Telecommunication System Engeering EEE 354

By Mian Ahmed YaserAssistant Professor EECIIT Lahore

COMSATS Institute of information and technology Lahore 9/8/20151Mian Ahmed YaserTelecommunication Systems EngineeringEEE354SIMPLE TELEPHONE CONNECTION

8 point Full Mesh

Star Connection

Sources, Sinks, Networks, Nodal PointsSourcesCall is initiated at a traffic SourceSinksCall is terminated at a traffic SinkNetworksA means of connecting subscribersNodal PointsSwitchesTrunks or JunctionsTelephone lines connecting one switch to another switch is called a junction or trunk lineLINESTelephone lines connecting subscribers to an exchange are called lines or loopsTRUNKSNumber of lines connecting one exchange to another exchange are called trunks or junctionsCONCENTRATIONRatio of lines to trunks is called CONCENTRATIONe.g. Two cities with 100 subscribers each are connected via a 6 line junction/trunk, then the concentration is 100:6= 16:1

Local Area Vs. Toll AreaLocal AreaA subscriber can call another subscriber without incurring extra charges for call-Charged as flat monthly rate or as time metered callToll AreaAlso called a long distance call. Separate details are available in the bill

TRAFFIC ENGINEERINGDimensioning the routeDetermine the number of trunks required on a route or connection between exchanges

To dimension a route we must have an idea of its usageRoute UsageRoute Usage is defined by two parametersCalling RateNumber of times a route or traffic path is used per unit periodOr-Call intensity per traffic path during busy hourHolding TimeAverage duration of occupancy of one or more paths by callsTRAFFIC PATHA ChannelA Time SlotA Frequency BandA LineA TrunkA Switch orA Circuit Over which individual communications pass in sequenceCarried Traffic and Offered TrafficCarried Traffic is the volume of traffic actually carried by the switchOffered Traffic is the volume of traffic offered to the switchTraffic Intensity over 24 hours of the day

Traffic is RandomTo dimension a traffic path or size a switch we must know the traffic intensity of normal busy seasonThere are weekly variations and daily variationsWhat is the busy hour in the previous graph?

BHBusy HourContinuous one hour period lying wholly in the time interval concerned for which the call attempts/traffic volume is maximumPeak Busy HourBusy hour each day , it is not the same over a number of daysTime Consistent Busy Hour1 hour period starting at the same time each day for which the average traffic volume is greatest over days under considerationBSBH and ABSBHBSBH: (Busy Season Busy Hour) Busiest Clock hour of the busiest weeks of the year with a certain GOS criteria appliedABSBH: Average Busy Season Busy Hour (used for trunk groups) with a certain GOS criteria appliedGOS: Grade of Service: e.g. for the ABSBH load a call requiring a circuit in trunk group should encounter ATB (All Trunks Busy) no more than 1% of the timeTRAFFIC MEASUREMENTAggregate of Telephone Calls over a group of circuits with regards to the duration of calls as well as their number

Traffic Flow (A)=C x TC=Number of Calls during 1 hour periodT=Average holding time (Hours)A is dimensionlessExampleAverage Holding Time=2.5 min

Calling rate in the BH of a particular day =237

Average Traffic Flow=237 x 2.5= 592.5 Cm (Call Minutes)

Or 592.5/60=9.87 Call Hours (Ch)

Traffic Density and Traffic IntensityTraffic Density: Number of Simultaneous Calls at a given momentTraffic Intensity: Average Traffic Density during one hour period

We use Traffic Intensity for calculation of dimensions of switchesTRAFFIC INTENSITYIts unit is Erlang: (dimensionless unit)One Erlang means a circuit occupied for one hourFor a group of circuits, Traffic Intensity in Erlangs is the number of Call-Seconds per Second or Call Hours per Hour

TRAFFIC INTENSITY (Example)A group of 10 circuits with a traffic of 5 Erlangs means that half of the circuits are busy at the time of measurementTraffic Measurement in USUC: Unit Call or Hundred Call-second i.e. CCSSum of the number of busy circuits provided the busy trunks were observed once every 100 seconds (36 observations in one hour)

Cs, Ch, TU, EBHC, CCS and Erlang1 Cs One or more calls having an aggregate duration of one second1 Ch is One or more calls having an aggregate duration of one hour1 TU is the average intensity in one or more traffic paths carrying an average traffic of 1Ch in 1 hour (the busy hour)1 TU=1 ErlangCs, Ch, TU, EBHC, CCS and ErlangEBHC (Equated Busy Hour Call)1 EBHC is the average intensity in one or more traffic paths occupied in the BUSY HOUR by one two minute call or an aggregate duration of 2 minutesSo,1 Erlang=30 EBHC=36 CCS=60 Cm (assuming 1 hour time interval)Traditional Traffic Measurements on TrunksPeg Count (Calls offered)Usage (CCS or Erlangs carried)Overflow (Calls encountering ATB-All Trunks Busy)

From the above measurements following terms can be calculated:Blocking ProbabilityMean Traffic load carried by the trunk groupTraffic Measurements for Short Term Network ManagementNetwork Congestion is detectedCalls offered (peg count) and overflow calls are used to calculate ACH (Attempts per Circuit per Hour)CCH (Connections per Circuit per Hour)Normal ConditionsACH is approximately equal to CCHAbnormal ConditionsACH High, CCH NormalHigh Traffic but low congestion, Normal Holding times for connected calls, meaning that most of the switched calls completedACH High, CCH HighHigh Traffic with High Congestion, short Trunk Holding times showing uncompleted call attempts being switched

Blockage, Lost Calls, Grade of ServiceBlockageAn exchange serving 5000 subscribers is dimensioned for 500 as 10% subscribers are expected to use the service at a time. 501th subscriber will not be able to reach the other end even if the other end is idle. This is the Lost Call or Blocked callLost CallsLost Call Attempts due to the limited dimension of the switch are lost callsGrade of ServiceProbability of meeting Blockage during the BH is GOS. Typically 1% i.e. p=0.01 (First call attempt)GOS (Example)GOS= (Number of Lost Calls)/(Total Number of Calls Offered)If Number of Seizures=354, and Number of Lost calls =6Then, GOS=6/(354+6)p=0.017

AvailabilitySwitches have Lines and TrunksWe can call them Inlets and OutletsIf each Inlet has access to an Outlet, the Availability is FULLWhen not all inlets have access to an Outlet it is called Limited AvailabilitySmaller exchange switches may achieve Full AvailabilityLarger Exchanges do not have full availability. To improve Availability Grading is used which means inter connecting switching sub-groups to make the switching load more Uniform.Many new digital switches Full availability is achievedA Switch with Limited Availability

A Switch with Full Availability

Handling of Lost CallsThere are three methods of Handling Lost Calls:LCH (Lost Calls Held)- US standardIt is assumed that caller will immediately re-dial as soon as he gets a congestion signal. (Calls are waiting at the users telephone). It is assumed that lost calls extend the average holding time theoretically, and in this case holding time is 0 and all the time is waiting timeLCC (Lost Calls Cleared)-European StandardAssumes that User will hang up and attempt some time later if he hears a congestion signal. Re-attempt is considered a new call. Erlang formula is based on LCCLCD (Lost Calls Delayed)User is put in a queue if there is congestion and is handled as follows:LIFOFIFORandomInfinite and Finite SourcesInfinite SourceProbability of call arrival is constant and does not depend on the occupancy of the systemInfinite number of call arrivals with infinitely small holding timeFinite SourceArrival rate is proportional to the number of sources that are not already engaged in sending a callNumber of sources offering traffic to a group of circuits is comparatively small as compared to the number of circuits

Probability Distribution CurvesCalls originating and arrivals at an exchange closely fit a probability distribution curves of Poisson distributionMean and VarianceMean is the x coordinate of the center of the area under the probability-density curve for populationStandard Deviation or Variance is the measure of how much the values or population are dispersed about the center or mean of the curve

Variance to Mean ratio (VMR)

Smooth, Rough and Random TrafficSmooth TrafficSmooth Traffic Behaves like filtered Random TrafficSmooth Traffic is the Traffic at the Exchange outletsFiltering is done by the Call blockage during the BHSmooth Traffic is characterized by a positive Binomial Distribution function or Bernoullis Distribution

Example of Bernoullis DistributionSubscribers make calls independently of each otherEach has a probability of p of getting engaged in a conversationIf n subscribers are examined, the probability that x of them will be engaged is

Smooth TrafficSmooth Traffic is assumed for small number of subscribers-200 is a breakpoint

Binomial Distribution of Rough TrafficRough Traffic is characterized by a negative indexLet k=Number of Traffic Sourcesq=Occupancy per source (between 0 and 1)

Where,R =Probability of finding x calls in progressRough Traffic is used for dimensioning Toll Trunks with alternate routing

Bernoullis Formula

Probability of finding x of s lines busyPoisson probability distribution can be derived from Binomial distribution,assuming that the number of subscribers (s) is very large And assuming that calling rate per subscriber is very low i.e. h is lowSuch that (s.h) =m is constantAssuming s tends to infinity, applying the limit

Call Holding TimesCall Holding Times are considered to have a negative exponential distribution

Where t/h is the average holding timeP is the Probability that a call lasts longer than an arbitrary time t

Smooth Random and Rough Traffic Probability Distributions

Erlang and Poisson Traffic FormulaWhile dimensioning a route, we need to find out the number of circuits that serve the routeSeveral formulas are available, but we select the formula on the basis of the following factors:Call Arrivals and Holding Time DistributionNumber of Traffic SourcesAvailabilityHandling of lost Calls

Erlang B Formula

Assumptions of Erlang B FormulaTraffic originates from an infinite number of sourcesLost Calls are Cleared assuming a zero Holding TimeNumber of Trunks is LimitedFull Availability existsTime Congestion and Call CongestionTime CongestionTime fraction of the hour during which all Trunks are busy simultaneouslyCall CongestionNumber of Calls that fail at first attemptAlternative Traffic Formula ConventionsPoisson Formula

Erlang B Formula

Erlang C Formula

Binomial Formula

Notations in the formulasA=Expected Traffic Density, expressed in busy hour erlangsP=Probability that the calls will be lost or delayed because of insufficient channelsn=number of channels in the group of channelss=number of sources in the group of sourcesp=probability that a single source will be busy at the instant of observation=A/sx=A variable representing a number of busy sources or busy channelsAssumptions for formulasPoisson FormulaErlang B FormulaErlang C FormulaBinomial FormulaInfinite SourcesInfinite SourcesInfinite SourcesFinite SourcesEqual Traffic Density per sourceEqual Traffic Density per sourceEqual Traffic Density Per sourceLost Calls held (LCH)Lost Calls cleared (LCC)Lost Calls Delayed (LCD)Lost Calls Held (LCH)Exponential Holding TimesCalls Served in the order of arrivalAssignmentUse MATLAB to implement Traffic formulas as per section 6.2 of Chapter 1 (TSE-Freeman)Erlang B FormulaPoisson FormulaErlang C FormulaCompare results with the book

Waiting System (Queueing)Based on LCD (Lost Calls Delayed)Serving Time:The a call takes to be served from the moment of arrival in the queue to the moment of being served by the switchIt is assumed that the call arrivals are random and PoissonianTraffic Engineer calculates the number of serving circuits or TrunksFollowing parameters are provided to the Traffic Engineer:Offered TrafficSize of QueueSpecified Grade of ServiceQueue DisciplineMethod by which a waiting call is selected to be served from the pool First Come First ServedIt requires equipment to keep order of the queue, so it is costlyRandomWaiting time of a call is disregarded and call is selected randomlyLast Come First ServedBulk Service DisciplineBatches of waiting calls are admittedPriority Service DisciplinePreemptiveNon Preemptive

Queuing System-GoSIn Queuing system, the Grade of Service is the Probability of Delay P(t)P(t) is the probability that a call is not immediately served and is delayed for a time more than t GoS based on Probability of Delay is used with Full Availability And Poissonian Call Arrival processProbability of Delay GoS is found using Erlang-C formula which assumes an infinity long queueGrade of Service is also expressed based onAverage delay on all calls orLength of the queue

Server-Pool TrafficServer Pools are groups of traffic resources which are used on shared basis like:Signaling RegistersOperator PositionsAS = (AT TS C)/TCwhere AS = server-pool traffic in erlangsAT = total traffic served in erlangsTS = mean server-holding time in hoursTC = mean call-holding time in hoursC = call-attempt factor (dimensionless)Server-Pool TrafficService requests that cannot be served are placed in a queue and served on FIFO basisServer Pool Traffic is directly related to:Offered TrafficServer Holding TimeCall Attempt FactorAnd inversely proportional to Call Holding TimeServer Pool-Total Traffic ServedTotal Traffic Served is the Total offered Traffic that requires the services of the specific Server-Pool for some portion of the callMean Server Holding Time is the Arithmetic Average of all Server Holding Times for the Specific Server-PoolTS = a T1 + b T2 + +k Tn where TS = mean server-holding time in hoursT1, T2, . . . , Tn = individual server-holding times in hoursa,b, . . . , k = fractions of total traffic served (a + b + +k = 1)Server-Pool ExampleDetermine the mean DTMF receiver-holding time for a switch where subscribers dial local calls using a 7-digit number and long-distance calls using an 11-digit number. Assume that 70% of the calls are local calls, the remainder are long-distance calls, and the typical signaling register holding times found in Table 1.1 are applicable.

Ts = (0.7)(8.1 s) + (0.3)(12.0 s) = 9.27 s

Call Attempt FactorsC adjust offered Traffic Intensity to compensate for call attempts that do not result in completed callsSo, Call Attempt Factor is inversely proportional to the fraction of completed callsC = 1/kwhere C = call-attempt factor (dimensionless)k = fraction of calls completed (decimal fraction)Server Pool TrafficDetermine the Call-Attempt Factor the data in Table 1.2, where 70.7% of the calls were completed

C = 1/k= 1/0.707= 1.414

Dimensioning and EfficiencyIf the number of trunks equals the Erlang load then 100% Efficiency is achievedAll trunks will be busy all the time or at least during the BHThis will not allow even a few moments for a trunk to be idle, while the switch decides to take the next callBut the customers will be unhappyDimensioning and Efficiency-GoSGoS is a measure of Customer satisfactionExampleThere are 100 trunks between cities X and Y. Tarrif depends on the Erlangs of carried trafficAssume $1 for Erlang-hour TrafficIf the upper limit of the Traffic on this route is 100 Erlangs, the maximum return on investment will be $2400 for one dayBut the customers are unhappy because they have to wait a lot before getting connected to the destinationDimensioning and Efficiency-GoSHow do we improve customer satisfaction?Take GoS=0.001, For 100 trunks 75.4 Erlangs of traffic will be supported75.4 Erlangs will earn $75.24, which is less than $2400 per dayNow reduce the GoS to 0.01, 100 trunks will support 84.06 Erlangs, meaning $84.06 for the BH per hourSo revenue has improved by reducing GoSAnother approach for saving money is to keep the traffic (Erlang load) constant and reduce the number of trunks as GoS is reducede.g. 70 Erlang of Traffic at p=0.001 requires 96 trunks and p=0.01 requires only 86 trunksAlternative RoutingAlternative Routing Improves efficiencyExample: X, Y and Z are three exchanges in their respective serving areasLet GoS=0.005It would require 67 trunks to carry 50 erlangs of traffic during the BH to meet that grade of service between X and Y Suppose that we reduced the number of trunks between X and Y , still keeping the BH traffic intensity at 50 erlangsWe would thereby increase the efficiency on the XY route at the cost of reducing the grade of serviceWith a modification of the switch at X, we could route the traffic bound for Y that met congestion on the XY route via Z.Then Z would route this traffic on the ZY linkThis is alternative routing in its simplest formCongestion probably would only occur during very short peaky periods in the BH, and chances are that these peaks would not occur simultaneously with peaks in traffic intensity on the ZY route. Furthermore, the added load on the XZY route would be very smallSome idea of traffic peakiness that would overflow onto the secondary route (X + Z + Y) is shown in the figure

One of the most accepted methods of dimensioning switches and trunks using alternative routing is the equivalent random group (ERG) method developed by Wilkinson which uses the mean M and the variance VHere the overflow traffic is the lost traffic in the Erlang B calculationsLet M be the mean value of that overflow and A be the random traffic offered to a group of n circuits (trunks). Then

When the overflow traffic from several sources is combined and offered to a single second (or third, fourth, etc.) choice of a group of circuits, both the mean and the variance of the combined traffic are the arithmetical sums of the means and variances of the contributorsThe basic problem in alternative routing is to optimize circuit group efficiency (e.g., to dimension a route with an optimum number of trunks)Thus we are to find what circuit quantities result in minimum cost for a given grade of service, or to find the optimum number of circuits (trunks) to assign to a direct route allowing the remainder to overflow on alternative choicesThere are two approaches to the optimizationThe first method is to solve the problem by successive approximations, which is better done by computersManual ApproachesEfficiency versus Circuit Group SizeA circuit group refers to a group of circuits performing a specific functionFor instance, all the trunks (circuits) routed from X to Y in Figure 1.8 make up a circuit group, irrespective of size.If we assume full loading, it can be stated that efficiency improves with circuit group size. Given p = 0.001, 5 erlangs of traffic requires a group with 11 trunks, more than a 2 : 1 ratio of trunks to erlangs, and 20 erlangs requires 30 trunks, a 3 : 2 ratioNote how the efficiency has improvedOne hundred twenty trunks will carry 100 erlangs, or 6 trunks for every 5 erlangs for a group of this size. Figure 1.10 shows how efficiency improves with group size.

Basic Network ConfigurationsA network in telecommunications may be defined as a method of connecting exchanges so that any one subscriber in the network can communicate with any other subscriberAssume that subscribers access the network by a nearby local exchangeThus the problem is essentially how to connect exchanges efficientlyThere are three basic methods of connection in conventional telephony: (1) mesh, (2) star, and (3) double and higher-order starThe mesh connection is one in which each and every exchange is connected by trunks (or junctions) to each and every other exchange as shown in Figure 1.11AA star connection utilizes an intervening exchange, called a tandem exchange, such that each and every exchange is interconnected via a single tandem exchange as in Figure 1.11BA double-star configuration is one where sets of pure star sub-networks are connected via higher-order tandem exchanges, as shown in Figure 1.11CThis trend can be carried still further to hierarchical networks

Mesh connections are used when there are comparatively high traffic levels between exchanges, such as in metropolitan networksStar network are applied when traffic levels are comparatively lowAnother factor that leads to star and multiple-star network configurations is network complexity in the trunking outlets (and inlets) of a switch in a full mesh. For instance, an area with 20 exchanges would require 380 traffic groups (or links), and an area with 100 exchanges would require 9900 traffic groups.A one-way group is best defined considering the connection between two exchanges, A and B. Traffic originating at A bound for B is carried in one group and the traffic originating at B bound for A is carried in another group, as shown in the following diagram:

Most networks compromises of mesh and star configurations. For instance, outlying suburban exchanges may be connected to a nearby major exchange in the central metropolitan areaThis exchange may serve nearby subscribers and be connected in mesh to other large exchanges in the city properAnother example is the citys long-distance exchange, which is a tandem exchange looking into the national long-distance network, whereas the major exchanges in the city are connected to it in meshAn example of a real life compromise among mesh, star, and multiple-star configurations is shown in Figure 1.12.A Telephone Network Serving a Small City

Higher Order Star Network

Higher Order Star NetworkAnother high-order star network is shown in Figure 1.11C. In these types of networks the higher layers are given more importance than the lower layers i.e. a hierarchical networkThere are three levels or ranks of exchanges in the figure.The smallest blocks in the diagram are the lowest-ranked exchanges, which have been marked with a 3 to indicate the third level or rank.There are restrictions or rules of traffic flow. As the figure is drawn, traffic from 3A1 to 3A2 would have to flow through exchange 2A1. Likewise, traffic from exchange 2A2 to 2A3 would have to flow through exchange 1ATraffic from any A exchange to any B exchange would necessarily have to be routed through exchange 1A.High Usage RouteThe next consideration is the high-usage (HU) route. For instance, if we found that there were high traffic intensities (e.g., >20 erlangs) between 2B1 and 2B2, trunks and switch gear might well be saved by establishing a HU route between the two (shown by a dashed line in Figure 1.13).Thus we might call the high usage route a highly traveled shortcut.HU routes could be established between any pair of exchanges in the network if traffic intensities and distances involved proved this strategy economicalWhen HU routes are established, traffic between the exchanges involved will first be offered to the HU route, and overflow would take place through a last choice route or, as shown in Figure 1.13, up to the next level and downIf routing is through the highest level of higher-order star network, we call this route the final route, a hierarchical network termVariations of Traffic FlowIn networks covering large geographic expanses and even in cases of certain local networks, there may be a variation in time of day of the BH or in the direction of traffic flowIn the United States, business traffic peaks during several hours before and after the noon lunch period on weekdays, and social calls peak in early eveningTraffic flow tends to be from suburban living areas to urban centers in the morning, and the reverse in the evening.In national networks covering several time zones where the differences in local time may be appreciable, long-distance traffic tends to be concentrated in a few hours common to BH peaks at both endsIn such cases it is possible to direct traffic so that peaks of traffic in one area fall into valleys of traffic in another (non-coincident busy hour)The network design can be made more economical if configured to take advantage of these phenomena, particularly in the design and configuration of direct routes versus overflowOne-way and Two way (both-way) CircuitsIn one-way circuits traffic from A to B is assigned to one group of circuits, and traffic from B to A is assigned to another separate groupIn both-way (or two-way) operation a circuit group may be engineered to carry traffic in both directionsThe individual circuits in the group may be used in either direction, depending on which exchange seizes the circuit first.In engineering networks it is most economical to have a combination of one-way and both-way circuits on longer routesSignaling and control arrangements on both-way circuits are substantially more expensiveHowever, when dimensioning a system for a given traffic intensity, fewer circuits are needed in both-way operation, with notable savings on low-intensity routes (i.e., below about 10 erlangs in each direction)For long circuits, both-way operation has obvious advantages when dealing with a non-coincident BHDuring overload conditions, both-way operation is also advantageous because the direction of traffic flow in these conditions is usually unequal.The major detriment to two-way operation, besides its increased signaling cost, is the possibility of double seizure. This occurs when both ends seize a circuit at the same time. There is a period of time when double seizure can occur in a two way circuit; this extends from the moment the circuit is seized to send a call and the moment when it becomes blocked at the other endSignaling arrangements can help to circumvent this problemLikewise, switching arrangements can be made such that double seizure can occur only on the last free circuit of a groupThis can be done by arranging in turn the sequence of scanning circuits so that the sequence on one end of a two-way circuit is reversed from that of the other endOf course, great care must be taken on circuits having long propagation times, such as satellite and long undersea cable circuitsBy extending the time between initial seizure and blockage at the other end, these circuits are the most susceptible, just because a blocking signal takes that much longer to reach the other endQuality of ServiceQuality of service also means how happy the telephone company (or other common carrier) is keeping the customer. For instance, we might find that about half the time a customer dials, the call goes awry or the caller cannot get a dial tone or cannot hear what is being said by the party at the other endAll these have an impact on quality of service.QoS means different for different servicesThe transmission engineer calls quality of service customer satisfaction, which is commonly measured by how well the customer can hear the calling party. It is called loudness rating and is measured in decibels. Lost calls or blockage certainly constitute another measure of service qualityIf this is measured in a decimal quantity, one target figure for grade of service would be p = 0.01. Other items listed under service quality are: Delay before receiving dial tone (dial-tone delay). Post-dial(ing) delay (time from completion of dialing a number to first ring of telephone called). Availability of service tones (busy tone, telephone out of order, ATB, etc.). Correctness of billing. Reasonable cost to customer of service. Responsiveness to servicing requests. Responsiveness and courtesy of operators. Time to installation of new telephone, and, by some, the additional services offered by the telephone company83Furthermore, each item on the list can be quantified, usually statistically, such as loudness rating, or in time, such as time taken to install a telephone