A bandwidth management scheme based on time multiplexing for wireless networks with predictive services

Nowadays, in wireless computing, guaranteeing a certain grade of Quality of Service (QoS) during wireless sessions is an important issue, especially when non-tolerant applications are running into the system. Mobility and channel quality often affect system performance, depending on the grade of mobility of hosts. In this paper, providing QoS to mobile hosts belonging to a certain service class (Mobility Independent Predictive - MIP) is the main issue and a statistical multiplexing algorithm is proposed, in order to save an enormous amount of resources; it is based on the Cell Stay Time (CST) distribution analysis for a 2-dimensional (2D) wireless environment under the Smooth Random Mobility Model (SRMM). There are many works in literature about resource pre-reservation: they offer good QoS guarantees, but a very low system utilization, because they do not care about the fact that a passive reservation can remain unused for long time. With the help of the CST statistical multiplexing introduced in this paper, better results are obtained in terms of system utilization, while the Quality of Service and prediction error are maintained at acceptable levels. The proposed idea has been validated through many simulation campaigns, with a good performance analysis in terms of some interesting parameters, like bandwidth system utilization or prediction errors. © 2009 IEEE.