Because of the computer network’s explosion-like development a big number of users communicate
at the same time. The data traffic could be easier and smoother, if it would be
fixed in space and constant in time. However neither of these conditions are guaranteed.
Due to the spacial movement we require wireless connections. At the section 2 we present
a review about these networks, among others we discuss Wi-Fi, LTE and 5G, and moreover
we explain the traffic’s handover among different channels, namely the origin and the
The traffic created in the network is dynamically changing in time. We should scale our
network according to the users. We can model the in-time changing processes with difference
equation systems, whose parameters depend on the users’ habit and on their download
rate. From the engineering point of view it is important to set our network to the required
reference value, so we would like to stabilize and control. For this purpose we explain the
mathematical background at section 3. One known model from the literature is reviewed
and generalized at section 4. However, this model is not fully capable of describing a real
life process, thus we created our new, own model. We explain at section 5, how to design
a network with the help of this model. Obviously because of the parameters’ estimation
value it may occur that the system will be overloaded. There are a lot of solutions to avoid
this. On one hand it is possible to stabilize our network to the required reference rate,
so that those users who cause the overload will be removed. However, this is not a user
friendly solution, that is why at section 6 we implement a new device for this problem.
This is the so called CAPEX-OPEX design.
Since in real life the connections are established with some kind of delay, we need to study
them. This is described at section 7, where in our previous stable network’s latency are
grown, thus bigger and bigger vibrations are generated that causes the network’s overload.
Thus during the design phase, the devices’ ad-hoc way implementation is not possible, so
we need to estimate the parameters as exactly as we can. It is required to be prepared for
a potential overload’s optimal management, and we should count with the real life process’
latency. We preformed these kind of investigations in this thesis with the help of our model,
furthermore we performed simulations with our program written in Wolfram Mathematica.