Adaptive IBLT Hierarchy for Burst Packet Loss and Latency Balance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing multicast networks face challenges in reliably transmitting data due to packet loss, especially with spikes or bursts, which can hinder downstream nodes from receiving necessary data and exceed bandwidth and latency limits.
Innovation Solution
Implementing an invertible bloom lookup table (IBLT) hierarchy within the network to manage packet loss by maintaining IBLTs at nodes, transmitting them based on network properties, and enabling peer-to-peer recovery of lost packets through IBLT operations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If IBLTs are transmitted frequently to downstream nodes to recover lost packets, then packet recovery reliability is improved, but bandwidth consumption increases
Solution Approach 1:
The system dynamically adjusts the IBLT transmission frequency based on observed packet loss characteristics. When packet loss is low, IBLT transmission frequency is reduced to save bandwidth. When packet loss increases or spikes are detected, the frequency is increased to improve recovery reliability. This dynamic adaptation resolves the contradiction by making the system responsive to actual network conditions rather than using a fixed transmission rate.
Solution Approach 2:
The system changes key parameters including IBLT transmission frequency, hierarchy depth, and table size based on network conditions. By adjusting these parameters dynamically, the system optimizes the balance between packet recovery reliability and bandwidth consumption, transmitting more IBLTs when needed and fewer when the network is stable.
2Loss of time
If IBLTs are transmitted with high frequency to reduce recovery latency, then packet recovery speed is improved, but bandwidth usage increases
Solution Approach 1:
The system dynamically adjusts IBLT transmission frequency based on real-time packet loss monitoring. During periods of stable transmission, frequency is lowered to conserve bandwidth. When packet loss spikes are detected or latency increases, the frequency is automatically increased to reduce recovery time, thus dynamically balancing latency and bandwidth usage.
Solution Approach 2:
The system transmits IBLTs in advance before packet loss is fully detected, maintaining a pipeline of recovery information at downstream nodes. This preliminary action reduces the effective recovery latency because recovery information is already available when loss is detected, rather than waiting for confirmation of loss before transmitting IBLTs.
3Reliability
If a deep IBLT hierarchy is implemented to increase packet recovery capacity, then recovery capability is improved, but memory requirements increase
Solution Approach 1:
The system segments the IBLT recovery function into a hierarchical structure with multiple layers. Each layer in the hierarchy maintains IBLTs for a specific subset of packets or time period. This segmentation allows the total recovery capacity to be distributed across multiple smaller memory structures rather than requiring one large IBLT, thus maintaining high recovery capacity while managing memory requirements through organized distribution.
Solution Approach 2:
The system implements a nested IBLT hierarchy where smaller IBLT structures are contained within larger hierarchical frameworks. Each level of the hierarchy provides recovery for specific packet subsets, with outer levels providing broader coverage and inner levels providing more granular recovery. This nesting allows efficient memory utilization by storing only the necessary IBLT data at each hierarchical level rather than duplicating full recovery capacity throughout.
4Reliability
If IBLT transmission frequency is increased to handle packet loss spikes, then recovery capability during spikes is improved, but average bandwidth usage increases
Solution Approach 1:
The system dynamically adjusts IBLT transmission frequency based on real-time detection of packet loss spikes. During normal operation with low packet loss, transmission frequency is kept low to minimize average bandwidth usage. When spikes are detected through monitoring mechanisms, the frequency is rapidly increased to provide adequate recovery capability during the spike period, then reduced again after the spike passes, thus adapting to transient conditions without permanently increasing average bandwidth consumption.
Solution Approach 2:
The system employs periodic monitoring of packet loss patterns and adjusts IBLT transmission frequency in periodic cycles. During periods between detected spikes, transmission occurs at a baseline low frequency. When spikes are detected, the system enters a high-frequency transmission mode for the duration of the spike event, then returns to baseline. This periodic adaptation allows the system to handle spikes effectively while maintaining low average bandwidth usage during stable periods.
Data Source
Figure 1
Figure 2
Figure 3~5
AI summary
The technology generally relates to systems and methods for transmitting data within a network. In accordance with aspects of the disclosure, nodes of the network may be configured to maintain invertible bloom lookup tables (IBLTs) based on received data packets and receive IBLTs from other nodes based on transmitted data packets. The IBLTs may be transmitted to downstream nodes in accordance with a frequency that allows for sufficient capacity of packet loss recovery, while also balancing bandwidth and latency requirements of the network.