Analysis of 5G New Radio Uplink Signals on an Analogue-RoF System Based on DSP-Assisted Channel Aggregation
Round 1
Reviewer 1 Report
The paper analyses two uplink physical channels (data and control) for new radio technology with improved performance, while using the Analogue Radio over Fiber (A-RoF) system as an alternative to inefficient on-off keying. It tests two strategies: frequency division multiple access and time division multiple access, and concludes that the latter achieves higher spectral efficiency and lower error rates.
I am not an expert on this
type of optical communications and have found an extensive amount of
very specific acronyms and terms that make the reading quite difficult. The diagram in Figure 1 is very helpful, but I would have liked to see
an explanation to go with Figures 2 and 3. It is complicated to understand the mechanisms without and accompanying description. The two techniques are the key of
the paper, and a bit more time could have been spent explaining how
their schemes work and what their parameters mean. Figure 16 is also shown without a good enough
explanation of what is being shown, why do results look like they do,
and which conclusions are extracted from them.
The results seem to clearly favour TDMA in every case. Do the authors have an intuition of some cases where TDMA would lose against FDMA? Describing some pros and cons for each of the two technologies and showing some kind of trade off would make the paper stronger.
In
general, I believe the results presented in this work are relevant.
However, the paper would benefit from better explanation of the systems
used and of some of the results.
Minor:
Line 118: which ARE the...
Author Response
Response to Reviewer 1 Comments
The paper analyses two uplink physical channels (data and control) for new radio technology with improved performance, while using the Analogue Radio over Fiber (A-RoF) system as an alternative to inefficient on-off keying. It tests two strategies: frequency division multiple access and time division multiple access, and concludes that the latter achieves higher spectral efficiency and lower error rates.
Point 1:
I am not an expert on this type of optical communications and have found an extensive amount of very specific acronyms and terms that make the reading quite difficult. The diagram in Figure 1 is very helpful, but I would have liked to see an explanation to go with Figures 2 and 3. It is complicated to understand the mechanisms without and accompanying description. The two techniques are the key of the paper, and a bit more time could have been spent explaining how their schemes work and what their parameters mean. Figure 16 is also shown without a good enough explanation of what is being shown, why results look like they do, and which conclusions are extracted from them.
Response 1: First of all we really appreciate your review and recommendation. We would like to mention that in the revised version of the paper all technical acronyms are clarified and shown in the redrawn Figure 1. Moreover, the blocks depicted in Figure 1 are now explained in detail. The TX DSP block at the transmitter side and the RX DSP block at the receiver side of Figure 1 are further analysed where the DSP blocks of the two techniques shown in Figure2 and Figure 3 split in to two subsections describing the principles of FDMA and TDMA.
Figure 16 was explained in the original paper as of Figure 15(by mistake) which shows the decoded constellation after channel equalization indicating that the symbol are distorted due to channel impairment. We corrected as Figure 15 to Figure 16 on line 197, 198 and 198 on the manuscript.
Point 2:
The results seem to clearly favour TDMA in every case. Do the authors have an intuition of some cases where TDMA would lose against FDMA? Describing some pros and cons for each of the two technologies and showing some kind of tradeoff would make the paper stronger.
In general, I believe the results presented in this work are relevant. However, the paper would benefit from better explanation of the systems used and of some of the results.
Response 2: Thank you for asking this essential point. The scope of this research is focused on the performance and spectral efficiency comparison between the two aggregation schemes in which TDMA is in favour for instance in Figure 4 it is more spectrally compact as you already noticed it. The other metric to compare the two techniques is latency. However, the latency measurement was not covered due to the fact that we perform the simulation based on offline DSP.
Point 3:
Minor:
Line 118: which ARE the...
Response 3: We correct it on the revised version.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
This paper provides a comparative study of TDMA and FDMA for 5G New Radio uplink physical channels, using simulation. The fact that the authors considered 5 GHz of bandwidth for the optical channels is strange, since the reviewer would expect the modulator or the photodetector to be band-limited, but not the optical channel. The used simulation setup and parameters seems complex, but no details are given such as how was the simulator implemented, in Matlab? absolutely no details are given on, e.g., pre-emphasis, up sampling, down-sampling, equalization, EVM estimation algorithm; it even seems that the authors used an existing simulator, given the lack of details. Also que quality of the figures is really poor and inconsistent across figures (different figure and font sizes).
The conducted analysis focused on the comparison of EVM vs clipping ratio and ODN loss, including a constellation diagram is shown in fig. 16, which allowed to conclude that TDMA saves ~1GHz of bandwidth and provides slight better EVM due to the fact of using a smaller amount of spectrum. This contribution is too simple, of low significance and its scientific soundness is really poor. For these reasons I recommend that the paper is rejected.
Author Response
Response to Reviewer 2 Comments
Point 1: This paper provides a comparative study of TDMA and FDMA for 5G New Radio uplink physical channels, using simulation. The fact that the authors considered 5 GHz of bandwidth for the optical channels is strange, since the reviewer would expect the modulator or the photodetector to be band-limited, but not the optical channel. The used simulation setup and parameters seems complex, but no details are given such as how was the simulator implemented, in Matlab? Absolutely no details are given on, e.g., pre-emphasis, up sampling, down-sampling, equalization, EVM estimation algorithm; it even seems that the authors used an existing simulator, given the lack of details. Also quality of the figures is really poor and inconsistent across figures (different figure and font sizes).
Response 1: We appreciate for these critical observations. On the revised version of the paper, we are targeting low-cost optoelectonics and we assume that the overall electrical-to-electrical transfer function has a pole at 5 GHz.
We re-plot Figure 1 and explained the detail of each functional blocks outlined on the figure. Furthermore, the offline DSP simulator set up implemented in MATLAB for the two techniques are subdivided into two subsections and the signal evolutions are then better elaborated. The qualities of the figures are also improved.
Point 2: The conducted analysis focused on the comparison of EVM vs clipping ratio and ODN loss, including a constellation diagram is shown in fig. 16, which allowed to conclude that TDMA saves ~1GHz of bandwidth and provides slight better EVM due to the fact of using a smaller amount of spectrum. This contribution is too simple, of low significance and its scientific soundness is really poor. For these reasons I recommend that the paper is rejected.
Response 2: Thank you for the feedback. In 5G scenario, we believe that one of the important parameters are EVM performance and spectral efficiency. For instance, given the same network deployment, TDMA seems to support more users without network failure than the corresponding FDMA. In the paper we didn’t figure out the latency gain between the two schemes, this is due to the fact that we emulate the system based on offline DSP.
Reviewer 3 Report
This paper is easy to read and understand.
It's good overall, however there are some small misses as follows.
Line 87 : optical distribution loss -> optical distribution network
Line 90 : DAC -> ADC
Line 98 : receiver signal processing -> receiver digital signal processing
Line120 : title of subsection 4.1 is incomprehensible
Line 197~ : Figure 15 -> Figure 16
In addition, I think author must explain about spectrum flatness about figures 4, 7, 8 and 9.
Author Response
Response to Reviewer 3 Comments
Point 1: This paper is easy to read and understand. It's good overall, however there are some small misses as follows.
Line 87: optical distribution loss -> optical distribution network
Line 90: DAC -> ADC
Line 98: receiver signal processing -> receiver digital signal processing
Response 1: We appreciate the observation. On the revised version we implemented all the comments.
Point 2: In addition, I think author must explain about spectrum flatness about figures 4, 7, 8 and 9.
Response 2: Thank you for the feedback. The frequency roll off towards higher frequency is due to the frequency response of the optical channel. We commented this behaviour on the revised version of the paper.
Round 2
Reviewer 2 Report
The authors have made some changes to the paper, including more details on the simulation setup, addressing one of the previous comments.
However, I maintain my previous position that this contribution is too simple, of low significance and its scientific soundness is really poor. For these reasons I still recommend that the paper is rejected.
