Predictive Time Slot Allocation for TSCH Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Deterministic networking in TSCH networks faces challenges in scalable and efficient time slot allocation, particularly in low power and lossy networks where dynamic traffic changes and resource constraints require adaptive scheduling to maintain low jitter and high packet delivery ratios.
Innovation Solution
A machine learning-based predictive time slot allocation system that uses time slot usage reports from network nodes to forecast demand changes, allowing for proactive adjustments in time slot assignments, reducing the need for centralized computations and minimizing control plane overhead.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If centralized time slot allocation is used in TSCH networks, then deterministic packet delivery is achieved, but control plane overhead and computational complexity increase
Solution Approach 1:
The patent applies preliminary action by having network nodes proactively report their time slot usage status before congestion occurs. Nodes monitor their buffer status and preemptively send usage reports to the centralized controller, allowing the controller to anticipate future congestion and adjust time slot allocations before packet loss occurs, thereby maintaining high delivery ratios while reducing reactive control overhead
Solution Approach 2:
The patent implements feedback mechanisms where network nodes continuously monitor and report their time slot usage, queue depth, and transmission status to the centralized controller. This feedback loop enables the controller to dynamically adjust time slot allocations based on actual network conditions, achieving deterministic delivery while optimizing control plane efficiency through targeted rather than continuous updates
2Adaptability or versatility
If frequent time slot allocation adjustments are made, then adaptability to traffic changes is improved, but control plane overhead increases
Solution Approach 1:
The patent applies partial action by having nodes report only when specific thresholds are exceeded (e.g., when queue depth reaches a certain level or utilization exceeds a threshold). This selective reporting approach provides sufficient traffic adaptation by triggering allocations only when genuinely needed, rather than continuously adjusting for minor fluctuations, thereby reducing control plane overhead while maintaining adaptability
Solution Approach 2:
The system uses preliminary action by configuring threshold-based triggers that anticipate when allocation changes will be needed. When nodes detect approaching thresholds in their buffer status or utilization, they proactively report, allowing the controller to make preemptive allocation adjustments before actual congestion occurs, balancing adaptability with reduced control overhead
3Measurement precision
If time slot usage reports are collected from all nodes, then scheduling accuracy is improved, but network overhead increases
Solution Approach 1:
The patent applies local quality by having different nodes report at different frequencies or with different levels of detail based on their local network conditions. Nodes experiencing high traffic or near capacity thresholds report more frequently and with greater detail, while nodes with stable low utilization report less frequently or not at all. This differentiated approach maintains scheduling accuracy for critical nodes while reducing overall network overhead from less critical nodes
Data Source
Figure 1
Figure 2
Figure 3
AI summary
In one embodiment, a device in a network receives one or more time slot usage reports regarding a use of time slots of a channel hopping schedule by nodes in the network. The device predicts a time slot demand change for a particular node based on the one or more time slot usage reports. The device identifies a time frame associated with the predicted time slot demand change. The device adjusts a time slot assignment for the particular node in the channel hopping schedule based on predicted demand change and the identified time frame associated with the predicted time slot demand change.