Randomized algorithms
Online computation and competitive analysis
Online computation and competitive analysis
Congestion control for high bandwidth-delay product networks
Proceedings of the 2002 conference on Applications, technologies, architectures, and protocols for computer communications
Competitive online scheduling for server systems
ACM SIGMETRICS Performance Evaluation Review
Product Line Selection and Pricing with Modularity in Design
Manufacturing & Service Operations Management
Scheduling with limited information in wireless systems
Proceedings of the tenth ACM international symposium on Mobile ad hoc networking and computing
Capacity of fading channels with channel side information
IEEE Transactions on Information Theory
IEEE Transactions on Information Theory
Fading channels: how perfect need "perfect side information" be?
IEEE Transactions on Information Theory
Exploiting wireless channel State information for throughput maximization
IEEE Transactions on Information Theory
Capacity and power allocation for fading MIMO channels with channel estimation error
IEEE Transactions on Information Theory
IEEE Network: The Magazine of Global Internetworking
Providing performance guarantees in multipass network processors
IEEE/ACM Transactions on Networking (TON)
A Combined Frequency Assignment and AP Scheduling for Throughput Maximization in IEEE 802.11 WLAN
Proceedings of International Conference on Advances in Mobile Computing & Multimedia
| Hi-index | 0.00 |
We consider a dynamic scheduling system where a single controller selects 'tasks' to service over U 'servers' of fluctuating quality/speed. The quality/speed of each server determines the likelihood of successful service should a task be assigned to that server. The goal is to maximize the total expected number of tasks successfully served over a fixed time horizon (aggregate throughput) given only one server can be used in each time slot. However, the state of the servers are not known to the scheduler with certainty; at best, only statistical distributions (estimates) of the realized server states are available. We consider how the uncertainty of server state information compromises the expected aggregate throughput compared to a 'clairvoyant' scheduler which has instantaneous, perfect information about the realized server states. The issue of operating in uncertain environments arises in a number of scheduling applications of interest from wireless applications to computing networks to revenue management systems. The results presented in this paper provide a framework for gauging the loss due to uncertainty in such scheduling systems. First, it is shown that opportunistic scheduling (on the server of current expected best state) is throughput optimal, under uncertain (unknown) server states. Then, the throughput of the 'clairvoyant' scheduler is found to be upper-bounded (in general) by U times the throughput under uncertain server states; this bound is tight. Third, for bimodal and uniform server qualities/speeds better bounds are obtained--down to a factor of 2. Of course, actual throughput loss due to server state uncertainty depends on the server state distributions which are available as partial information to the scheduler. Finally, via numerical experiments we evaluate the throughput loss in various operational scenarios for wireless packet scheduling applications.