Dynamic control of processing systems for jobs with decaying value
- In myriad processing systems, delays can degrade the internal state of jobs in the system. Explicitly modeling the decay in internal job value allows us to generalize the concept of a "hard" deadline to that of a "soft" deadline: while hard deadlines are characterized by abrupt value decay, soft deadlines allow for gradual value decay. We leverage optimal control and stochastic simulation methodologies to study two models that incorporate job value decay as an essential modeling attribute. The first model, motivated by patient triage problems, features jobs that are characterized by stochastic service times and deterministically decaying internal value. Since the internal value of a job models the overall health of a patient, we aim to find a non-preemptive scheduling policy that maximizes the total expected value of the jobs after they have completed service. Finding an optimal scheduling policy is computationally intractable so we choose to focus on some heuristics. We provide an- alytic performance guarantees and use numerical experiments to compare the policies in a variety of scenarios, including patient triage scenarios. The second model, motivated by wireless streaming systems, is a service rate control model in which jobs decay in value only upon reaching the server. When a job has zero value, the job is ejected from the queue. In packet streaming, this value decay ensures that the stream does not get "stuck" on any single packet. We formulate an optimal control problem in which power is used to modulate the service rate. We characterize the monotonicity structure of the optimal control and use these analytic results to build power control schemes that adapt to the backlog and decaying value. Simulations show that our proposed schemes outperform benchmark algorithms.
|Type of resource
|electronic; electronic resource; remote
|1 online resource.
|Stanford University, Department of Electrical Engineering.
|Statement of responsibility
|Submitted to the Department of Electrical Engineering.
|Thesis (Ph.D.)--Stanford University, 2017.
- © 2017 by Neal Maneck Master
- This work is licensed under a Creative Commons Attribution Non Commercial 3.0 Unported license (CC BY-NC).
Also listed in
Loading usage metrics...