QoS Provisioning in Wireless Data Networks under Non-Continuously Backlogged Users

This paper addresses the problem of efficient resource allocation in CDMA wireless networks, supporting non-real-time services with various long-term QoS requirements, by exploiting the advantages of opportunistic scheduling. Most of the existing work covers the analysis of the system under the assumption of infinite backlogs. In this paper, we remove this assumption and analyze the behavior of the system with respect to its overall achievable performance and its ability to gratify users’ individual QoS requirements under different fundamental opportunistic scheduling policies, for the more realistic case of non-continuously backlogged users. Motivated by the corresponding results that indicate the unsatisfactory performance of these schedulers when considering finite backlogs, and in order to overcome these inefficiencies, a new non-work-conserving opportunistic scheduling framework is proposed which is seamlessly integrated within the operation of the various schedulers. The corresponding results demonstrate that the proposed approach results in the fulfillment of users’ QoS prerequisites while still maintaining the opportunistic nature of the schedulers and thus achieving high system throughput.