A Disruption Tolerant Distributed Routing Algorithm in LEO Satellite Networks
Round 1
Reviewer 1 Report
The authors represent disruption tolerant distributed routing algorithm, which makes low earth orbit (LEO) constellation network resilient to packet loss. To describe the proposed system, the authors presents the LEO satellite network model, system overview, and detailed description of algorithms, which compose the proposed system. Also, the authors conduct simulation to represent the performance of the system.
However, the assumptions and presentations of this paper needs to be improved.
The mathematical notation in the manuscript should be represented with italic and subscripts of variables must be clearly marked. Without this, the manuscript is hard to read and the reader might be confused.
In line 80, the authors claim that there is no large-scale buffer to support the buffering of data packets. However, there are no reference and evidence about the limited buffer.
In section 3, the title is 'Overview of our scheme'. But, this section represents about detailed operation procedure which is hard to read. Therefore, the authors should represent a simplified operation overview of the overall system at the beginning of the section.
In line 293, the authors represents the propagation delay t1 as (3Lh1+Lh3+2Lv)/c, but it seems to be fixed to (2Lh1+Lh3+2Lv)/c. Also, calculation result at line 303-311 should be fixed.
In line 321, the authors said that "the LEO satellite network has relatively dense business at low latitudes, and relatively few at high latitudes, ~". However no evidence nor reference is represented in the manuscript.
In section 5.2.1, the authors set link failure probability as 0.0001, 0.001, 0.002, and 0.003. However, the probability range seems to be narrow. The authors should describe the reason for determining the link failure probability. If not, the authors presents more simulation result with wider link failure probability range.
In section 5.2.2, the authors set average link load as 0.2, 0.24, 0.28, 0.33, and 0.38. This range also seems to be narrow. As the result of section 5.2.1, the authors should describe the simulation result with reason or add additional simulation results.
In section 6, the conclusion has four sentences, which briefly summarizes the manuscript. In this conclusion, the authors suggests an novel idea which could improve fault tolerance of LEO satellites network. However, the authors presents their achieved results with only one sentence. To improve the manuscript, a more detailed explanation or meaning of the result is required.
Also, the manuscript should be reviewed in detail with following points.
- There are many typos
- All equation and algorithms should be represented in their appropriate format for better readability.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The paper presents a disruption-tolerant routing approach for LEO constellations with unstable laser links, called DTDR. The topic is interesting and timely, and the paper is well written and easy to follow.
However, this work misses significant background on LEO constellation routing, a widely studied topic. This weakness forbids a proper assessment on what is the concrete contribution to the state-of-the-art.
On the one hand, the paper ignores generic routing algorithms for delay/disruption tolerant networks, such as Contact Graph Routing (CGR), which should be reviewed. On the other hand, the proposed scheme is not compared with classical (and generic) distributed routing protocols like AODV and OLSR which could also react to failures (i.e., topology changes) in mesh networks like the one proposed by the authors.
Moreover, a plethora of routing solutions specific for LEO constellations exits, to name a few:
Virtual topology (network is discretized into sixed time slices)
- Dynamic Virtual Topology Routing (DVTR)
- Finite State Automata (FSA)
- Compact Explicit Multi-Path Routing (CEMR)
- Explicit Load Balancing (ELB)
- Priority-based Adaptive Routing (PAR)
- Dynamic Detection Routing Algorithm (DDRA)
Virtual node (network is discretized into fixed ground locations)
- Localized Zone Distributed Routing (LZDR)
- Datagram Routing Algorithm (DRA)
- Low-complexity Probabilistic Routing Algorithm (LPR)
- Destruction-resistant On-Demand Routing (DODR)
Although authors cite some of them as previous work, the manuscript is not clear on which is the main difference/contribution of this work. The argumentation of using link load information against link state is poor and unconvincing. Furthermore, to prove the authors’ claim on DTDR, the evaluation section should in fact compare DTDR with well-chosen candidates among the existing solutions besides DRA and the SPF baseline.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
The authors have improved the manuscript based on given suggestion. However, there are some points to improve, therefore I composed replies of the authors response.
Point 1: The authors corrected their mathematical notation in appropriate form. However, equation (5) is not aligned as other equations. Also, the calculation methods, including hop direction decision, are hardly distinguished from other plain text. This should be improved to increase the readability and quality of the manuscript.
Point 2: I understand and agree with the authors response which the FPGA has small-sized internal BRAM, and only with internal BRAM, the buffer stores data only for very small duration (6.28 ms) with maximum ISL link rate. If there are other works about small-sized buffer of satellites, please add more reference of the works.
Point 3: The authors improved their section with adding overview subsection. However, brief description of the authors’ proposed algorithm is required for better readability. Also, I suggest the title of section 3 and subsection 3.1 as “Disruption Tolerant Distributed Routing Algorithm” and “Algorithm Overview”.
Point 4: I understand the authors response.
Point 5: I checked the reference which the authors mentioned.
Point 6 & 7: Thanks for add more simulation results. However, I suggest adding supportive data about the average link failure probability and the average load of network link.
Point 8: The conclusion is enriched with more information about the authors' proposed routing scheme.
Point 9: The manuscript has been polished than older version. However, there are still some points to be improved.
i) There are still some typos in the manuscript (ex: low earth orbit – line 10)
ii) All calculation methods would be distinguished than other plain texts. I suggest the authors to write the calculation methods as algorithm separately like other manuscripts.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report
Authors have applied sufficient updates to satisfy my previous concerns. No further comments.
Author Response
Thank you for your review of our manuscript.
