Scheduling in Packet Switched Cellular Wireless Systems

In cellular wireless networks where users have independent fading channels, throughput for delay tolerant applications has been greatly increased on the downlink by exploiting the multiuser diversity inherent in the system using opportunistic schedulers at the base station. An interesting question is how such schedulers will provide Quality of Service(QoS) guarantees for a mix of data traffic and traffic from delay-sensitive multimedia applications. In the first part of this dissertation, we completely characterize the scheduled rate, delay and packet service times experienced by mobile users in a packet-switched cellular wireless system in terms of a configurable scheduler metric. The metric used has a general form, combining an estimate of a mobile user’s channel quality with the scheduling delay experienced by the user. A continuous approximation enables us to compute closed form expressions for the scheduler statistics. In addition to quantifying the scheduler performance, our analysis highlights the inherent trade-off between system throughput and the delay experienced by mobile users with opportunistic scheduling. We also use this analysis in to study the effect of prioritized voice users on data users in a cellular wireless system with delay constrained opportunistic scheduling. Our statistical analysis of the forward link is validated by extensive simulations of a system architecture based on the CDMA 1xEV-DO system. The increase in data traffic from mobiles to the base station has led to a growing interest in a scheduled reverse link in the 1xEV-DO system. We address the reverse link scheduling problem in a multi-cell scenario with interference constraints both within and outside the cell. This approach leads to a co-operative scheduling algorithm where each base station in a cellular network maximizes the sum of mobile data transmission rates subject to linear constraints on (1) the maximum received power for individual mobiles (2) the total interference caused by scheduled mobiles to (a) traffic and control channels of other mobiles within the cell and (b) mobiles in neighboring cells. Simulations of the reverse link structure based on the 1xEV-D0 system highlight the distinct advantages of this algorithm in ensuring predictable inter-cell interference and higher aggregate cell throughputs.