| Abstract |
In recent years, Internet technology has emerged as a global communication means. New services are defined in order to evolve in a true integrated services network. These new services are ought to provide a level of quality to users. On the other hand, wireless data networks are evolving with tremendous rates and are becoming an important part of communications. Up to now, the main application of most mobile radio systems, like the pan-European GSM (Global System for Mobile communication), has been mobile telephony. The impressive growth of cellular mobile telephony as well as the number of internet users promises an exciting potential for a market that combines both innovations: cellular wireless data services. Within the next few years, there will be an extensive demand for wireless data services. More precisely, users will demand high performance wireless access to the Internet. An important research issue is to extend the QoS models and the algorithms developed for the wireline networks for the wireless networks as well. In order to do so, the differences between the wireless and the wireline environment must be defined. The fundamental differences are the following: Channel capacity is varying dynamically. The channel errors depend on the position of the mobile station and are bursty in nature. Mobile stations are contenting to gain access to the wireless channel. Mobile stations do not have knowledge of the system state. This means they do not know how may other stations are trying to gain access to the channel. Scheduling must be done separately for the uplink and downlink. Finally, mobile stations are often restrained by lack of computational power and lack of battery power. Until now, there are some efforts for the development of scheduling algorithms, which take into account one or a few of the above factors. We studied the factors that play an important role to the scheduling in wireless networks and more precisely for GPRS, the data transmission extension of GSM. We propose are own algorithm under the name LUB. With our scheduling algorithm, the maximum delay of the system has a lower upper bound. The CAC (Call Admission Control) process therefore can admit more flows without increase of the maximum delay.
|
Αλγόριθμοι Χρονοπρογραμματισμού για δίκτυα GPRS
