Decision Tree Link Selection for Multi-PHY Mesh Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Multi-PHY mesh network nodes experience increased latency due to retransmission attempts over primary wireless links, and the secondary powerline medium's throughput is often underutilized as nodes prefer the primary medium, leading to suboptimal bandwidth utilization.
Innovation Solution
Implementing decision trees that use transmission metrics to dynamically select between radio frequency (RF) and powerline communication (PLC) links on a per-packet basis, leveraging algorithms like ID3, C4.5, random decision forests, and gradient boosted decision trees to maximize total bandwidth and minimize collisions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If mesh network nodes perform multiple retransmission attempts over the primary wireless medium, then data reliability is improved, but latency increases due to retransmission time and backoff penalty
Solution Approach 1:
The patent implements dynamic medium selection by switching between RF and PLC links based on real-time link quality metrics. The system dynamically evaluates transmission success rates, packet loss, and latency for each medium and adapts the selected path accordingly, allowing nodes to transition between retransmission strategies and medium switching to optimize the balance between reliability and latency.
Solution Approach 2:
The system changes the operational parameters by switching communication mediums based on measured performance metrics. When RF link quality deteriorates (high packet loss, low signal strength), the system transitions to PLC medium, and vice versa. This parameter change approach allows the system to maintain reliability while minimizing latency by selecting the optimal medium for each transmission condition.
2Reliability
If mesh network nodes are configured to communicate only over the secondary medium when the primary medium fails, then reliability is maintained, but throughput is underutilized
Solution Approach 1:
The patent implements dynamic load balancing across multiple mediums by continuously monitoring link quality metrics and switching between RF and PLC links based on real-time conditions. This dynamic approach allows the system to utilize both mediums for simultaneous transmissions, maximizing throughput while maintaining reliability through selective medium switching rather than exclusive reliance on a single medium.
Solution Approach 2:
The system segments the communication load by dividing packets into different streams that can be transmitted over different mediums simultaneously. By segmenting the data flow and assigning different packets to RF or PLC channels based on link quality, the system achieves parallel transmission and maximizes overall throughput while maintaining reliability through distributed communication paths.
3Productivity
If mesh network nodes use decision trees for medium selection, then throughput load balancing is optimized, but device complexity increases
Solution Approach 1:
The patent implements self-service by having each mesh network node autonomously collect link quality metrics, evaluate transmission performance, and make independent medium selection decisions using embedded decision trees. Each node maintains its own set of transmission metrics and decision logic, eliminating the need for centralized control and reducing overall system complexity while achieving optimized throughput load balancing through distributed intelligence.
Solution Approach 2:
The system uses feedback mechanisms where nodes continuously monitor transmission success rates, packet loss, and latency for both RF and PLC links. These feedback metrics are fed into decision trees that automatically adjust medium selection strategies. The feedback loop enables the system to learn from past transmissions and make increasingly optimal decisions, achieving throughput optimization through adaptive, data-driven medium selection.
Data Source
AI summary
In one embodiment, a technique for load balancing of throughput for multi-PHY networks using decision trees is provided. A first device of a mesh communication network may collect at least one transmission metric indicative of a primary link and a secondary link between the first device and a second device of the mesh communication network. The first device may provide the at least one transmission metric as input to one or more decision trees comprising one or more attributes that are each indicative of a threshold for a corresponding transmission metric. The first device may obtain an output from the decision tree comprising a selection of either the primary link or the secondary link. The first device may send, based on the output from the decision tree, one or more packets to the second device using the selected link.