Optimization-Based Multi-Objective Task Offloading in IoT–Fog–Cloud Systems: An Energy–Delay Trade-off Framework 

Research Context: This is an ongoing thesis research project focused on the problem of task offloading in IoT–Fog–Cloud computing architectures, where devices must decide whether to process tasks locally or offload them to fog or cloud nodes. The core challenge is simultaneously minimizing energy consumption and execution delay while satisfying hard constraints such as task deadlines, memory limits, and node capacity. The work begins by rigorously formulating this as a constrained multi-objective optimization problem with well-defined variables, objectives, feasibility rules, and benchmark scenario specifications.

Paragraph 2 — Methodology: The proposed solution involves implementing a baseline NSGA-II evolutionary solver, then enhancing it with a hybrid operator — either a Differential Evolution-based crossover or a local refinement module — to improve Pareto-front quality, convergence stability, and solution diversity. Ablation studies will be conducted to isolate the contribution of each component, and a practical selection rule will be defined for choosing operating points on the Pareto front. The final phase involves a reproducible experimental suite benchmarking the proposed approach against standard baselines across varying loads and system configurations, supported by statistically rigorous comparative analysis.

Full project Desciption and Requiremnts :

Optimization-Based Multi-Objective Task Offloading in IoT–Fog–Cloud Systems: An Energy–Delay Trade-off Framework 

Objective 1 — Formulate the optimization problem 

Description (short): Build a constrained multi-objective model for offloading decisions that minimizes energy and delay under deadline and resource limits. 

Expected outputs: 

• Mathematical formulation (variables, objectives, constraints). 
• Feasibility handling rules (deadline/memory/capacity, one-choice-per-task). 
• Defined evaluation metrics and benchmark scenario specifications. 

Objective 2 — Develop the optimization solver 

Description (short): Implement a baseline multi-objective optimizer (e.g., NSGA-II) and a hybrid enhancement (e.g., DE-based operator or local refinement) to improve search quality and stability. 

Expected outputs: 

• Implemented solver(s): baseline + improved variant, with documented settings. 
• Ablation results showing the contribution of the enhancement module. 
• Pareto-front generation and a practical selection rule for operating points. 

Objective 3 — Validate performance and robustness 

Description (short): Benchmark the proposed solver against standard baselines across different loads and system settings with statistically supported analysis. 

Expected outputs: 

• Reproducible experimental suite (scenarios, seeds, parameter settings). 
• Result artifacts: Pareto fronts, energy–delay curves, convergence/runtime, feasibility rate. 
• Comparative analysis and final conclusions (including limitations and reproducibility notes). 

References:

https://link.springer.com/article/10.1007/s12652-021-03388-2