2 out of 1000 calls allowed to be lost 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 GoS = Lost Traffic / Offered Traffic= (A-A0) / A A - Offered traffic, A0 - Carried Traffic (A Ao) - Lost Traffic
CETTM
MTNL
Quality of Service
More general term than GoS includes other factors like Quality of Speech, Error free transmission capability, etc. Loss system: Circuit switched networks When overloaded, a user call is blocked and user has to make a retry Delay System:Packet switched (store & forward) network
Delay when extended beyond limits becomes a loss system In a store & forward network if the queue becomes full, then further requests have to be rejected
CETTM
MTNL
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
CETTM
MTNL
Queuing Model: Parameters for Delay System Service Delays Flow Control To prevent loss, queue of traffic is cleared to an acceptable limit Flow Control technique used to prevent loss of traffic
CETTM
MTNL
Traffic Engineering
Designing a cost effective network which provides the required Quality of Service under varied traffic conditions demands a formal scientific basis
Traffic Engineering provides a means to determine the quantum of exchange equipments required to provide a particular level of service for a given traffic pattern and volume.
CETTM
MTNL
Erlang-Bformula
A: User traffic described by offered traffic A N: Network described by number of channels n E: Quality-of-Service described by blocking probability E Robust to the traffic process
CETTM
MTNL
Economy of scale
CETTM
Technological developments
MTNL
Networking development Packet based transfer mode Packetized voice Wireless access networks Mixed core networks Photonic backbone networks Centralized & decentralized control
CETTM
Services development Differentiated services Narrowband & broadband Real-time services: Delay sensitive Jitter (delay variation) sensitive Non-real-time services Packet loss sensitive Best effort services
MTNL
CETTM
MTNL
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
QoS can only be guaranteed by ressource reservation End toend 1. Bandwidth based mechanism Separation:Implylowutilization>=highcost Minimum bandwidth guaranteed => worse case guarantee Sharing:Implyhighutilization>=lowcost 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
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
MTNL
Type 1: Load 0.1 erlang, mean service time 0.1 s Type 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 s Preemptive resume: W1 = 0.0056 s W2 = 14.46 s (twice as many type 1 jobs as of type 2)
Processor sharing: all users share the available capacity Generalized Processor sharing: maximum capacity for each user Robust to the service time (file size) Mean performance measures are the same as forErlangswaitingtimesystem This model is applicable for Best Effort traffic (Web traffic)
CETTM
MTNL
A service type is characterized by Qosparameters(discussedabove) Trafficcharacteristics 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
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
CETTM
MTNL
Performance monitoring
CETTM
MTNL
Simplifying assumption
Validation of models
Traffic Forcasting Estimation of parameters value
Relevant parameters
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
General and operational aspects Short,mediumandlongtermapplications Operationalrequirements Directmeasurementsvs.calldetailedrecords. Technical aspects Measurementsprinciples Criteria to choose the length of the read-out period: statistical confidence, stationarity of the process Normal and high loads Estimationofofferedtraffic Measurements requirements PSTN&N-ISDN , B-ISDN and SS No.7
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
MTNL
Forecasting of new services There are not historical data Techniques: market research, expert opinion and sectorial econometrics Combination of techniques, adjustments after implementation
CETTM
GOS Objectives:Principles
MTNL
Losses, delays..
MTNL
Traffic routing
Hierarchical or non-hierarchical Fixed or dynamic (time-, state- or eventdependent routing) Network traffic management controls Maintain the throughput under overload or failure conditions Protective or expansive controls
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
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
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 cost optimisation
CETTM
MTNL
No
Accepted
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 Packet-level traffic controls Similar to a multi-rate circuit-switched network
CETTM
MTNL
Accepted
No
Accepted
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 Packet-level traffic controls Similar to a multi-rate circuit-switched network
CETTM
MTNL
Framework Principles and definitions: CAC role Strategies for logical network configuration Packet-level traffic controls 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
CETTM
MTNL
Dimensioning of signalling networks Performance under failures and traffic overload Maximumdesignlinkloadmax Acceptable performance at load 2max
Allocation and dimensioning of intelligent network resources Particular features, e.g., mass calling situations Fast implementation of new services with uncertain forecast Quick and flexible procedures for allocation and dimensioning
CETTM
MTNL
Traffic controls Guidelines for the choice of control parameters Requirements on node-level overload controls Harmonisation principles: multivendor, multioperator
PERFOMANCE MONITORING
CETTM
MTNL
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