User Performance Modeling for Network Optimization

No Thumbnail Available
Journal Title
Journal ISSN
Volume Title
Sähkötekniikan korkeakoulu | Master's thesis
Ask about the availability of the thesis by sending email to the Aalto University Learning Centre oppimiskeskus@aalto.fi
Date
2015-12-14
Department
Major/Subject
Networking Technology
Mcode
S3029
Degree programme
TLT - Master’s Programme in Communications Engineering (TS2005)
Language
en
Pages
67 + 8
Series
Abstract
With the advancement in networking technology, the growth of interactive applications have risen dramatically. Many interactive activities such as teleoperation, web search, business applications, video conferencing, remote desktop control and gaming are very popular in the recent years. The effectiveness and ease of access of these applications depend on the quality of service (QoS) provided by the network. QoS parameters such as bandwidth, loss rate, delay, jitter and bit rate are examined to optimize the network. However, user performance in accomplishing the goals of these activities are not taken into account for optimization of network. In this thesis, we define and model the user performance with respect to QoS parameters. However, user performance is difficult to model because it is user-specific activity and subsumes multiple interrelated variables like success, speed and accuracy. We use first person shooter game as the platform for the user performance modeling with reference to Fitts’ law pointing task. User performance model exerts speed and accuracy in the form of the user metrics for this application. The result of the experiment shows that loss rate and latency have impact on the user performance metrics: (1) Throughput of pointing task and efficiency of first person shooter game linearly varies with latency and loss rate; (2) The relation between throughput and efficiency has positive correlation. Statistical testing, analysis of variance (ANOVA) is used to determine the effect of latency and loss on user performance. On the basis of data analysis, user performance modeled as a linear function QoS parameter. Furthermore, we demonstrate the user performance models can contribute in the network optimization by using the dynamic QoS management. In addition, we try to relate the user performance metrics as the objective parameters of the Quality of Experience.
Description
Supervisor
Oulasvirta, Antti
Thesis advisor
Oulasvirta, Antti
Keywords
quality of experience, quality of service, network optimization, Efficiency, speed-accuracy, user performance
Other note
Citation