ICACT20260302 Question.1
Questioner: yangyn@dsp.ac.cn 2026-02-04 오후 4:01:05 |
ICACT20260302 Answer.1
Answer by Auhor zhaoyb@dsp.ac.cn 2026-02-04 오후 4:01:05
Chrome  Click!! |
| How does the per-interface cache system in ISMCC dynamically transition between states (e.g., Increase, Multipath, Flush) based on the MI state machine, and what role does the parameter λ play in controlling out-of-order packets? |
Thank you for your question. The per-interface cache system is consisted of many sub-caches, and the states of sub-caches are ralated to the corresponding MI states. As the occupancy of the sub-cache belows m and the corresponding MI states is Cache, the sub-cache is in Incress mode. As the sub-caches reach m and the corresponding MI states is Multipath\Backpressure\DisableBackpressure mode, the sub-cache is in Multipath mode.Finally, when the MI dimiss from the Multipath mode, the sub-cache switches to Flush mode until it is empty.The setting of parameter λ is to prevent the oscillations of MI state, which might cause some out-of-order packets, but within the limitation of 2λ. |
ICACT20260302 Question.10
Questioner: shil@dsp.ac.cn 2026-02-11 오후 12:04:44 |
ICACT20260302 Answer.10
Answer by Auhor zhaoyb@dsp.ac.cn 2026-02-11 오후 12:04:44
Chrome Click!! |
| How does ISMCC decide when to switch between caching and multipath transmission based on the relationship between sub-cache occupancy (n) and the estimated out-of-order threshold (m)? |
Thank you for your question. ISMCC manages the transition between caching and multipath transmission by continuously evaluating the real-time relationship between two key variables: the current sub-cache occupancy (n) and the estimated out-of-order data threshold (m). The value of m represents the exact amount of buffered data required to offset the propagation delay difference between the primary path and the detour path. When congestion occurs, the network node evaluates m and the sub-cache occupancy n. When n << m, ISMCC prioritizes caching to mitigate congestion. When n approaches m, ISMCC initiates per-packet multipath transmission. When n >> m, ISMCC reaches its limit at the current node, and backpressure is triggered. |
ICACT20260302 Question.11
Questioner: zhaozx@dsp.ac.cn 2026-02-11 오후 12:04:10 |
ICACT20260302 Answer.11
Answer by Auhor zhaoyb@dsp.ac.cn 2026-02-11 오후 12:04:10
Chrome Click!! |
| Three major data transmission challenges are highlighted: burstiness, high loss cost, and traffic congestion. Can you explain why increased link capacities amplify traffic burstiness, and what specific impacts does this burstiness have on WAN performance (e.g., flow completion time, resource utilization)? |
Thank you for your question. When link capacities increase, the physical time it takes to serialize and transmit a fixed amount of data shrinks dramatically. For examples, if an AI application needs to send a 1 GB payload over a 1 Gbps link, it takes 8 seconds of sustained, relatively steady traffic. If that link is upgraded to 100 Gbps, that same 1 GB payload hits the network as a massive spike lasting just 0.08 seconds.Traffic burstiness creates a mismatch between average bandwidth and instantaneous demand,which causes congestion and packet loss. It can bring Poor Network Resource Utilization, Degradation of End-to-End Goodput and Volatile Flow Completion Times to WANs performance. |
ICACT20260302 Question.2
Questioner: yangyn@dsp.ac.cn 2026-02-09 오전 12:35:42 |
ICACT20260302 Answer.2
Answer by Auhor zhaoyb@dsp.ac.cn 2026-02-09 오전 12:35:42
Chrome Click!! |
| What are the most promising extensions for ISMCC mentioned in the conclusion, such as adaptation to multi-hop scenarios, and how might machine learning be integrated for better traffic prediction? |
Thank you for your question.In the conclusion of the paper, the most promising extension for ISMCC is enhancing its adaptability to complex WAN conditions, specifically multi-hop scenarios. The current evaluation focuses on simpler topologies to isolate system behavior. However, in a real-world multi-hop environment, coordinating backpressure signals across multiple nodes and ensuring stability without causing oscillation in the state machine will be critical challenges to address in future work.Regarding the integration of Machine Learning (ML),We think it could significantly enhance ISMCC in traffic prediction and dynamic threshold tuning. |
ICACT20260302 Question.3
Questioner: tanghr@dsp.ac.cn 2026-02-09 오전 12:41:58 |
ICACT20260302 Answer.3
Answer by Auhor zhaoyb@dsp.ac.cn 2026-02-09 오전 12:41:58
Chrome Click!! |
| How does the weight update algorithm in the cache system ensure fair traffic allocation? Readers might be confused by the logic of ideal weight versus practical allocation in Algorithm 1. |
Thank you for your question.The Weight Update Algorithm (Algorithm 1) ensures fairness by dynamically balancing "theoretical entitlement" with "actual usage capability". We calculate the ideal weight based on the number of active flows. Algorithm 1 then adjusts the weights to align the sending rates with these ideal weights, thereby achieving fairness. For specific implementation details, please refer to the seventh paragraph of Section II(B). |
ICACT20260302 Question.4
Questioner: xueyy@dsp.ac.cn 2026-02-09 오전 12:43:52 |
ICACT20260302 Answer.4
Answer by Auhor zhaoyb@dsp.ac.cn 2026-02-09 오전 12:43:52
Chrome Click!! |
| How does ISMCC reduce packet reordering at the receiver through caching and multipath coordination? |
Thank you for your question. ISMCC reduces packet reordering by using the cache as a "time buffer" to synchronize the arrival times of packets traveling over different paths. Instead of blindly sending packets over a slower secondary path (which would cause them to arrive late and out of order), ISMCC forces packets on the primary path to wait in a queue until their total delay matches the detour paths delay. |
ICACT20260302 Question.5
Questioner: xueyy@dsp.ac.cn 2026-02-09 오전 12:46:27 |
ICACT20260302 Answer.5
Answer by Auhor zhaoyb@dsp.ac.cn 2026-02-09 오전 12:46:27
Chrome Click!! |
| What is the role of the “backpressure signal” in ISMCC, and how is it propagated between network nodes? |
Thank you for your question. The Backpressure Signal acts as a "last resort" safety mechanism to prevent buffer overflows when local resources (cache and multipath capacity) are exhausted.The signal propagates hop-by-hop from the point of congestion back to the source.Once the congestion clears in the congested node, it sends a "Disable Backpressure" signal the its upstream node. |
ICACT20260302 Question.6
Questioner: xueyy@dsp.ac.cn 2026-02-09 오전 12:54:01 |
ICACT20260302 Answer.6
Answer by Auhor zhaoyb@dsp.ac.cn 2026-02-09 오전 12:54:01
Chrome Click!! |
| Why does ISMCC introduce a “multipath slow start” mechanism, and what problem does it address? |
Thank you for your question. The Multipath Slow Start mechanism is introduced to overcome a specific synchronization issue that gets the system stuck in the Increase states.The sending rate grows slowly in standard transport-layer behavior (Addititve Increase), and causes the target threshold to grow at the same time as the cache occupancy is growing.By multipath Slow Start,the congestion window size grows exponentially in Increase mode,activating transmission over the secondary path. |
ICACT20260302 Question.7
Questioner: wangsy@dsp.ac.cn 2026-02-09 오전 12:56:30 |
ICACT20260302 Answer.7
Answer by Auhor zhaoyb@dsp.ac.cn 2026-02-09 오전 12:56:30
Chrome Click!! |
| You mentioned that multiple paths and cache nodes can cause packet disorder in the same flow and reduce goodput. So, what is the core method of ISMMC to reduce packet disorder?
Make the packets of the same flow follow the same path as much as possible?
Or should we conduct per-packet delay prediction and scheduling to make the arrival sequence more consistent? |
Thank you for your question. ISMCC reduces packet disorder using a "Delay Equalization" strategy, rather than the two methods you guessed.ISMCC explicitly uses per-packet multipath transmission. It splits a single flow across multiple paths to maximize bandwidth utilization.It does NOT use complex per-packet prediction. Instead of calculating the arrival time for every single packet (which is computationally expensive), it uses the Cache Occupancy as a simple, mechanical proxy for delay. |
ICACT20260302 Question.8
Questioner: shil@dsp.ac.cn 2026-02-11 오전 11:47:08 |
ICACT20260302 Answer.8
Answer by Auhor zhaoyb@dsp.ac.cn 2026-02-11 오전 11:47:08
Chrome Click!! |
| How does ISMCCs per-interface cache system work to absorb traffic bursts and reduce out-of-order packet delivery at the receiver? |
Thank you for your question. ISMCC handles traffic spikes locally at the network node,instead of immediately dropping packets or waiting for end-host transport layers to react to congestion. To reduce out-of-order packet delivery, ISMCC ties the activation of multipath routing to the specific occupancy level of the cache, ensuring packets from different paths arrive simultaneously. |
ICACT20260302 Question.9
Questioner: shil@dsp.ac.cn 2026-02-11 오전 11:48:42 |
ICACT20260302 Answer.9
Answer by Auhor zhaoyb@dsp.ac.cn 2026-02-11 오전 11:48:42
Chrome Click!! |
| What role does the backpressure signal play in coordinating between different network nodes and the sender in ISMCC? |
Thank you for your question. In the ISMCC framework, the backpressure signal serves as a critical communication mechanism to propagate congestion states upstream, allowing intermediate network nodes and the original sender to adjust their transmission behaviors and prevent buffer overflow. It is primarily triggered when ISMCC reaches its operational limit at a single network node. |
