Intelligent Resource Allocation for Immersive VoD Multimedia in NG-EPON and B5G Converged Access Networks
Round 1
Reviewer 1 Report
Comments and Suggestions for Authors1- The contribution of this work is low, kindly make additional effort to your contribution.
2- Based on suggestion no. 1 proved more results to enhance your contribution.
3- The abstract must be contain some results from the proposed work.
4- The conclusion must be contain the best numerical results of this work
5- The authors must be provide a comparison table to compare between this work and the published related work.
Author Response
Reviewer #1:
Comments and Suggestions for Authors
Comment 1- The contribution of this work is low, kindly make additional effort to your contribution.
Response 1: We thank the reviewer for the feedback. In the revised manuscript, we have emphasized the novelty of our work by highlighting the following changes in our manuscript:
“The main contribution of this work is the design and implementation of a comprehensive SDN-based intelligent resource allocation framework tailored for converged NG-EPON and B5G X-haul access networks, addressing the stringent requirements of immersive VoD (IVoD) services. The proposed unified architecture introduces an SD-OLT with FPGA and ANN capabilities. These components enable real-time traffic prediction and dynamic bandwidth control. A predictive bandwidth allocation mechanism based on an ANN with a backpropagation learning model is enhanced by a dynamic threshold adjustment scheme that continuously fine-tunes prediction accuracy according to the network feedback. The Immersive media service–dynamic bandwidth allocation (IMS-DBA) algorithm is developed to operationalize this model.“
Comment 2- Based on suggestion no. 1 proved more results to enhance your contribution.
Response 2: Thanks for your valuable suggestion and comment. To further enhance the contribution of our work, we have improved Section 6. System Performance and Discussion and The following changes have been made in the manuscript:
“The computational complexity of the ANN backpropagation model integrated within the FPGA is evaluated based on the number of neurons, layers, and input features. For a network with layers, each having neurons, and a sliding window of input time steps, the forward and backward pass complexity is estimated as:
Given our implementation with 3 layers (input, one hidden, output), each with ≤100 neurons, and a sliding window size of 5, the model’s per-inference runtime remains in the sub-millisecond range, suitable for real-time operations on FPGA hardware. This runtime is further optimized by parallelizing matrix operations using dedicated logic blocks on the FPGA. We have also evaluated the total control overhead regarding the network's number of ONUs (N). The IMS-DBA operates on a cycle-based loop, and its computational cost grows linearly with the number of ONUs, i.e.,
,
where represents the per-ONU prediction and bandwidth grant computation (which remains constant and hardware-accelerated). In our OPNET simulations with 32 active SD-ONUs, the total computational cycle per DBA round is below 1 ms, including ANN prediction, bandwidth grant calculation, and ANN model update. This aligns with the cycle time (1.0 ms) and does not introduce processing bottlenecks.”
We have further elaborated and discussed the evaluation of each simulation result. The following changes have been made in the manuscript:
“6.1. Mean Packet Delay: All cases exhibit a similar lower EF delay across all load conditions due to its higher transmission priority, as shown in Fig. 5(a). Both IMS-DBA and LSTM-GARAA fairly maintain EF delay below 1.0ms during light load. However, IMS-DBA consistently delivers slightly lower EF delay at a higher traffic load compared to LSTM-GARAA up to around 80% traffic load, beyond which the delay increases sharply. A similar trend is shown for AF traffic, including immersive video, moderately sensitive to delay in Fig. 5(b). IMS-DBA outperforms LSTM-GARAA in all three traffic configurations, especially when the load crosses 80% and the delay increases more noticeably. The BE traffic is given the lowest priority and hence incurs higher delay under traffic load. While both schemes show increasing BE delay with traffic load, IMS-DBA maintains a more stable delay progression as shown in Fig. 5(c). Finally, the Total End-to-End delay reflects the aggregate EF, AF, and BE delay to reflect overall network responsiveness as shown in Fig. 5(d). The delay curve remains relatively smooth until the traffic load reaches 80%, beyond which the network gets congested. However, IMS-DBA avoids sudden spikes in delay while efficiently adjusting grants based on the feedback loop enabling IVoD and other delay-sensitive traffic to be handled more efficiently.
6.2. Packet Drop: Across all traffic configurations, IMS-DBA achieves significantly lower BE packet drop comparing LSTM-GARAA. For instance, in LSTM-GARAA-163 and LSTM-GARAA-153, the BE drop occurs at 60% traffic load while the IMS-DBA exhibits 70% with a drop rate reaching a maximum at 11.5% when compared to almost 14% in LSTM-GARAA. The gap between LSTM-GARAA and IMS-DBA becomes more significant at 90%-100% traffic load. This reflects IMS-DBA predictive learning and dynamic adjustment through a feedback mechanism.
6.3. System Throughput: Across all load levels, IMS-DBA consistently achieves marginally higher throughput than LSTM-GARAA. At 100% offered load, IMS-DBA peaks at approximately 21.9 Gbps, whereas LSTM-GARAA reaches around 21.3 Gbps. This improvement is noteworthy in access networks where optimizing throughput promptly leads to improved utilization and decreased service latency, particularly for high-bandwidth applications like IVoD. The performance advantage of IMS-DBA can be attributed to its intelligent bandwidth allocation technique based on historical data and real-time adjustments which leads to overall improved throughput.
More significantly added another section 6.4 for AF jitter, the following changes have been added to the manuscript:
6.4. Jitter
: The variation in delay between successive packets (i.e., jitter) is a critical performance metric for immersive VoD (IVoD) and real-time multimedia applications. High jitter can result in frame misalignment, buffering, or interruptions in playback, especially for XR-based content where synchronization is paramount. Figure 8 illustrates the AF jitter comparison between IMS-DBA and LSTM-GARAA across the three traffic scenarios (case 163, 153, 145) and varying offered load conditions. Across all scenarios, IMS-DBA outperforms LSTM-GARAA in minimizing jitter values. Notably, at lower load (10%-50%), both schemes show similar performance, although IMS-DBA maintains slightly less. However, as the traffic load increases, the difference becomes more prominent between the two schemes, with LSTM-GARAA-163, the AF Jitter reaches close to 0.75 while the IMS-DBA fairly maintains below 0.45. Moreover, the smoother slope of the IMS-DBA jitter curve indicates its ability to make real-time adjustments with ANN predictions and dynamic threshold tuning.
Comment 3- The abstract must be contain some results from the proposed work.
Response 3: Thank you for your valuable suggestions, the abstract has been revised consistently to include results from the simulation. Specifically, we mention the reduced mean packet delay and improved packet drop achieved by our proposed technique IMS-DBA when compared to LSTM-GARAA under various traffic scenarios.
“Immersive content streaming services are becoming increasingly popular in video on demand (VoD) due to the rising interest in extended reality (XR) and spatial experiences. Unlike traditional VoD, immersive VoD (IVoD) offers more engaging and interactive content beyond conventional 2D video. IVoD requires substantial bandwidth and minimal latency to deliver its interactive XR experiences. This research examines intelligent resource allocation for IVoD services across NG-EPON and B5G X-haul converged networks. A proposed software-defined networking (SDN) framework employs artificial neural networks (ANN) with a backpropagation technique to predict bandwidth control, based on traffic patterns and network conditions. The new immersive video storage, field-programmable gate array (FPGA), Queue Manager, and logical layer components are added to the existing OLT and ONU hardware architecture to implement the SDN framework. The SDN framework manages the entire network, predicts bandwidth requirements, and operates the immersive media dynamic bandwidth allocation (IMS-DBA) algorithm to efficiently allocate bandwidth to IVoD network traffic, ensuring that QoS metrics are met for IM services. Simulation results demonstrate that the proposed framework significantly enhances the mean packet delay by up to 3% and improves packet drop probability by up to 4% as traffic load varies from light to high in different scenarios, leading to enhanced overall QoS performance.”
Comment 4- The conclusion must be contain the best numerical results of this work
Response 4: We have updated the conclusion to include numerical highlights from our evaluation. The following changes has been made to the conclusion which is annexed and elaborated:
“This research investigates the challenges and solutions associated with intelligent resource allocation for IVoD services over advanced networks, including NG-EPON and B5G X-haul access networks. IVoD, characterized by incorporating XR and spatial technologies, transcends conventional VoD by offering dynamic, interactive, and engaging experiences. These advancements require substantial high bandwidth and ultra-low latency to facilitate seamless delivery. Further, this study presents a resource management framework that utilizes SDN to monitor and optimize optical resources. Predictive technique, ANN with backpropagation, is employed to forecast bandwidth demands based on historical traffic data, internet behavior, and network conditions. The backpropagation techniques constantly adjust thresholds to align actual and estimated usage, thereby decreasing latency and increasing throughput, thus enhancing QoS. The FPGA and IVoD storage components are added to the OLT hardware architecture to effectively handle the training and testing of data utilizing the ANN backpropagation model and server mechanism designed for caching immersive video content, adapted to predicted traffic patterns and bandwidth utilization. A new IMS-DBA algorithm is proposed to handle network traffic efficiently based on the classified priorities. The methodology focuses on dynamic adjustments to bandwidth predictions through a feedback loop derived from network usage reports, thereby minimizing over- and under-predictions. This iterative process enhances resource utilization and prioritizes high-priority traffic while accommodating lower-priority demands as opportunities arise. The IMS-DBA is compared with the LSTM-GARAA to examine the efficiency of the proposed DBA. Simulation results show that the proposed IMS-DBA improves overall QoS metrics regarding mean packet delay, packet drop probability, and system throughput. The proposed IMS-DBA improves the mean packet delay by up to 3% and packet drop probability by up to 4% in different traffic loads and patterns. In conclusion, this research study illustrates that integrating ML models for predictive bandwidth allocation and a closed-loop feedback system enables scalable, efficient, and responsive IVoD delivery to user demands. These innovations establish a foundation for tackling future challenges in immersive multimedia services, ensuring sustained high performance and accessibility within increasingly complex network environments. Our future work focuses on implementing resource allocation with dynamic ML model implementation in the real test bed.”
Comment 5- The authors must be provide a comparison table to compare between this work and the published related work.
Response 5: Thank you for your valuable suggestions. We have added a new comparison table (Table 1 in related work section), that contrasts our proposed ANN through the backpropagation technique. The following manuscript has been updated:
“Table 1 shows the overall comparative studies of estimating bandwidth schemes.”
Table 1. Estimating bandwidth schemes.
Reference paper |
Method/Technique |
ML |
[13] |
Constant bit rate |
No |
[14] |
Packet arrival time |
No |
[15] |
Regression model |
No |
[16] |
Supervised learning |
Yes |
[17] |
ANN through a discreet dynamical system |
Yes |
[18] |
ANN through MLP |
Yes |
[Our technique] |
ANN through backpropagation |
Yes |
Author Response File: Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsThe paper Intelligent Resource Allocation for Immersive VoD Multimedia in NG-EPON & B5G Converged Access Networks is focused on modern NG-EPON networks and it proposes an innovative and intelligent resource allocation technique to cover VoD multimedia demands. The method is based on a centralized framework of Software-Defined Networking (SDN), integrating Artificial Neural Networks (ANN) with backpropagation technique for predictive bandwidth control. Moreover, an innovative Immersive Media Dynamic Bandwidth Allocation (IMS-DBA) algorithm is proposed that allocates bandwidth based on the traffic types and other QoS metrics.
First, the paper is well written and organized. The language level throughout the entire paper is very good and I was able to notice only some minor language errors and typos, on the other hand, I recommend performing proofreading prior to its publication. The language is technically sound, all abbreviations are properly described. The paper is well organized. It contains sufficient theoretical background and formulation of the problem together with the introduction into the topic and the survey of existing sources. The following part describes the proposed framework for centralized storage, delivery and control of demands in PON network, while the next section contains the methodology and the description of the proposed IMS-DBA algorithm. Next, the ANN Predicted Bandwidth Allocation Scheme is introduced. The next section contains the simulations and comparisons of the proposed method and the existing technique. Finally, conclusions are provided.
The framework proposed in section 3 is a centralized node providing the control of bandwidth allocation in the PON. In my opinion, this is a disadvantage and the drawback of the proposed technique, as the building of such centralized control center would require additional resources.
Next, I miss any computational demand evaluation of the proposed algorithm, the article contains no evaluation of the complexity, time consumption and computer time calculation consumption of the proposed method. Moreover, since the entire PON with its all active ONUs should be controlled using the proposed technique, the simulation should include the computational complexity analysis for controlling of entire network.
According to the simulation parameters in Table 1, only 32 active ONUs are considered. I recommend repeating the simulation with 64 units as well.
Figure 5 contains simulations of transmission delays. However, especially for multimedia services, the value of delay deviation between the frames delivered for each service is important.
Due to that I recommend performing some major improvements in the article.
Author Response
Reviewer #2:
Comments and Suggestions for Authors:
The paper Intelligent Resource Allocation for Immersive VoD Multimedia in NG-EPON & B5G Converged Access Networks is focused on modern NG-EPON networks and it proposes an innovative and intelligent resource allocation technique to cover VoD multimedia demands. The method is based on a centralized framework of Software-Defined Networking (SDN), integrating Artificial Neural Networks (ANN) with backpropagation technique for predictive bandwidth control. Moreover, an innovative Immersive Media Dynamic Bandwidth Allocation (IMS-DBA) algorithm is proposed that allocates bandwidth based on the traffic types and other QoS metrics.
First, the paper is well written and organized. The language level throughout the entire paper is very good and I was able to notice only some minor language errors and typos, on the other hand, I recommend performing proofreading prior to its publication. The language is technically sound, all abbreviations are properly described. The paper is well organized. It contains sufficient theoretical background and formulation of the problem together with the introduction into the topic and the survey of existing sources. The following part describes the proposed framework for centralized storage, delivery and control of demands in PON network, while the next section contains the methodology and the description of the proposed IMS-DBA algorithm. Next, the ANN Predicted Bandwidth Allocation Scheme is introduced. The next section contains the simulations and comparisons of the proposed method and the existing technique. Finally, conclusions are provided.
Comment 1- The framework proposed in section 3 is a centralized node providing the control of bandwidth allocation in the PON. In my opinion, this is a disadvantage and the drawback of the proposed technique, as the building of such centralized control center would require additional resources.
Response 1: We appreciate this insight however, as our centralized architecture does introduce some infrastructure overhead, it enables global network visibility, and efficient resource management, and supports AI integration, which is difficult in decentralized systems.
In our context, the centralized SD-OLT plays a pivotal role in enabling real-time, network-wide predictive resource allocation using the ANN-based bandwidth prediction engine. This level of coordination and responsiveness would be challenging to achieve with a fully distributed architecture, particularly when delivering immersive VoD (IVoD) services that are highly sensitive to latency, jitter, and bandwidth fluctuations.
We have added the following text in section 3.2 SD-ONU operations for traffic classification the manuscript as well so as to make our claim strong:
“While the centralized SD-OLT does introduce certain overheads, its benefits, including intelligent traffic control, dynamic bandwidth allocation, and immersive service quality assurance, significantly outweigh the drawbacks; hence, the central office SD-OLT predominantly governs most functions and operations.”
Comment 2- Next, I miss any computational demand evaluation of the proposed algorithm, the article contains no evaluation of the complexity, time consumption and computer time calculation consumption of the proposed method. Moreover, since the entire PON with its all active ONUs should be controlled using the proposed technique, the simulation should include the computational complexity analysis for controlling of entire network.
Response 2: We appreciate the reviewer’s valuable comment. To address this, we have included in the revised manuscript where we provide a quantitative analysis of the computational demands of the proposed ANN-based IMS-DBA scheme. The following changes has been updated in section 6 System Performance Evaluation & Discussion:
“The computational complexity of the ANN backpropagation model integrated within the FPGA is evaluated based on the number of neurons, layers, and input features. For a network with layers, each having neurons, and a sliding window of input time steps, the forward and backward pass complexity is estimated as:
Given our implementation with 3 layers (input, one hidden, output), each with ≤100 neurons, and a sliding window size of 5, the model’s per-inference runtime remains in the sub-millisecond range, suitable for real-time operations on FPGA hardware. This runtime is further optimized by parallelizing matrix operations using dedicated logic blocks on the FPGA. We have also evaluated the total control overhead regarding the network's number of ONUs (N). The IMS-DBA operates on a cycle-based loop, and its computational cost grows linearly with the number of ONUs, i.e.,
,
where represents the per-ONU prediction and bandwidth grant computation (which remains constant and hardware-accelerated). In our OPNET simulations with 32 active SD-ONUs, the total computational cycle per DBA round is below 1 ms, including ANN prediction, bandwidth grant calculation, and ANN model update. This aligns with the cycle time (1.0 ms) and does not introduce processing bottlenecks.”
Comment 3- According to the simulation parameters in Table 1, only 32 active ONUs are considered. I recommend repeating the simulation with 64 units as well.
Response 3: Thank you for your insightful suggestion. While we acknowledge the merit of evaluating the system under a larger number of ONUs, we respectfully clarify that the choice of 32 active ONUs in the simulation was made deliberately based on a balance between computational realism, comparability with previous works (e.g., [6],[10],[20]), and alignment with practical deployment scales in NG-EPON testbeds.
Scientifically, the performance pattern of the proposed IMS-DBA framework is not expected to significantly change with an increased number of ONUs (e.g., 64 instead of 32), because of two main reasons:
- Linear Scaling of the Algorithm: The proposed IMS-DBA algorithm and the ANN-based bandwidth prediction scale linearly with respect to the number of ONUs, as noted in our computational complexity analysis (Section 6.). Each ONU is handled independently within a cycle, and the aggregate bandwidth is dynamically allocated based on real-time demand and priority. Therefore, doubling the number of ONUs would proportionally distribute the available bandwidth across larger entities but would not alter relative performance trends (e.g., mean packet delay, packet drop, throughput behavior) across traffic classes or offered load levels.
- Bandwidth Contention and Saturation Behavior: Our simulations intentionally span low to high offered load conditions (10% to 100%) across multiple traffic scenarios (case 163, 153, 145). These scenarios effectively simulate network saturation and contention conditions that would also occur in a 64-ONU setup, but at proportionally lower individual ONU loads. The pattern of performance degradation under load (e.g., increased packet delay after 80% offered load) results from bandwidth saturation, not the absolute number of ONUs. Therefore, the observed trends—particularly the superior performance of IMS-DBA over LSTM-GARAA—remain valid and consistent under different ONU counts.
Comment 4- Figure 5 contains simulations of transmission delays. However, especially for multimedia services, the value of delay deviation between the frames delivered for each service is important.
Response 4: We sincerely thank the reviewer for highlighting this important aspect. We fully agree that in multimedia service delivery, especially for immersive VoD (IVoD), delay deviation (i.e., jitter) between successive frames is a critical metric, as high jitter can lead to frame skipping, buffering, or degraded immersive experiences.
To provide quantitative support, we have now included an additional metric in the revised manuscript—Standard Deviation of Packet Delay (Jitter)—in a new figure (Figure 8) and accompanying text in Section 6.4: Jitter. The results confirm that the IMS-DBA framework maintains consistently lower jitter levels across all three traffic scenarios compared to the LSTM-GARAA baseline, particularly under higher load conditions where maintaining temporal consistency is more challenging. The following changes have been added in the manuscript.
“6.4. Jitter
: The variation in delay between successive packets (i.e., jitter) is a critical performance metric for immersive VoD (IVoD) and real-time multimedia applications. High jitter can result in frame misalignment, buffering, or interruptions in playback, especially for XR-based content where synchronization is paramount. Figure 8 illustrates the AF jitter comparison between IMS-DBA and LSTM-GARAA across the three traffic scenarios (case 163, 153, 145) and varying offered load conditions. Across all scenarios, IMS-DBA outperforms LSTM-GARAA in minimizing jitter values. Notably, at lower load (10%-50%), both schemes show similar performance, although IMS-DBA maintains slightly less. However, as the traffic load increases, the difference becomes more prominent between the two schemes, with LSTM-GARAA-163, the AF Jitter reaches close to 0.75 while the IMS-DBA fairly maintains below 0.45. Moreover, the smoother slope of the IMS-DBA jitter curve indicates its ability to make real-time adjustments with ANN predictions and dynamic threshold tuning.”
Due to that I recommend performing some major improvements in the article.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsThe revised version is fine
regards
Reviewer 2 Report
Comments and Suggestions for AuthorsI recommend to accept the paper.