Beyond 100G: All-Optical Processor for High-Capacity Access~Networks

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

In the paper “Beyond 100G: All Optical Processor for High-Capacity Access Networks,” the authors propose an all-optical processor (AOP) that leverages optical signal processing to greatly enhance the system efficiency of high-speed Passive Optical Networks (PONs). Specifically, two functionalities are realized: all-optical equalization and chromatic dispersion compensation. The authors comprehensively introduce the processor’s design, operating principles, simulation studies, and system validations, demonstrating significant improvements in signal quality and overall network performance. The paper is written clearly, with the background introduced effectively, the motivation behind the AOP well-justified, and the AOP design novel and powerful.

That being said, I would support the work for publication. However, several specific issues still need to be justified and addressed:

1.        Please clarify the purpose of the two \tau boxes in Figure 1 within the path switch path 0 located at the bottom left corner of the figure. Similarly, there are two T boxes in Figure 2, positioned at the bottom left corner before the second coupler in the system. What distinguishes the functionality of the first time delay box from the second? If the duplication is a typographical error and there should be only one time delay box, why is a time delay necessary in the 50/50 branch, given that the time delay for each of the four separate paths will eventually be added into the system? If this time delay box is necessary, why is it missing in the bottom left corner of Figure 3?

 

2.        Traditional DSP techniques implement all-optical equalization, chromatic dispersion compensation, and correct optical nonlinearities. The proposed AOP only addresses power equalization and chromatic dispersion, limiting its feasibility in actual PON systems. If DSP is still needed to correct optical nonlinearities, why not use it to handle chromatic dispersion and power equalization as well? This would simplify system construction. Please elaborate.

 

3.        While I trust the authors' claim that the overall PON system energy efficiency should be much better using AOP compared to DSP, more quantified energy consumption discussion should be included. Specifically, the discussion should address how much power will be spent on AOP’s training, inference, and operation.

 

4.        In Figure 4, the processor is placed before the SSMF. From my understanding, the processor should be used to correct the chromatic dispersion inside the SSMF. Is there any difference of placing the processor in front of or after the SSMF? 

 

5.        Following question 4, how robust is this AOP to environmental variations? In real life, temperature variations and strains on the optical fibers change the chromatic dispersion of the optical signals constantly, while AOP is trained using the system parameters under one specific environmental condition. If the AOP cannot adapt to these environmental changes, the practicality and significance of this work are highly questionable.

 

6.        How long and how many computational resources does it take to tune the AOP parameters? This is important for the scalability of this technology. In the paper, the authors proposed a four-channel system. However, in the industry, PON systems can easily scale to 10+ or even 30+ channels. It would be of interest for readers to understand the limits and current status of training AOP. Additionally, if the authors further tune the parameters, will the signal quality be further improved?

7. Is there any physical or computational understanding for why 12FFE performs better than the optical processor over short transmission distances, as shown in Figure 5a? 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The authors presented an All optical processors for 100G PON and 200G PON.

The manuscript is written clearly, however I feel like few more plots can used to make the paper impactful.

For example: Authors only presented the eye diagram....

a. I recommend to put the optical spectrum before and after AOP at a particular receive power.

b. What is the frequency response of the AOP at a particular operating condition. What are the values of the delay, and attenuator.

c. How to implement the AOP practically? Please add discussion sections about the implementation of the AOP and how you control so many delay lines and phase shifter together given that the couplers are not identical in practice.

d. Why is the reason of the eye is not symmetric? What is the linearity (RLM) of the eye?

e. Please provide all the details so that an independent researcher can reproduce the results without any trouble.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf