The evolution of wireless telecommunication networks has experienced during the last decade an unprecedented development. From the third generation, this development has achieved considerable jumps either on bandwidth or on the types of offered services. However, this multitude of services poses incessantly the problem of quality of service (QoS), especially with the increased demand for these services.
Evaluation and improvement of performance in wireless cellular networks is a challenge for researchers. This research topic is approached in two complementary ways:
The main objective of this chapter is to propose models for managing queue to improve performance for a heterogeneous traffic (formed in real-time and non real time flows) to an end user in a cellular network. Moreover, the model studied is situated in relation to a classification of existing queue models in the literature.
2.1 Real Time and Non Real Time Services Classification
In recent years, the performance of mobile cellular telecommunication networks have been growing continuously by increasing the hardware capacity, and new generations of cellular networks offer more bandwidth resources. With this development, new generation of mobile wireless networks have to support multimedia applications that generate traffic having diverse QoS requirements.
We assume that we can always improve performance and guarantee the QoS by increasing microprocessor speed or even building specialized hardware to replace inefficient software components. However, for a predetermined set of available resources, it becomes important to take additional measures to effectively implement the QoS in network.
Typically service classification in new generations of cellular networks is based on the two parameters: delay and packet loss. The UMTS standard, for example, defines four QoS classes (Sanchez & Thioune, 2004) which are the conversational class, streaming class, interactive class and background class. These classes mainly differ in the way they balance between the transmission delay and reliability.
When classifying services according to their delivery requirements, the concept of Real Time (RT) and Non Real Time (NRT) services is introduced. Usually, RT services have been considered to impose strict delay requirements on the end to end communication. As a result, the involved network nodes in the RT traffic have to transfer the packets within a maximum tolerable delay. Due to these severe delay constraints, the error correction possibilities of this type of communication are very limited. On the other hand, NRT traffic is commonly considered as error sensitive, though with less demanding delay constraints than RT traffic. These characteristics of NRT traffic allow for link and also end-to-end level recovery mechanisms, enabling an error free delivery of the payload.
The above mentioned QoS classes in UMTS can be grouped by these two categories. Namely, the conversational and streaming traffic can be identified with RT services, whereas the interactive and background services belong to a NRT traffic pattern. For clarity and structuring purposes, this paper distinguishes between NRT and RT traffic. NRT (i.e. best effort) services require payload to be transferred error free, whereas delay requirements still allow for end-to-end error recovery mechanisms such as carried out by TCP. In contrast, RT services have delay requirements which exclude end-to-end retransmission mechanisms. Hence, they are using unreliable transport protocols like the UDP.
This classification into real time (RT) and non real time traffic can be adopted in all wireless cellular networks.