FFSCN: Frame Fusion Spectrum Center Net for Carrier Signal Detection

Round 1

Reviewer 1 Report

The authors presented FFSCN: Frame Fusion Spectrum Center Net for Carrier Signal Detection.

The reviewer has the following comments.

1.     How the FFSCN can effectively detect the multi-carrier and single carrier modulation signals on broadband and outperform other deep learning-based methods in accuracy and efficiency. Can authors justify using other proposed models?

2.     How the authors justifying the proposed work if an algorithm is missing?

3.     The authors should incorporate the comparison table and then show the advantage of the proposed work with the previously published work.

Author Response

The authors presented FFSCN: Frame Fusion Spectrum Center Net for Carrier Signal Detection.

The reviewer has the following comments.

Point 1: How the FFSCN can effectively detect the multi-carrier and single carrier modulation signals on broadband and outperform other deep learning-based methods in accuracy and efficiency. Can authors justify using other proposed models?

Response 1: Thank you very much for your comment. We notice our summary description of the proposed FFSCN is not very exact, and we have modified it in the abstract (P1), introduction (P3), and conclusion (P15) sections. Moreover, this paper mainly focuses on locating and estimating the sub-carrier signals on a broadband power spectrum, which is an improvement and expansion of SCN. We have compared the FFSCN with our prior work SCN and FCN. Also, we compared the FFSCN with another similar work: SigdetNet.

Point 2: How the authors justifying the proposed work if an algorithm is missing?

Response 2: We appreciate your helpful comment. We proposed 3 FFSCN models, FFSCN-R, FFSCN-MN, and FFSCN-FMN, and they can all effectively solve the multi-carrier signal detection problem in the braodband power spectrum. Meanwhile, the 3 model algorithms are derived from the gradual evaluation of the SCN model. We kept the three algorithms in the paper to describe improvements' thought processes completely. If one of them is missing, it may cause our proposed model algorithms to appear abrupt, but it does not affect the remaining algorithms to be described in detail and specifically.

Point 3: The authors should incorporate the comparison table and then show the advantage of the proposed work with the previously published work.

Response 3: We greatly appreciate your comment. We updated the experiments section and added some descriptions of Table 1. Details are as follows:

Firstly, to demonstrate the effectiveness of the proposed FFSCN models, we compare the performance with other deep learning-based methods, including the SCN [14], the FCN [12], and SigdetNet [13]. As can be seen in Table 3, our proposed FFSCN-FMN models outperform the other models. The performances of SigdetNet and FCN degrade more than other models. With the IoU threshold increase, all the model's detection performances degrade, but the proposed FFSCN models perform more robustly than our previous SCN model overall. Moreover, from the table, we also conclude that the residual backbone performs better than the MobileNetV3 backbone. However, the fusion MobileNetV3 backbone achieves the best performance, which indicates that the multiple-times feature fusion is superior to the one-time fusion strategy.

Reviewer 2 Report

Dear authors, first of all I would like to acknowledge your work in the field of electronics. Please, find below my suggestions.

An extensive language check is necessary to avoid constructions like "Recently studies", where either a comma is missing or the sentence is not correctly formulated.

I believe the introduction should be improved: more details and state of the art should be provided.

Data has not been made available, please consider data sharing.

All the best!

Author Response

Point 1: An extensive language check is necessary to avoid constructions like "Recently studies", where either a comma is missing or the sentence is not correctly formulated.

Response 1: Thank you very much for the comment. We have checked and revised the sentences with spelling or grammatical errors by using the "Track Changes" function.

Point 2: I believe the introduction should be improved: more details and state of the art should be provided.

Response 2: We greatly appreciate your comment. We have added some more details in the introduction section. We presented it below:

There are many algorithms for carrier signal detection. Energy detection [3,4,5] is a non-coherent detection method that detects the carrier signal based on the sensed energy. Although it is simple and needs no prior knowledge, the detection performance is subject to the uncertainty of the received signal noise power. Cyclostationary feature detection [6,7,8] exploits the periodicity in the received narrowband signal to identify the presence of carrier signal, and it is robust to noise uncertainties. In contrast, this method needs prior knowledge and costs high computational complexity and detection time. At the same time, energy detection and cyclostationary feature detection methods do not estimate the parameters of the carrier signal, only focusing on its presence.

Some improvements have been noted using the double-thresholds method [9,10] to overcome these shortcomings. Moreover, by using signal properties such as amplitude, slope, deflection width, or distance between neighboring deflections, Kim et al. [11] proposed using a slope tracing-based algorithm to separate the interval of the carrier signals. But these thresholds methods have to face the critical issue of discovering the proper threshold values.

Point 3: Data has not been made available, please consider data sharing.

Response 3: Thank you for your comment. We have shared our dataset on onedrive (https://1drv.ms/u/s!AvmAsnk8HBhuiwMRK8kQWKzktiIw?e=YRZFay) and gitee (https://gitee.com/yulongpo/ffscn_dataset).