ICACT20260355 Question.1
Questioner: pangb@dsp.ac.cn 2026-02-05 오후 4:31:56 |
ICACT20260355 Answer.1
Answer by Auhor xueyy@dsp.ac.cn 2026-02-05 오후 4:31:56
Chrome  Click!! |
| What is the content of the feedback returned to the sender in the closed-loop feedback system? Does it indicate the success of the decoding result, or does it confirm the successful transmission of each data packet? |
This closed-loop feedback is not a per-packet ACK/confirmation mechanism; instead, it is a per-round (or per-window τ) feedback that combines decoding-state information and arrival statistics. Specifically, it indicates whether decoding succeeds (via fail_t) and uses s_t to reflect the overall arrival ratio of coded packets in that round, but it does not confirm the successful transmission of each individual data packet. |
ICACT20260355 Question.10
Questioner: gongh@dsp.ac.cn 2026-02-05 오후 9:16:21 |
ICACT20260355 Answer.10
Answer by Auhor xueyy@dsp.ac.cn 2026-02-05 오후 9:16:21
Chrome Click!! |
| Why does random routing lead to “long-tail latency”? |
Random routing leads to long-tail latency because each packet’s next hop is chosen independently at random, so different packets end up experiencing large variations in path length and link quality. As a result, a small fraction of packets traverse “bad” paths, suffer abnormally long transmission times, or even time out and get lost, and these stragglers dominate the overall session completion delay, producing the typical long-tail effect. |
ICACT20260355 Question.11
Questioner: wangsy@dsp.ac.cn 2026-02-08 오후 2:24:16 |
ICACT20260355 Answer.11
Answer by Auhor xueyy@dsp.ac.cn 2026-02-08 오후 2:24:16
Chrome Click!! |
| What role does the Raptor code play in this solution? And which type of errors/packet losses does it mainly aim to address? |
In the ARRCT solution, the Raptor code serves as a rateless fountain code that provides forward error correction (FEC) capability. It enables the recovery of all original data from a number of received encoded packets slightly greater than the number of source packets, without requiring retransmissions. This significantly enhances transmission reliability and packet-loss resilience. The Raptor code primarily addresses random packet losses and long-tail latency caused by the unpredictability of random routing paths and dynamic network conditions. It is particularly effective in environments with high packet loss rates and fluctuating path quality, mitigating the increase in session completion time due to packet loss and timeout retransmissions. |
ICACT20260355 Question.12
Questioner: shil@dsp.ac.cn 2026-02-09 오전 11:53:04 |
ICACT20260355 Answer.12
Answer by Auhor xueyy@dsp.ac.cn 2026-02-09 오전 11:53:04
Chrome Click!! |
| How does the ARRCT mechanism utilize Raptor coding to address the long-tail latency issue in random routing systems? |
ARRCT employs Raptor coding for forward error correction, adopting a “redundancy-for-waiting” strategy to recover data without retransmissions under random packet loss, thereby significantly reducing long-tail latency. |
ICACT20260355 Question.13
Questioner: shil@dsp.ac.cn 2026-02-09 오전 11:58:42 |
ICACT20260355 Answer.13
Answer by Auhor xueyy@dsp.ac.cn 2026-02-09 오전 11:58:42
Chrome Click!! |
| What is the role of the feedback-driven adaptive redundancy control in ARRCT, and how does it adjust redundancy based on real-time network conditions? |
The feedback control dynamically adjusts redundancy based on real-time packet arrival rate and decoding success/failure: increasing redundancy for reliability under poor network conditions, and reducing it for efficiency when the network is favorable. |
ICACT20260355 Question.14
Questioner: hejf@dsp.ac.cn 2026-02-11 오후 8:29:12 |
ICACT20260355 Answer.14
Answer by Auhor xueyy@dsp.ac.cn 2026-02-11 오후 8:29:12
Chrome Click!! |
| In this paper, how does the adaptive redundancy control decide how much extra redundancy (repair symbols) to add when using Raptor coding? |
Extra redundancy = k×(r_t+1 — r_t) the increase in coded packets from current round to next round.Key logic: Lower arrival rate or decoding failure triggers more added redundancy; historical smoothing prevents drastic fluctuations. |
ICACT20260355 Question.15
Questioner: hejf@dsp.ac.cn 2026-02-11 오후 8:30:12 |
ICACT20260355 Answer.15
Answer by Auhor xueyy@dsp.ac.cn 2026-02-11 오후 8:30:12
Chrome Click!! |
| What is the key benefit of combining random routing with Raptor coding compared to using either one alone? |
Achieves significant transmission efficiency and robustness gains without compromising the system’s original covertness and anti-attack capabilities. |
ICACT20260355 Question.2
Questioner: pangb@dsp.ac.cn 2026-02-05 오후 4:32:10 |
ICACT20260355 Answer.2
Answer by Auhor xueyy@dsp.ac.cn 2026-02-05 오후 4:32:10
Chrome Click!! |
| At which layer do the mentioned algorithms, such as ARRCT, operate? Are they applied at the application layer as coding schemes, or at the transport layer? |
It operates at the application layer. |
ICACT20260355 Question.3
Questioner: tanghr@dsp.ac.cn 2026-02-05 오후 8:55:02 |
ICACT20260355 Answer.3
Answer by Auhor xueyy@dsp.ac.cn 2026-02-05 오후 8:55:02
Chrome Click!! |
| In the ARRCT control law, how do parameters λ (forgetting factor) and κ (failure penalty) affect the smoothness and responsiveness of redundancy adjustment? Why does a λ closer to 1 make the system smoother but potentially less responsive? |
In ARRCT, λ acts as an inertia/smoothing weight on the previous redundancy rt: as λ approaches 1, the update relies more on historical values and less on the latest observation st，o redundancy changes become smoother and less affected by instantaneous randomness, but the system may respond more slowly to sudden network changes; κ is a failure penalty that is applied when decoding fails (failt =1),pushing redundancy up more aggressively to quickly escape an under-provisioned regime and restore decodability. |
ICACT20260355 Question.4
Questioner: yangyn@dsp.ac.cn 2026-02-05 오후 8:58:27 |
ICACT20260355 Answer.4
Answer by Auhor xueyy@dsp.ac.cn 2026-02-05 오후 8:58:27
Chrome Click!! |
| The conclusion mentions future work on theoretical analysis of the control law and overhead profiling. What specific tools from stochastic approximation will be used, and how will you address the feedback overhead for small data blocks? |
Concretely, this will follow the standard stochastic-approximation route for noisy iterative updates (e.g., ODE-based/stability-style analysis) to explain when and how
𝑟𝑡 tracks the optimum in random-routing environments.
For small data blocks, I will explicitly address the fact that the benefit of adaptation can be reduced because feedback and state-evaluation overhead is no longer negligible. I will therefore profile the overhead quantitatively—both bandwidth consumption and processing time—and incorporate these measurements (including practical encoding/decoding costs) into the completion-time model, so the evaluation reflects the net end-to-end gain when blocks are small. |
ICACT20260355 Question.5
Questioner: yangyn@dsp.ac.cn 2026-02-05 오후 9:00:50 |
ICACT20260355 Answer.5
Answer by Auhor xueyy@dsp.ac.cn 2026-02-05 오후 9:00:50
Chrome Click!! |
| Why was Raptor coding specifically chosen over other FEC schemes for integration with random routing, and how does its rateless property complement the unpredictability of KRRD paths. |
I chose Raptor coding because random routing (KRRD) makes end-to-end delivery highly unpredictable, and Raptor’s rateless nature fits that uncertainty: I can generate as many coded packets as needed, and the receiver can decode once it has collected enough packets. |
ICACT20260355 Question.6
Questioner: yangyn@dsp.ac.cn 2026-02-05 오후 9:06:07 |
ICACT20260355 Answer.6
Answer by Auhor xueyy@dsp.ac.cn 2026-02-05 오후 9:06:07
Chrome Click!! |
| The results in Figure 7 show that ARRCTs speedup advantage becomes significant only for data sizes above 20MB. What causes this threshold effect, and how does the feedback overhead impact small-scale transmissions? |
In Fig. 7, this “20 MB threshold” is mainly because ARRCT introduces extra state evaluation and closed-loop feedback per round, and when the data block is small (e.g., 1–10 MB) the baseline transmission time is already short, so this fixed/near-fixed overhead becomes noticeable and can even offset the benefit of adaptive redundancy—hence the speedup stays around 1 or dips slightly below 1. As the data size grows, that overhead is amortized, while static redundancy increasingly mismatches the time-varying network conditions, causing avoidable overhead; therefore, beyond ~20 MB the adaptive scheme’s advantage becomes consistently significant. |
ICACT20260355 Question.7
Questioner: yangyn@dsp.ac.cn 2026-02-05 오후 9:10:45 |
ICACT20260355 Answer.7
Answer by Auhor xueyy@dsp.ac.cn 2026-02-05 오후 9:10:45
Chrome Click!! |
| What are the two main problems that ARRCT aims to address? |
ARRCT aims to address tSecond, using a fixed redundancy with Raptor coding tends to mismatch the time-varying network conditions: too little redundancy can cause decoding failures, while too much redundancy introduces unnecessary bandwidth and delay overhead. ARRCT therefore uses feedback on decoding outcome and delivery statistics to adapt redundancy and balance reliability against overhead.o main problems. First, in random-routing networks (e.g., KRRD), path unpredictability can lead to high loss and long-tail latency / prolonged session completion time. |
ICACT20260355 Question.8
Questioner: hejf@dsp.ac.cn 2026-02-05 오후 9:13:12 |
ICACT20260355 Answer.8
Answer by Auhor xueyy@dsp.ac.cn 2026-02-05 오후 9:13:12
Chrome Click!! |
| Why is Raptor coding more suitable than an ACK/retransmission mechanism in a random-routing network? |
In a random-routing network, end-to-end delivery is highly unpredictable, so per-packet ACK/retransmission is hard to track reliably and adds extra RTT and control overhead. In contrast, with Raptor coding the receiver can decode once it collects enough coded packets within a window，and the sender can adapt redundancy 𝑟 using aggregate feedback such as the observed success ratio 𝑠𝑡 and decoding failure indicator failt，without requiring packet-level acknowledgments. |
ICACT20260355 Question.9
Questioner: zhaoyb@dsp.ac.cn 2026-02-05 오후 9:14:23 |
ICACT20260355 Answer.9
Answer by Auhor xueyy@dsp.ac.cn 2026-02-05 오후 9:14:23
Chrome Click!! |
| Which performance metric does ARRCT mainly improve? |
ARRCT mainly improves end-to-end session completion time / session completion latency, and the paper quantifies this benefit using the speedup metric (fixed-redundancy session time divided by adaptive-redundancy session time). |
