Content Dependent Representation Selection Model for Systems Based on MPEG DASH

Round 1

Reviewer 1 Report

The current paper needs major revision, and the major concerns are as follows:

1. there are some parts not well explained. e.g. what is the relationship between frames and segments? e.g.  (page3) what is the reason behind the claim "Also segments with shorter duration ensure the lower number of stalling and cutoff events during the video streaming", why is this true?
2. what does SI and TI information mean in this paper?
3. why precoding is required in existing works?
4. the streaming schemes used in the simulation are not new
5. how about simulating the performance using PSNR
6. there are many new video transmission schemes, the authors may need to discuss more recent works. Here are some examples: Power-Efficient Wireless Streaming of Multi-Quality Tiled 360 VR Video in MIMO-OFDMA Systems, TWC2021; Joint Resource Allocation and 3D Aerial Trajectory Design for Video Streaming in UAV Communication Systems, IoTJ 2021

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

This paper presents a model that can be used to determine representation sets based only on spatial (SI) and temporal information (TI) of selected video sequences. The author believes that compared with the existing methods, the model is easy to use to select the representation sets is its biggest advantage, the latter mainly involves multiple pre-coding and quality calculations of the encoded video sequences.

 

The main comments are as follows:

 

1. From my point of view, your summary did not capture the point. You said that you proposed a method to speed up the model, but you used two algorithms to test, and I did not find that you combined them.

 

2. In the third and fourth chapters of your article, I only saw you listed a lot of existing formulas and diagrams. Are these two testing algorithms your original algorithm? I have not seen what you call "Content dependent representation selection model"

 

3. You described your experimental part very briefly, unlike your third and fourth chapters, which have a lot of graphs. And I didn't see the correlation between your theoretical part and experimental part. Which algorithm is the A-Stream system based on? So I think your final data is meaningless.

 

Because of the above points, I rejected your article

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

This article presents a methodology and a notation of a model for selecting the optimal representation set that takes into account the spatial and temporal information of video sequences. The model was tested using the Basic Adaptation algorithm and the Segment Aware Rate Adaptation algorithm and two different network scenarios. Compared to the segmentation available in the relevant literature, the proposed segmentation obtains better Structural Similarity Index Measure (SSIM) values in 92% of cases. Even though the idea is great but I believe this work is far from the journal scope. 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The authors have answered the questions.

Author Response

Thank you very much for the valuable comments and suggestions for our manuscript.

Reviewer 2 Report

I am glad to see that the authors tried their best to revise the manuscript. But there are some concerned questions:

1. Abstract is too long, authors should refine the background.
2. It also exists many grammaly issues, such as "Solutions in the avail- 485 able research that addressed the problem of defining the optimal representation sets are 486 mostly proprietary, do not ensure all needed information of reproduction, do not provide"
3. Figure11 needs to imporve to clarify the method.
4. Some related work may be added, such as"Deep-IRTarget: An Automatic Target Detector in Infrared Imagery using Dual-domain Feature Extraction and Allocation", "Multi-camera multi-player tracking with deep player identification in sports video", "Deep-learning-based burned area mapping using the synergy of Sentinel-1&2 data".

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Authors addressed all my concerns and I am satisfied with the current version. 

Author Response

Thank you very much for the valuable comments and suggestions for our manuscript.