Augmented Multi-Link Neighbor Lists for Wireless Roaming
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing wireless networking technologies face challenges in optimizing multi-link network environments, particularly in generating accurate and efficient neighbor lists for devices with multiple connections, which affect performance metrics such as Received Signal Strength Indicator (RSSI), Quality of Experience (QoE), and compatibility with legacy devices.
Innovation Solution
The implementation of augmented neighbor lists generated through machine learning processes that evaluate multiple links based on performance metrics like RSSI, time to last mile performance, and QoS, and format the lists for compatibility with legacy devices, including grouping BSSIDs to ensure seamless connectivity and improved QoE.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional neighbor lists are used in multi-link network environments, then device compatibility is maintained, but network performance and throughput are suboptimal
Solution Approach 1:
The neighbor list is segmented into multiple independent link-specific neighbor lists, each optimized for specific link characteristics. This allows the system to provide targeted recommendations for each link while maintaining overall compatibility with legacy devices that may only utilize a subset of these link-specific lists.
Solution Approach 2:
The patent introduces a new dimension to neighbor lists by creating link-specific neighbor lists that operate parallel to the traditional single neighbor list. This multi-dimensional approach enables simultaneous support for both legacy devices (using traditional lists) and modern multi-link devices (using link-specific lists), resolving the compatibility-performance contradiction.
2Reliability
If link-specific neighbor lists are generated for each link, then network optimization is improved, but device complexity and processing requirements increase
Solution Approach 1:
Link-specific neighbor lists are generated in advance based on historical performance data and link characteristics, rather than being computed in real-time during connection establishment. This preliminary generation reduces processing complexity during actual device connections while maintaining high reliability through pre-optimized recommendations.
Solution Approach 2:
The system creates simplified copies of the traditional neighbor list structure for each link, maintaining the familiar format and processing requirements of legacy lists while adding link-specific optimizations. This copying approach allows modern devices to utilize multiple link-specific lists without requiring fundamentally new processing mechanisms.
3Measurement precision
If machine learning processes are used to evaluate multiple links, then recommendation accuracy is improved, but computational overhead and processing time increase
Solution Approach 1:
Machine learning models are trained offline using historical network data to establish performance patterns and link characteristics. During runtime, the pre-trained models perform rapid inference to generate link-specific neighbor lists, significantly reducing the computational time required compared to real-time training while maintaining high recommendation accuracy.
Solution Approach 2:
The system applies machine learning selectively to evaluate only the most promising candidate links based on initial screening criteria, rather than exhaustively analyzing all possible links. This partial application of ML processes achieves sufficient recommendation accuracy while minimizing computational overhead and generation time.
Data Source
AI summary
Various embodiments described herein are configured to provide an augmented neighbor list during wireless network roaming. The augmented neighbor list can be formatted such that the neighbor recommendations provided to a mobile computing device during roaming include not just a single neighbor recommendation, but also a neighbor recommendation for each link associated with the mobile computing device. In this way, multi-link operations can become more efficient such that neighbor lists are augmented to add the per-link recommendations. Furthermore, one or more machine learning processes can be utilized to select the best per-link neighbor recommendation based on a series of inputs configured by the network administrator. The augmented neighbor list may also be formatted such that legacy devices may still be compatible by formatting the augmented neighbor list such that only one recommendation occurs. Sometimes, the recommendations are configured per mobile computing device link instead of the available network device links.