Self-Organized Edge Task Orchestration for Dynamic Ad Hoc Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing orchestration systems like Kubernetes are inefficient in highly dynamic mobile ad-hoc networks due to reliance on a single central controller, leading to challenges in resource management and reliability in environments with rapidly changing network topology and node availability.
Innovation Solution
A decentralized, self-organized network node design that distributes computational tasks across multiple nodes without a central manager, using a request-response mechanism and event-driven architecture to ensure robust and reliable processing even in volatile environments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a central controller is used for task orchestration, then resource management is simplified, but reliability deteriorates in highly dynamic networks with node volatility and connectivity issues
Solution Approach 1:
The patent extracts the central controller from the orchestration system, removing the single point of failure. Each node now independently performs orchestration functions that were previously centralized, distributing control across the network to maintain reliability while nodes dynamically join and leave.
Solution Approach 2:
The patent segments the centralized orchestration functionality into distributed components at each node. Each node maintains local task state and makes independent decisions, dividing the monolithic control system into autonomous units that can operate independently when network conditions change.
2Productivity
If a centralized orchestration system is used, then task coordination is efficient, but adaptability to network topology changes deteriorates
Solution Approach 1:
The patent implements dynamic adaptation where each node continuously monitors network conditions and adjusts task execution accordingly. Nodes can dynamically change their role from executor to coordinator based on current network state, allowing the system to adapt to topology changes while maintaining efficient task coordination through local decision-making.
Solution Approach 2:
Each node autonomously manages its own task state and makes coordination decisions without requiring central controller intervention. This self-service capability allows nodes to independently adapt to network changes, selecting appropriate tasks from available work packages based on local resource availability and network conditions.
3Adaptability or versatility
If nodes frequently join and leave the network, then network flexibility is improved, but data loss increases due to volatile connections
Solution Approach 1:
The patent implements comprehensive feedback mechanisms where nodes continuously exchange status information about task progress and data receipt. When a node leaves or reconnects, the feedback loop enables other nodes to detect the change and retransmit any lost data packets, ensuring complete task execution state is preserved despite network volatility.
Solution Approach 2:
The system prepares for potential node failures by implementing redundant data transmission and state synchronization before nodes leave. Task state information is propagated to multiple nodes in advance, creating a cushion against data loss when nodes dynamically join or leave the network.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Embodiment provide a network node for an edge computing network, wherein the network node is configured to obtain input data to be processed by a computational task, wherein the computational task comprises a plurality of computational steps, wherein the network node is configured to distribute a processing of at least one computational step of the plurality of computational steps over at least one other network node of the edge computing network, in order to distribute the computational task over different nodes of the edge computing network.