Joint Trajectory Planning, Service Function Deploying, and DAG Task Scheduling in UAV-Empowered Edge Computing

Round 1

Reviewer 1 Report

This manuscript presents a joint optimization architecture for UAV position, service function deployment, and dependent task scheduling, in multi-UAV edge computing scenario. After formulating the system model, the authors establish a mixed 0-1 integer programming problem, aiming to minimize the maximum completion time of tasks from the ground devices. Genetic algorithm is adopted for solving the problem, and the detailed design of the individuals is illustrated with the overall working flow. Simulation results show that the proposed method can converge to a stable state, and it outperforms other algorithms in terms of the fitness value and completion time. This paper is well-structured and is easy to follow, and the results have validated the correctness of the design. The following concerns should be addressed before publication:

1. Table I compares this work with existing work. However, only a few of the references are included in the table. Most efforts mentioned in Section II should also be included in the table to highlight the innovation.

2. SF Instances or SI are used in Fig. 1. However, I cannot find the descriptions about SI in the context. The authors need to provide the exact definition of SI in the manuscript or the caption of Fig. 1.

3. In Fig. 6, the proposed method converges after 600 rounds iteration. What is the criteria for a good convergence ability of an algorithm? This should be clarified and justified. 

In general, the manuscript is well-structured and easy to read, but still has some grammatical errors. A complete proofreading will improve the quality of this paper. 

A few inconsistencies are scattered throughout the manuscript, for example, what does 'overall DAG' mean in Table II? 'subtask' and 'sub-task'. 

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

In this paper, the authors provide an optimization scheme of position deployment, service function deployment, and DAGs task scheduling for UAV-enabled edge computing systems. They build a 0-1 integer optimization problem and design a GA-based algorithm, termed GA-JoDeS, to achieve the objective of minimizing the completion time. The performance of GA-JoDeS is evaluated by comparing it against three baseline methods in a series of simulations. The proposed system model is interesting, however, the description of the paper is confusing and lacks sufficient experimental evidence. Please see my detailed comments below.

1. In Table 1, please further provide a definition of ‘complex tasks’. It is not clear what the difference is between the complex tasks and the regular ones.

2. The analysis of algorithm complexity is necessary since the tasks are ‘delay sensitive’, as the authors state in Line 132. The time complexity of the proposed algorithm should meet the real-time requirement.

3. The convergence time of the algorithm and the situation of the computing hardware platform should be further explained and demonstrated in the simulation.

4. What is the difference between the proposed GA-JoDeS and GA? As the difference between their performance is marginal in Fig. 7 and Fig. 8, it is necessary to further demonstrate the contribution of GA-JoDeS.

5. The three comparison benchmarks are all heuristic algorithms, but their performance significantly depends on the initial value. Therefore, it is recommended to further enrich the simulation conditions, such as different initial deployments of UAVs and mobile devices, to indicate the universality of algorithms.

6. The words ‘Trajectory Planning’ in the title is inaccurate since trajectory is a collection of positions in different time slots. However, the authors only optimize the fixed position in this paper.

7. Several minor mistakes exist: (1) The units of the coordinates are missing in Fig. 6 to Fig. 8; (2) Typographical errors. Fig. 10 should be placed before the Conclusion; (3) English grammatical errors, such as ‘G individuals is generated in the population’ in line 411.

8. It is advisable to thoroughly proofread the paper for English writing.

It is advisable to thoroughly proofread the paper for English writing.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

The layout of the paper needs to be further improved, for example, the proportion of Figure 5 is too large, and there is a large blank on page 18.

Avoiding colloquial expressions.

Author Response

Thank you very much for your valuable comments and suggestions, which are very helpful to the improvement of paper quality. According to your suggestions, the layout of the paper has been revised, including reducing the size of Figure 5 and the blank space of the document on page 18, etc., thank you again for your suggestion.

Author Response File: Author Response.pdf