# Teletraffic engineering/How is telephony traffic simulated

Module by: Nomfundo N. Dlamini

## Summary

Telephone networks are very large and complex networks. This is due to the technological development and the ever demanding requirement for multiple services that these networks need to support. This poses a need to forecast the behavior of a network that should a certain condition prevail that a specific remedy be available. This can only be done by simulation, which is a modeling tool in great use today. It is an economic in a way that financially and chronologically represents the system's traffic behavior. By building a model, the computing power of the simulation is accessible. Not only is it accessible, there is also an increasing availability of low cost computer hardware and the development of specialized powerful and flexible simulation languages contributes the formation of a simulation for a preferred modelling technique.

## What is simulation?

Simulation can be defined as the imitation of a system or environment in order to predict the actual behavior. In telecommunication systems, it is the most quantitative modelling technique used and its purpose is to model large, complex stochastic systems for forecasting or performance measurement purposes. Simulation is a cost-effective way of pre-testing proposed systems, plans, or policies before incurring the expense of prototypes, field tests, or actual implementations 

Network traffic simulation usually follows the following four steps

• Modelling the system as a dynamic stochastic (i.e. random) process
• Generation of the realizations of this stochastic process
• Measurement of Simulation data
• Analysis of output data

## How is telephony traffic simulated?

Telephony traffic is simulated by using two simulation methods 

1. Discrete/Discrete event simulations (next event approach)
2. Continuous simulations (fixed time increment approach)

#### Discrete simulations

Discrete simulations are due to occurrence of an event. The system is dynamic and stochastic. It contains a number of states, modelled using a set of variables. A change in value of variables represents an event and the system's state changes. State equations are used in the representation of discrete simulations. The dynamic nature of the system gives it the characteristic of being in motion, thus constantly changing its position.

When designing a discrete system, there has to be an object and movement of the object through the system. The object may possess attribute values and may interact other certain conditions, leading to an event taking place which in turn changes the state of the system. Since a discrete simulation model may contain contain different objects which in turn lead to different events, a scheduler has to be used. A scheduler performs the complex task of keeping track of all objects and scheduling all events.An example of a discrete simulation model is a queueing system consisting of one server and one queue. Consider customers paying at a supermarkert till being served by the shop attendant. The assumption made is that they stand in a queue and are served by shop attendant at some specifired service time distribution and then leave system.Further, if two events are defined, an arrival and a departure event, also two state variables (the length of the queue and whether the server is idle or busy) we can draw up a possible sequence of events and system state changes.

```Example
```

Table 1: Sequence of events for a simple queueing system (adopted from ). File:QueueSystem1.JPG

#### Continous simulations

These are usually employed when state variables of the system change continuously. There is a mechanism to advance time in fixed increments. Simulation models contain one or more state equations which attempt to model variations in the state of the system over time. The equations are differential equations that track the system with reference to time.

#### Exercise

What type of traffic simulation model would you classify an SMS service? Support your answer.

Solution

The SMS service has the ability to queue during blocking in the network. So it is a discrete simulation model.

## Implementation and applications of Simulations

Simulations of a telecommunications system are usually implemented in a computer program. The program can be written in a simulation-specific language or in a general-purpose language. Simulation-specific languages such as OPNET or GPSS are rapid for development. However, sometimes the flexibility of a general purpose language may be preferred --- a well-known example is the ns-2 simulator ns-2 written in C++. There are many applications that can make use of simulation as an analytical tool. While simulation does require extensive resources, it is still a relatively cost-effective method of pre-testing potential systems.It helps further in confirming and verifying the performance of implemented systems.

Simulations models come with a lot of advantages which are stated as follows.

• Simulation modelling may be applied to the analysis of a very wide variety of systems. Most complex, real-world systems with significant random elements cannot be accurately described by a mathematical model, which can be evaluated analytically. Simulation is often the only type of investigation possible.
• Simulation modelling allows the detailed study of the interactions between system components. It allows the manager or analyst to make detailed decisions about the system and then judge the impact and consequences of these decisions. The model may be used for detailed sensitivity analyses to decide which controllable variables have the greatest impact on system performance.
• Alternative system designs or operating policies can be compared to estimate the system performance. Thus we can choose the best design.
• Simulation models and the principles of the simulation technique are often easy to understand. It is also easier to convince the non-simulation expert of the validity of the model. The simulation approach encourages active participation by management and operating staff in the modelling exercise.
• The process of building a simulation model may provide valuable insight and understanding about the system characteristics and how it operates. The value of this advantage is often underestimated.

Since simulations are merely a way of predicting the network performance, problems are bound to arise since this is not a real case scenario. What follows are some of the problems that may be encountered when running simulations.

• Selecting the correct level of detail (or level of abstraction) for a simulation is a difficult problem. Too little detail can produce simulations that are misleading or incorrect, but adding detail requires time to implement, debug, and later change. Furthermore, it slows down simulation and can be a distraction from the research problem at hand.
• Running effective simulations comes with experience. This means it might actually take longer to get the correct results which are a through reflection of the network performance.
• Accurate simulation model development requires extensive resources.

## Statistical Issues in Simulation Modelling

#### Input Data

Simulation models are generated from a set of data taken from a stochastic system. It is necessary to check that the data is statistically valid by fitting a statistical distribution and then testing the significance of such a fit. Further, as with any modelling process, the input data's accuracy must be checked and any outliers must be removed.

#### Output Data

When a simulation has been completed, the data needs to be analysed. The simulation's output data will only produce a likely estimate of real-world events. Methods to increase the accuracy of output data include: repeatedly performing simulations and comparing results, dividing events into batches and processing them individually, and checking that the results of simulations conducted in adjacent time periods “connect” to produce a coherent holistic view of the system.

#### Random numbers

As most systems involve stochastic processes, simulations frequently make use of random number generators to create input data which approximates the random nature of real-world events. Computer generated [random numbers] are usually not random in the strictest sense, as they are calculated using a set of equations. Such numbers are known as pseudo-random numbers. When making use of pseudo-random numbers the analyst must make certain that the true randomness of the numbers is checked. If the numbers are found not to behave in a sufficiently random fashion, another generation technique must be found. Random numbers for the simulation are created by a random number generator.