טכניון מכון טכנולוגי לישראל
הטכניון מכון טכנולוגי לישראל - בית הספר ללימודי מוסמכים  
M.Sc Thesis
M.Sc StudentOklander Boris
SubjectDelay Jitter Analysis in Communication Networks and
Real-time Systems
DepartmentDepartment of Electrical Engineering
Supervisor Professor Moshe Sidi
Full Thesis textFull thesis text - English Version


Abstract

Information flow in communication networks and real-time systems grows rapidly. One of the notable examples for this kind of flow is Voice over Internet Protocol (VoIP) streaming. This real-time traffic is sensitive to the delay jitter of the packets arriving at the receiver side. Delay jitter is the variation in packet transit delay caused by routing, queuing, scheduling and serialization effects on the path through the packet-switched network. The quality of the received information at the recipient side is strongly affected by delay jitter. Reduction of the delay jitter is generally done by incorporation of a playback mechanism called jitter buffer, which regulates the stream and prevents unacceptable levels of distortion. The delay jitter process and the jitter buffer mechanism are thus recognized as important, affecting the Quality of Service (QoS) subjects whose behavior merits investigation.

In this work we study the delay jitter process for a variety of packet arrival and service processes in classical queueing systems. Analytical expressions for delay jitter are derived and some surprising and non-intuitive results are explored. After studying statistical characteristics of the delay jitter, performance evaluation of jitter buffer is done. Particularly, we model the operation of jitter buffer by a finite capacity queueing system and analyze it to establish the relations between such quantities as initial playout delay, delay jitter and loss. These relations are used together with voice quality evaluation methodologies to optimally choose the controlling parameters of the jitter buffer. The analytical results developed in this work are applicable for incorporation in playout algorithms to achieve better voice quality.