Deadline Aware Edge Task Offloading Under Network and Queue Uncertainty

Authors

  • Mayank Singh Student, Government Engineering College, Modasa
    Author
  • Mohit Malhotra Student, Government Engineering College, Modasa
    Author
  • Riya Tripathi Student, Government Engineering College, Modasa
    Author
  • Shubham Pandey Assistant Professor, Government Engineering College, Modasa
    Author

DOI:

Keywords:

Edge computing; computation offloading; deadline probability; queueing theory; energy optimization; mobile computing; resource allocation.

Abstract

Edge computing can reduce the energy consumed by mobile devices, but offloading decisions become unreliable when average latency is treated as a guarantee that a task will finish on time. This paper develops an analytical framework that chooses between two local processor settings and a remote edge server. The framework combines a device energy model, transmission time, an exponential queueing model, and a requirement that each selected operating mode meet a deadline with at least 95 percent probability. A constructed workload is evaluated under three wireless conditions and several server arrival rates. At an arrival rate of 80 tasks per second, the probability constrained policy consumes 0.11425 joules per task and achieves a calculated deadline success probability of 99.50 percent. A policy based only on mean latency consumes less energy but achieves 93.44 percent success. These are model calculations rather than measurements from deployed equipment. The analysis explains why queue uncertainty, local fallback capacity, and the boundary of the energy accounting must be considered together when designing an offloading controller.

Downloads

Published

2026-11-14

How to Cite

[1]
Mayank Singh , “Deadline Aware Edge Task Offloading Under Network and Queue Uncertainty”, Int. J. Curr. Sci. Eng. Intell. pp. 1-5, 2026-11-14 doi: .