The evolution of applications on wireless networks has grown in recent years, due to
the increased number of smartphone users, tablets and others. The availability of
demanding services such as video transmission, affects Quality Experience (QoE) and
Quality of Service (QoS) provided to domestic users and trade, this had stimulated the
study of new resource management techniques networks, aiming to provide quality
services to a customer each increasingly demanding. This thesis presents a
methodology Intelligence Artificial using a Bayesian network with a hybrid evaluation
strategy analyzing the behavior metrics QoE and QoS in the LAN network design
wireless. The diversity of the place of Measurements chosen compound materials such
as brick, glass, wood and concrete. It was necessary first to map all the points to be
measured before and after deliberately placing each barrier outdated the signal. Metrics
as level Receiver Signal Strength Intensity signal (RSSI) Jitter, delay end to end
network for the video transmission, PeakSignal-to-NoiseRatio (PSNR) and Structural
Similarity (SSIM) were collected during the Measurements. And using the Bayesian
Network inferences were made for each metric and could not find satisfactory results
for the proposed solution assist the wireless network planning in indoor environments.
Enabling demonstrate that up to 10 meters away from the transmitter, the signal has its
best power, and delay metrics in order to have more than 65% probability that the lower
delay range and following this optimum performance the Jitter has more than 65%
probability in this lower range. And the QE metrics, PSRN and SSIM have a similar
behavior and has more than 80% probability of getting your greater value, and
consequently the video has its best reception. These results show that does not preclude
the use of this proposal in other situations.
|
