Link Prediction in Mobile Computing Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In mobile computing, predicting dynamic link patterns is challenging due to limitations in understanding link appearance and disappearance, leading to inefficient communication and high energy consumption, especially in scenarios like vehicular and opportunistic networks.
Innovation Solution
A method and system for link prediction in mobile computing networks that breaks the network into snapshots, models historical link data using kernel regression, and evaluates ranking indices like the Adar-Adar index and contact frequency to predict future link probabilities, handling sparse data with community detection and approximation algorithms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If traditional network protocols are used with stable routing metrics, then routing is simple to implement, but end-to-end communication becomes inefficient in mobile environments
Solution Approach 1:
The patent applies dynamics by transitioning from stable routing metrics to dynamic routing metrics that adapt to changing network conditions. The system continuously updates routing decisions based on real-time link quality assessments, allowing the routing protocol to respond to mobility-induced network changes while maintaining implementation feasibility through structured metric updates.
Solution Approach 2:
The patent implements feedback mechanisms by continuously monitoring link quality and using this information to adjust routing decisions. The system evaluates link metrics, compares them against thresholds, and dynamically modifies routing paths based on observed network conditions, creating a closed-loop control system that improves communication efficiency.
2Productivity
If link prediction is performed to optimize communication mechanisms, then communication efficiency improves, but understanding of link patterns is required which currently lacks
Solution Approach 1:
The patent applies preliminary action by performing link prediction in advance before actual data transmission is needed. The system proactively identifies potential future links and pre-establishes routing paths based on predicted link patterns, allowing the network to prepare for upcoming communication needs and reduce latency when links actually form.
Solution Approach 2:
The patent introduces an intermediary link prediction module that bridges the gap between observed network data and routing decisions. This intermediary component analyzes historical and current link patterns, applies prediction algorithms, and provides actionable insights to the routing protocol, making the complex task of link pattern understanding manageable and systematic.
3Ease of operation
If static link prediction methods from complex networks are applied, then implementation is simple with similarity-based algorithms, but they treat networks as static snapshots and cannot predict dynamic link appearance and disappearance
Solution Approach 1:
The patent transforms static link prediction methods into dynamic ones by incorporating time-varying metrics and continuous network state monitoring. The system uses historical link data, current network conditions, and temporal patterns to predict future links, enabling the simple similarity-based algorithms to adapt to dynamic mobile network environments while maintaining their implementation simplicity.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention provides a method and system for link prediction in a mobile computing network comprising: c) breaking the mobile computing network into a sequence of snapshots at different times, each snapshot representing a link pattern of the mobile computing network; b) based on the snapshots obtained in step a) providing historical link data, representing the history of connected nodes in the mobile computing network; c) based on the historical link data obtained in step b), modeling the probability of links in the next snapshot of the mobile computing network using kernel regression; d) ranking of the obtained probability of links according to step c) in a descending order to obtain a link pattern in the next snapshot ; and e) based on the results obtained in step c) and d) outputting of the link probability for each individual link and/or the predicted link pattern in the next snapshot.