A Comprehensive Model of Android Software Aging and Rejuvenation Considering Battery Saving

Round 1

Reviewer 1 Report

This paper presents a summarized survey of literature on Android software aging and rejuvenation models. These models are extended by the authors   considering different aging levels, mobile device activity, and battery status. The proposed models extend the ones described by other authors e.g. [17,20] and refer to previous publications of the submitting authors [22-25]. Hence, it is needed to clearly and more substantively comment new contributions in relevance to theses publications, especially those of Yakovena, et al., are figures 2-7 original for this paper or did you take them (or modified) from your previous publications? I suggest to shorten the introduction and introduce an additional section, e.g.  Problem statement (comprising literature review).

Line 61 the sentence “The first …”lacks verb. Line 179, here you mention measured aging metrics – I suggest to present and comment these metrics , how to select their thresholds?  Line 200 – FDT parameter needs explanation, line 211 – high performance level is not defined (how to identify it practically?), similarly aging process is observed – in line 218 – what are the symptoms to observe,

Line 238 – I propose to comment cold and warm rejuvenation (for some readers this ma not be obvious), lines 277-288 you specify assumed values of some parameters , some comment is needed whether they are typical or can vary in practice?

line 289 – what is the original model (fig 1, or another one ?), lines 306-312 – these observations should be better justified, 

Section 3.2 – you refer to various states but their definition is lacking, maybe it is in your previous publications?? Equations (1) are original or taken from your publication?

Metrics in lines 396-401 are probably described in your papers [24,25], they should be defined in this paper. Lines 431-433 needed additional studies should be commented. The conclusion is too vague, what do you mean by software quality? (line 458).

The presented results refer to performed simulations, it is a pity that there are no references to real systems and results from the field?

Author Response

Thank you for your valuable comments and suggestions. We have addressed all your comments in the updated version of the paper.

This paper presents a summarized survey of literature on Android software aging and rejuvenation models. These models are extended by the authors   considering different aging levels, mobile device activity, and battery status. The proposed models extend the ones described by other authors e.g. [17,20] and refer to previous publications of the submitting authors [22-25]. Hence, it is needed to clearly and more substantively comment new contributions in relevance to theses publications, especially those of Yakovena, et al., are figures 2-7 original for this paper or did you take them (or modified) from your previous publications? I suggest to shorten the introduction and introduce an additional section, e.g.  Problem statement (comprising literature review).

• Figures 5–7 are original to this paper, and the comprehensive model presented in Fig. 7 is the main contribution of this study.
• The outline of the new contributions of the paper was added at the end of the Introduction section.
• The introduction was shortened, and the Problem statement section was added.

Line 61 the sentence “The first …”lacks verb.

• We corrected the sentence.

Line 179, here you mention measured aging metrics – I suggest to present and comment these metrics , how to select their thresholds?

• The metrics are described and discussed in Section 4 along with their thresholds. We added a reference to this section of the paper.

Line 200 – FDT parameter needs explanation,

• FDT is “Frame Draw Time” – time needed to draw a frame.

line 211 – high performance level is not defined (how to identify it practically?), similarly aging process is observed – in line 218 – what are the symptoms to observe,

• Aging metrics are used to observe these symptoms, as explained in Section 4. The corresponding sentence with the explanation was added.

Line 238 – I propose to comment cold and warm rejuvenation (for some readers this ma not be obvious)

• Cold and warm rejuvenation procedure description was added. The rejuvenation procedure can be divided to “cold” and “warm” one [24]. "Cold" rejuvenation occurs by rebooting the entire device, which gives the best result of improving system performance, but takes the most time and can interrupt the user's work. In turn, rejuvenation by restarting the application ("warm" rejuvenation) takes the least time, but the effectiveness of improving performance is not as high and, also, it is possible to interfere with the user's activities or the performance of important tasks.

lines 277-288 you specify assumed values of some parameters , some comment is needed whether they are typical or can vary in practice?

• The explanation was added in the text.

line 289 – what is the original model (fig 1, or another one ?)

• The model presented in Fig. 5 is mentioned, and we add the indication.

lines 306-312 – these observations should be better justified, 

• We added some explanation and justification in the text. A comparison of the obtained results shows that considering the battery charge has a significant impact on planning the time of software rejuvenation. Thus, if one schedules the software rejuvenation procedure without considering the battery state, the estimated starting time of this procedure will be significantly longer. However, during this time, the device's battery will most likely enter a low charge state, and performing the rejuvenation procedure will further discharge it, which will have a negative impact on the user experience.

Section 3.2 – you refer to various states but their definition is lacking, maybe it is in your previous publications?? Equations (1) are original or taken from your publication?

• We added the state definitions to all the models. Equations (1) are original for this publication.

Metrics in lines 396-401 are probably described in your papers [24,25], they should be defined in this paper.

• A description of the metrics was added to the paper. Activity Launch Time is the time needed to launch an activity in Android OS [28]. Frame Draw Time is the time needed to draw a frame. It is necessary that FDT does not exceed 16ms to allow frame refresh rates above 60 fps, which is a typically convenient value. The ratio of missed or delayed frames that were not displayed to the total number of frames that should have been displayed is the Junky Frames Ratio.

Lines 431-433 needed additional studies should be commented.

• We add some comments to this part of the discussion.

The conclusion is too vague, what do you mean by software quality? (line 458).

• Software failure is an important part of its reliability, and according to ISO/IEC 25010:2011, of software quality. The proposed models of aging and rejuvenation can be used to enhance software quality in terms of decreasing the probability of aging-related failures. The corresponding explanation was added to the text.

The presented results refer to performed simulations, it is a pity that there are no references to real systems and results from the field?

• Typical systems that use software rejuvenation methods are transaction processing systems, web servers, spacecraft systems etc. A real example can be the method of Apache web server rejuvenation, which terminates and rebuilds processes after a certain number of requests have been executed. Another approach is to reboot virtual machines running in a cloud computing environment. Telecommunications corporation AT&T implemented real-time billing software rejuvenation. The concluding section was updated with these references.

Reviewer 2 Report

Dear authors, 

  you can find my minor comments in the attached file. I think the scope is quite interesting, but I see to much pictures very similar to those reported in ref. 23. I would suggest to improve the quality of pictures and to try to explain better the states and transition in the graph so that is more highglighted your contribution. 

Comments for author File: Comments.pdf

Author Response

Thank you for reading our paper in detail, and for your valuable comments and suggestions. We have addressed all your comments in the updated version of the paper.

Line 60: can you provide here a referece please?

• We added the proper references.

Line 82: restarting or re-installing the operting system, in my opinion, is not a so good example. I would know if there are most effective tecquiniques.

• We added an example of another rejuvenation technique. Restarting or re-installing the OS can restore the “pure” initial state – this is the most effective from the point of aging view, but sure not so good from the user’s point of view. The paper is not focused on the rejuvenation techniques by themselves, but on the proper time for the rejuvenation procedure. Hence, we do not describe them in detail.

Fig. 3: this picture is very well detailed but I wuld invite to better explain in the text all th transitions. Also the acronyms of state are not so intuitive and then to read and unserstand the picture the reader needs to come back often to the state explanation.

• Figures 3, 4, and 7 were completely redrawn for better quality and understanding, and the proper reference was put near Fig. 4.

Fig. 4: I would suggest to improve the quality of pictures and to specify it is taken by [23]

• Figures 3, 4, and 7 were completely redrawn for better quality and understanding, and the proper reference was put near Fig. 4 (though it was indicated in the text, line 260).

Fig. 5: this picture seems equal to to that reported in [23]

• We carried out a new set of experiments, and Figures 5 and 6 were replaced with the newly obtained results. Though it does not change the conclusions.

Fig. 6: this picture seems equal to to that reported in [23]

• We carried out a new set of experiments, and Figures 5 and 6 were replaced with the newly obtained results. Though it does not change the conclusions.

Line 320: following explanation would be reported before Figure 4.

• The proper explanation was placed before Fig. 4.

Round 2

Reviewer 1 Report

The corrections of the paper are satisfactory.