Choose one of the following and work in a team of 2–3.
(A) Multi-user offloading. Extend Lab OS13.2: users also choose their transmit power or the edge charges a price per offloaded task. Show how the equilibrium, the average latency and the edge utilization change with the price, and find the price that minimizes average latency.
(B) Deadline-aware placement. Extend Lab OS13.4 with (1) dropping tasks that cannot meet their deadline and (2) phone energy in the cost. Generate random workloads (e.g., 1,000 tasks, Poisson arrivals, sizes from a distribution) and compare your policy with all-local, all-edge and the original policy.
(C) Workflow offloading. Model an AR pipeline (capture → pre-process → detect → track → render) as a DAG and schedule it on device, edge and cloud with HEFT (Chapter 6 lab), adding communication costs from the network model of this chapter. Study how the schedule changes with bandwidth.
For your choice, write a 2–3 page report: problem and model, algorithm, experimental setup (parameters, number of runs, random seeds), results (at least two graphs or tables), discussion (when does your method help, when not?) and one idea for future work. Cite at least two research papers.