Enhanced Clustering MAC Protocol Based on Learning Automata for UV Networks

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Combining UV communication with an Ad hoc network to construct a UV network, can enable wireless multi-hop communication among UV communication terminals and expand the constrained communication distance. UV networks have gradually become a topic of considerable research interest. In this paper, the authors proposed an enhanced clustering time division multiple access MAC protocol based on clustering and learning automata for multinode UV networks, which is valuable. The writing is standard and the chart is clear. The paper presents timely contribution with analytical and practical work.

To me, this work is suggested to be published in Photonics. There are a few things that need to be addressed before proceeding for publication.

1. For introduction part, what are some application scenarios for UV networks? Moreover, it seems that the authors suggested the main applications are about security. What is it that makes them particularly secure? Please clarify.

2. Indicate any safety concerns from the use of UV light?

3. The performance of the proposed protocol as authors called is better than the conventional ones. However, will there be some trade-off on the complexity of the system?

4. Language and Style: While the overall writing is clear, consider revisiting some sentences for improved conciseness. Additionally, ensure consistent use of terminology throughout the paper.

5. Bibliography should be enhanced. Strengthen the discussion by explicitly connecting your findings to existing bibliography.

Comments on the Quality of English Language

 Minor editing of English language required

Author Response

Combining UV communication with an Ad hoc network to construct a UV network, can enable wireless multi-hop communication among UV communication terminals and expand the constrained communication distance. UV networks have gradually become a topic of considerable research interest. In this paper, the authors proposed an enhanced clustering time division multiple access MAC protocol based on clustering and learning automata for multi-node UV networks, which is valuable. The writing is standard and the chart is clear. The paper presents timely contribution with analytical and practical work.

To me, this work is suggested to be published in Photonics. There are a few things that need to be addressed before proceeding for publication.

1. For introduction part, what are some application scenarios for UV networks? Moreover, it seems that the authors suggested the main applications are about security. What is it that makes them particularly secure? Please clarify

Answer:

We appreciate this concern and strongly agree that the application scenarios for Ultraviolet (UV) communication and the reason for UV being secure communication need to be supplemented. According to reviewer’ advice, we have added the corresponding explanation.

UV communication, with its excellent characteristics of non-line-of-sight (NLOS) propagation, high confidentiality, and flexible networking, has gradually assumed significant importance in research studies. However, UV NLOS communication relies on the scattering of aerosol particles in the atmosphere, leading to excessive path loss, and the point-to-point communication distance is usually less than 500 m, with the limitations of light power and photodetectors. Therefore, combining UV communication with an ad hoc network to construct a UV network can expand the restricted communication distance, which meets the special needs of covert communication and has high prospects for application. The application scenarios of UV network are mainly secure communication, including military network, emergency services, disaster recovery and other complex electromagnetic environments.

The secure properties of UV include strong anti-interference ability, good confidentiality, and low position resolution:

1. Strong anti-interference ability. On the one hand, owing to the serious loss of optical signals in the atmosphere, it is difficult to interfere with UVC from a long distance. UVC, on the other hand, is largely unaffected by radio-frequency signals because of its special frequency band.
2. Good confidentiality. Owing to the scattering and absorption of atmospheric gas molecules and aerosol particles in the transmission process, the signal attenuation is serious, and it is difficult to eavesdrop outside the range of communication.
3. Low position resolution. The UVC band is the non-visible band, and the naked eye cannot distinguish the signal source position through observation. Moreover, in the NLOS mode, the scattering angle is large and the directivity is weak. Even if the scattered light signal can be detected, the signal source orientation cannot be identified, which can provide strong protection for the signal source and its surrounding facilities.

The main modification is the 1st paragraph of the INTRODUCTION section on page 1.

 

2. Indicate any safety concerns from the use of FSO UV light?

Answer:

Thanks for your kindly expressed concern. Now we have added a note about safety concerns to the manuscript.

When UV is used for indoor communications, the International Commission on Non-Ionizing Radiation Protection and the International Electrotechnical Commission govern UV radiation exposure power limits. In the UV-C band, the allowable continuous exposure limits at 200, 270, and 280 nm are 100, 3, and 3.4 mJ/cm² for an 8 h exposure, respectively.

The National Institute for Occupational Safety and Health also proposed the UV exposure standard for conventional UV sources. This standard is based on an envelope action spectrum, which combines the action spectrum for photokeratitis with Caucasian skin erythema over a highly-sensitive UV wavelength range, that is, 200 to 315 nm. The proposed standard defines 4.6 mJ/cm² as the maximum safe or permissible exposure to 260 nm and 4.7 mJ/cm² corresponding to 290 nm. As defined by this standard, any exposure that is below these limits is considered safe for humans. This is an important safety consideration that a UVC system design should obey.

When UV light is used for outdoor communication, the safe power of UV can be used at a higher level. In addition, the UV source has the advantages of high confidentiality, non-line-of-sight communication, and strong anti-interference. These characteristics are rarely available in other signal sources, and UV has great potential in military communication. It should be noted that the high-power UV light source does cause some damage to the eyes and skin. Therefore, when it is used, the power of the light source should be strictly controlled according to the safety regulations.

The main modification is the 1st paragraph of the INTRODUCTION section on page 1.

 

3. The performance of the proposed protocol as authors called is better than the conventional ones. However, will there be some trade-off on the complexity of the system?

Answer:

Thanks for your kindly mention. Considering the characteristics of the physical UV channel, the CL-LA MAC protocol is proposed, which is based on the clustering topology and learning automata (LA) algorithm. In order to achieve dynamic time-slot-allocation and fully utilize the channel, the overhead of reinforcement learning (RL) algorithm is added to the frame structure. The complexity of the system does increase somewhat, but it is still within acceptable limits. Compared with other protocols that can achieve similar functions, the CL-LA protocol does not involve complex handshake mechanisms, which are greatly simplified in complexity. In addition, the performance of the proposed protocol is better than the conventional ones.

The main modification is the Discussion section on page 15.

 

4. Language and Style: While the overall writing is clear, consider revisiting some sentences for improved conciseness. Additionally, ensure consistent use of terminology throughout the paper.

Answer:

Thank you for your valuable advice. We acknowledge that there is room for improvement in terms of English usage in our manuscript. We have thoroughly revised the manuscript to enhance the clarity and fluency of the language and ensure consistent use of terminology. Furthermore, we have sought an official professional language editing services to help improve the language quality of our manuscript.

The main modifications include the INTRODUCTION section on page 1, the Simulation and Analysis section on page 10, and the Discussion section on page 15.

 

5. Bibliography should be enhanced. Strengthen the discussion by explicitly connecting your findings to existing bibliography.

Answer:

Thank you for your careful reading! We regret that we did not explicitly explain the connection of the findings to existing bibliographies; here, we have added the corresponding explanation and bibliographies. The main modifications include the Introduction section, and Discussion section.

1. Introduction

Ultraviolet (UV) networks are mobile networks that use UV radiation as communication carriers to realize wireless multi-hop communication among UV communication terminals [1]. These networks can be applied to military networks, emergency services, disaster recovery, and other complex electromagnetic environments due to their excellent non-line-of-sight (NLOS) communication, high security, and all-weather operation [2]. The secure properties of UV networks include strong anti-interference ability, good confidentiality, and low position resolution [3]; thus, they have become a research hotspot recently [4]. Unfortunately, high-power UV light sources cause damage to the eyes and skin; there-fore, the power of the light source should be strictly controlled according to safety regulations [5].

Properly setting up and optimizing the media access control (MAC) protocol is significant for improving the network performance [6]. This protocol is one of the key technologies for realizing communication through UV networks. At present, the UV MAC protocols are relatively lacking and fall into two main categories [5, 7]. The first class concerns competition-based protocols, which require less control overhead and are more suitable for changes in network topology [8]. However, as traffic loads increase, there are more transmission collisions, resulting in the network performance significantly deteriorating. Furthermore, competition-based protocols do not ensure the quality of service (QoS) and bounded network delays in highly dense scenarios.

On the other hand, the second class has received increasing attention. In competition-free MAC protocols, a certain channel is allocated to a single terminal at a time. When a terminal transmits data on this channel, no other terminal competes for channel re-sources. The competition-free protocol can guarantee the QoS of the data, and its performance is better than that of the competition-based MAC protocol under a high traffic load. Liu et al. [9] proposed that the competition-free protocol had better QoS than the competition-based protocol that worked based on the carrier sense multiple access (CSMA) mechanism.

Studies on using a competition-free MAC protocol have been carried out to provide an optimized access mechanism in the bandwidth-constrained solar-blind UV band [10]. Compared with other MAC protocols, the time division multiple access (TDMA) protocol is a representative competition-free protocol with several preponderances, such as convenient networking, good communication reliability, and bounded network delay [11]. However, there is an unavoidable defect in this protocol regarding its fixed-slot allocation, that is, the time slot allocated to the terminals is inversely proportional to the number of nodes, resulting in a longer network delay and unsatisfactory throughput in multi-node networks [12].

The traditional TDMA protocol can be improved by employing the concept of clustering in cognitive radio ad hoc networks [13]. In the clustering protocol, the number of nodes that interfere with one other is limited, solving the problem of the network performance rapidly deteriorating as the number of nodes increases [14]. Moreover, this protocol provides convenient topology adjustment in the cluster [15, 16].

In practical network scenarios, the load and channel-resource requirements of each terminal significantly vary, and the fixed-slot allocation mechanism cannot fully utilize the channel. Therefore, in this study, we optimized the clustering mechanism even further using a reinforcement learning (RL) algorithm. To address the channel utilization problem, the cluster leader (CL) uses a smart learning automata (LA) model to monitor the intra-cluster transmissions and learn the traffic parameters of its cluster nodes (CNs), avoiding any complications [17]. The LA can help the CL optimize the allocation of intracluster time slots thereby maximizing the channel utilization.

2. Discussion

At present, the UV protocols are relatively lacking, especially in multi-node networks [5,7]. For the application in large-scale UV node networking, this study improves the conventional TDMA protocol based on the clustering topology and LA algorithm. The simulation results reveals that the CL-LA MAC protocol obtains better network performance, proving its suitability for multi-node UV networking.

The overhead of the RL algorithm is added to the frame structure to achieve dynamic time-slot-allocation and fully utilize the channel. The complexity of the system does increase but still within acceptable limits. Compared with other protocols that can achieve similar functions, the CL-LA protocol does not involve complex handshake mechanisms, simplifying its complexity. In addition, the performance of the proposed protocol is better than the conventional ones [27]. The CL-LA MAC protocol provides practical guidance for the development of UV multi-node networking.

As the cluster head, the CL had to frequently send important control data. However, because the FIFO method was adopted in the current CL-LA mechanism, forwarded packets from the CNs frequently occupied the data-transmission time slot of the CL, resulting in a loss in the packet overflow and a low throughput for the CL, owing to the limited cache. Therefore, in the future, we aim to configurate various caches to classes of service or set a higher priority for the CL data to guarantee the transmission of control data.

The main modifications include the INTRODUCTION section on page 1, the Discussion section on page 15, and the References on page 16.

Author Response File: Author Response.doc

Reviewer 2 Report

Comments and Suggestions for Authors

The authors propose an enhanced clustering Time Division Multiple Access (TDMA) MAC protocol based on clustering and learning automata for UV networks. This article is of significant interest to the research community in the UV networks field. However, there are some issues that require attention:

 

• The authors should include a comparative analysis with relevant prior research findings to contextualize their contributions and clarify advancements.
• Most of the figures are unclear.
• The references list should be expanded to include more recent sources.
• Some of the lengthy equations in section 4 could be moved to an appendix for better organization of the paper.
• In Figures 10, 12, 14, and 16, the authors should explain why the throughput remains constant regardless of the increase in data arrival intensity.
• The authors should explain the reason of not using a high data rate (higher than 50 kbps) in the simulations.

Comments on the Quality of English Language

The English language is fine.

Author Response

The authors propose an enhanced clustering Time Division Multiple Access (TDMA) MAC protocol based on clustering and learning automata for UV networks. This article is of significant interest to the research community in the UV networks field. However, there are some issues that require attention:

1. The authors should include a comparative analysis with relevant prior research findings to contextualize their contributions and clarify advancements.

Answer:

We are very grateful to the reviewer for reviewing the paper so carefully. We regret that we did not explicitly explain the connection of the findings to relevant prior research findings. According to reviewer’ advice, we have added the comparative analysis and existing bibliographies. The main modifications include the Introduction section, and Discussion section.

1. Introduction

Ultraviolet (UV) networks are mobile networks that use UV radiation as communication carriers to realize wireless multi-hop communication among UV communication terminals [1]. These networks can be applied to military networks, emergency services, disaster recovery, and other complex electromagnetic environments due to their excellent non-line-of-sight (NLOS) communication, high security, and all-weather operation [2]. The secure properties of UV networks include strong anti-interference ability, good confidentiality, and low position resolution [3]; thus, they have become a research hotspot recently [4]. Unfortunately, high-power UV light sources cause damage to the eyes and skin; there-fore, the power of the light source should be strictly controlled according to safety regulations [5].

Properly setting up and optimizing the media access control (MAC) protocol is significant for improving the network performance [6]. This protocol is one of the key technologies for realizing communication through UV networks. At present, the UV MAC protocols are relatively lacking and fall into two main categories [5, 7]. The first class concerns competition-based protocols, which require less control overhead and are more suitable for changes in network topology [8]. However, as traffic loads increase, there are more transmission collisions, resulting in the network performance significantly deteriorating. Furthermore, competition-based protocols do not ensure the quality of service (QoS) and bounded network delays in highly dense scenarios.

On the other hand, the second class has received increasing attention. In competition-free MAC protocols, a certain channel is allocated to a single terminal at a time. When a terminal transmits data on this channel, no other terminal competes for channel re-sources. The competition-free protocol can guarantee the QoS of the data, and its performance is better than that of the competition-based MAC protocol under a high traffic load. Liu et al. [9] proposed that the competition-free protocol had better QoS than the competition-based protocol that worked based on the carrier sense multiple access (CSMA) mechanism.

Studies on using a competition-free MAC protocol have been carried out to provide an optimized access mechanism in the bandwidth-constrained solar-blind UV band [10]. Compared with other MAC protocols, the time division multiple access (TDMA) protocol is a representative competition-free protocol with several preponderances, such as convenient networking, good communication reliability, and bounded network delay [11]. However, there is an unavoidable defect in this protocol regarding its fixed-slot allocation, that is, the time slot allocated to the terminals is inversely proportional to the number of nodes, resulting in a longer network delay and unsatisfactory throughput in multi-node networks [12].

The traditional TDMA protocol can be improved by employing the concept of clustering in cognitive radio ad hoc networks [13]. In the clustering protocol, the number of nodes that interfere with one other is limited, solving the problem of the network performance rapidly deteriorating as the number of nodes increases [14]. Moreover, this protocol provides convenient topology adjustment in the cluster [15, 16].

In practical network scenarios, the load and channel-resource requirements of each terminal significantly vary, and the fixed-slot allocation mechanism cannot fully utilize the channel. Therefore, in this study, we optimized the clustering mechanism even further using a reinforcement learning (RL) algorithm. To address the channel utilization problem, the cluster leader (CL) uses a smart learning automata (LA) model to monitor the intra-cluster transmissions and learn the traffic parameters of its cluster nodes (CNs), avoiding any complications [17]. The LA can help the CL optimize the allocation of intracluster time slots thereby maximizing the channel utilization.

2. Discussion

At present, the UV protocols are relatively lacking, especially in multi-node networks [5,7]. For the application in large-scale UV node networking, this study improves the conventional TDMA protocol based on the clustering topology and LA algorithm. The simulation results reveals that the CL-LA MAC protocol obtains better network performance, proving its suitability for multi-node UV networking.

The overhead of the RL algorithm is added to the frame structure to achieve dynamic time-slot-allocation and fully utilize the channel. The complexity of the system does increase but still within acceptable limits. Compared with other protocols that can achieve similar functions, the CL-LA protocol does not involve complex handshake mechanisms, simplifying its complexity. In addition, the performance of the proposed protocol is better than the conventional ones [27]. The CL-LA MAC protocol provides practical guidance for the development of UV multi-node networking.

As the cluster head, the CL had to frequently send important control data. However, because the FIFO method was adopted in the current CL-LA mechanism, forwarded packets from the CNs frequently occupied the data-transmission time slot of the CL, resulting in a loss in the packet overflow and a low throughput for the CL, owing to the limited cache. Therefore, in the future, we aim to configurate various caches to classes of service or set a higher priority for the CL data to guarantee the transmission of control data.

The main modifications include the INTRODUCTION section on page 1, the Discussion section on page 15, and the References on page 16.

 

2. Most of the figures are unclear.

Answer:

Thank you for pointing out this problem. We apologize for the unclear figures. We have redrawn all the figures and resized the figures in the revised manuscript to increase the readability of the manuscript.

The main modifications are the figures.

 

3. The references list should be expanded to include more recent sources.

Answer:

We appreciate it very much for this good and sweet suggestion, and we have added the recent references to support the motivation and importance of this study.

The main modifications is the References on page 16.

 

4. Some of the lengthy equations in section 4 could be moved to an appendix for better organization of the paper.

Answer:

Thanks for your kindly mention. We have adapted parts of the mathematical analysis (section 4) to the Appendix for a better reading experience.

The main modifications include the Stable Cache Probability Distributions Based on MC section (section 4) on page 7 and the Appendix section on page 17.

 

5. In Figures 10, 12, 14, and 16, the authors should explain why the throughput remains constant regardless of the increase in data arrival intensity.

Answer:

Thank you for your valuable advice. According to reviewer’ advice, we have added the relevant explanations.

The channel is set to a single wavelength channel, and multiple network nodes share a single wavelength channel. The communication between nodes needs to obtain channel resources under the CL-LA protocol.

Firstly, the intensity of the data arrival increased, leading to an increase in the throughput. With the continuous increase of intensity, each time slot in the network is gradually occupied and the channel tends to be saturated. Even if the data intensity continues to increase, the network throughput cannot increase and remains constant.

So there will be an upper bound on the throughput in Figures 10, 12, 14, and 16.

The main modification is the 4th paragraph of the Simulation and Analysis section on page 11.

 

6. The authors should explain the reason of not using a high data rate (higher than 50 kbps) in the simulations.

Answer:

Thank you for your careful reading! It is our negligence not to describe the reason of not using a high data rate in the simulations. According to reviewer’ advice, we have added the reason for data rate selection.

With the development of UV light sources, photoelectric detection techniques, and modulation coding modes, much higher rates than several Mbps are achieved in the UV point-to-point communication system [A. Vavoulas, H. G. Sandalidis, N. D. Chatzidiamantis, Z. Xu, and G. K. Karagiannidis, “A survey on ultraviolet C-band (UV-C) communications,” IEEE Commun. Surv. Tut. 21(3), 2111–2133 (2019)]. However, the UV networks is mainly limited by the networking method, topology control and terminal movement, and the actual working rate can only be limited to below 100 kbps [J. B. Mao, J. H. Li, J. Y. Wang, et al, “A novel ultraviolet communication channel access protocol based on competition mechanism,” Optik 273, 170426 (2023)].To ensure reliable transmission of information, the data rate is selected as 50 kbps in the simulations of multi-node UV network.

In view of the high requirement for real-time performance in military application scenarios of UV network, the UV network is currently mainly used to transmit instructions and control signals, and it is sufficient for this kind of service at several ten kbps. The target data rates reach several Mbps to realize real-time image signal transmission in the secure military application field.

The main modification is the 2nd paragraph of the Simulation and Analysis section on page 10.

Author Response File: Author Response.doc