Teletrac Engineering

January 2013

Teletrac Engineering

January 2013

1 / 18

Trac Engineering

Trac engineering uses statistical techniques such as queuing theory to predict and engineer the behavior of telecommunications networks such as telephone networks or the Internet.
One of the important steps of teletrac engineering determines number of channels required on a route or a connection between two MSCs1 . Another important steps of teletrac engineering is to ensure the desired

Mobile switching centre server which is a 3G term

Teletrac Engineering January 2013 2 / 18

Busy hour
Busy hour - uninterrupted period 60 min during which the trac volume is highest Used for trac dimensioning Can be: xed/mobile In heterogeneous networks busy hours for dierent trac types may not coincide In cellular network, busy hour occurs at dierent time for dierent cells

Figure : Typical daily trac usage in a cellular system.

Teletrac Engineering January 2013 3 / 18

Teletrac Engineering

Service Level Service Level can be divided into two main areas :
1

Dial tone delay: The maximum waiting time to hear a dial tone after removing the hand-set from the hook. Service blocking probability:
the probability that service delay will exceed some specic value or The probability that the call will be denied or blocked

Teletrac Engineering

January 2013

4 / 18

Teletrac Engineering

Service or call blocking probability is known as the grade of service (GoS) GoS = GoS = Block Call Total Call (1) (2)

10 1 Block Call = = Serviced calls + Blocked calls 380 + 10 39

Teletrac Engineering

January 2013

5 / 18

Basic Queueing Theory

Figure : Number of customers as a function of time (t ): Number of customers who have come to the system and not rejected in the interval (0, t) (t ): Number of customers who have left the system and completed served in (0, t) N(t): (t ) (t ) Number of customers in system at time t (t ): Total time that all customers together have spent in the system during the interval
Teletrac Engineering January 2013 6 / 18

Basic Queueing Theory - Birth-Death Processes

A Birth-Death is a special type of discrete-time or continuous time Markov Chain with the restriction that at each step, the state transitions, if any, can occur only between neighbouring states. The underlying Markov process representing the number of customers in such systems is know as a birth-and-death process. The birth-death terminology is used to represent increases and decreases in the population size. The corresponding events in queueing systems are arrivals and departures.

Teletrac Engineering

January 2013

7 / 18

Basic Queueing Theory - Birth-Death Processes

Again using the birth (arrival)-death (departure) terminology, when the population size is n, let n and n be the innitesimal transition rates (generators) of birth and death, respectively. When the population is the number of customers in the system, n and n indicate that the arrival and service rates depend on the number in the system. If the process is a Birth-death process and if the current state Xn is i, then the above condition implies that the next state Xn+1 can only be i-1, i or i+1. Based on the properties of the Poisson process, i.e., when arrivals are in a Poisson process and service times are exponential, we can make the following probability statements for a transition during (t, t + t ]:
Teletrac Engineering January 2013 8 / 18

Basic Queueing Theory - Birth-Death Processes

birth(n 0): P (one birth) P (no birth) P (more than one birth) death(n > 0): P (one death) P (no death) P (more than one death) = = = n t + o (t ) 1 - n t + o (t ) o (t ) = = = n t + o (t ) 1 - n t + o (t ) o (t )

t) where o (t ) is such that o( t n 0 as t 0. Since a continouse-time process is being considerd, we need to focus on changes in the process over time interval o (t ) as o (t ) 0. Let k be the birth rate in state k2 Similarly, let k be the death rate in state k.
2

Note that is these statements the o (t ) terms do not specify actual values. In each of the two cases, the o (t ) terms sum to 0 so that the total probability of the three events is equal 1.
Teletrac Engineering January 2013 9 / 18

Basic Queueing Theory

The innitesimal transition rates of previous equation lead to the following generator matrix for birth-and-death process model of the queueing systems: 0 0 1 (1 + 1 ) 1 A= 0 ( + ) 2 2 2 2 . . . . . . . . . . . . .. .

The generator matrix A leads to the following forward Kolmogorov equations for Pin (t )(Pin (t ) Pn (t ) and insert the initial state i only when needed.) P0 = 0 P0 (t ) + 1 P1 (t ), Pn (t ) = (n + n )Pn (t ) + n1 Pn1 (t ) + n+1 Pn+1 (t ), n = 1, 2, . . .
Teletrac Engineering January 2013 10 / 18

Basic Queueing Theory

Figure : State Transition Diagram (Birth-Death)

Teletrac Engineering

January 2013

11 / 18

Basic Queueing Theory - Another Way

Teletrac Engineering

January 2013

12 / 18

Basic Queueing Theory - Another Way

Let Q(t) be the number of customers in the system at time t. Dene Pik (t ) = P [Q (t ) = k |Q (0) = i ] Considering transitions over a time interval t as t 0 , we get that P {state k to state k+1 in time t } = Pk ,k +1 = k (t ) P {state k to state k-1 in time t } = Pk ,k 1 = k (t ) P {state k to state k in time t } = Pk ,k = 1 (k + k )(t ) P {other transitions in t } = 0 Considering the transitions of the process Q(t) for the state transitions between the time instant t and the time instant t + t we have (in other words (t, t + t ]) we have
P0 (t + t ) = P0 (t )[1 0 t ] + P1 (t )1 t Pk (t + t ) = Pk (t )[1 (k k )t ] + Pk 1 (t )k 1 t + Pk +1 (t )k 1 t with
k =0

Pk (t ) = 1
Teletrac Engineering January 2013 13 / 18

Basic Queueing Theory

Figure : State Transition Diagram (Birth-Death)

Teletrac Engineering

January 2013

14 / 18

Erlang-B Example
A cell site has 5 FDMA radios. The average call origination rate is 60 calls per hour. If the call holding times are distributed exponentially with an average of 90 sec, calculate the GOS Average birth rate = 60 calls/hour
1 H

Average death rate = Trac Intensity A=

= =

1 90/3600 60 40

= 40 calls/hour

= 1.5 Erlangs

GoS AC /C ! B [A = 1.5, C = 5] = C Ai
i =0 i !

1.55 /5!
1.50 0!

1.51 1!

1.52 2!

1.53 3!

1.54 4!

1.55 5!

0.0142

Teletrac Engineering

January 2013

15 / 18

ErlangB Formula

Teletrac Engineering

January 2013

16 / 18

Cell Capacity

Average number of MSs requesting service (Average arrival rate): Average length of time MS requires service (Average holding time): T Oered load: A = T where a is in Erlangs e.g., in a cell with 100 MSs, on an average 30 requests are generated during an hour, with average holding time T=360 seconds Then, arrival rate =30/3600 requests/sec A completely occupied channel (1 call-hour per hour) is dened as a load of one Erlang, i.e., A=
30 calls 3600 sec

360 sec call

= 3 Erlangs

Teletrac Engineering

January 2013

17 / 18

Capacity of a Cell

This is Erlang B formula B(C, A) In the previous example, if C = 2 and A = 3, the blocking probability B(2, 3) is AC /C !
C Ai i =0 i ! 32 /2! 2 3i i =0 i !

B [2, 3] = =

= 0.529

So, the number of calls blocked 30 0.529 = 15.87

Teletrac Engineering

January 2013

18 / 18