An existing metric algorithm in wireless mesh networks (WMNs) adopts samples generated by itself. In other words, the existing algorithm sends some probe frames and then counts the returned acknowledgments (acks). This leads to two inherent defects: wasting limited bandwidth to send too many probe frames; and detecting metric changes insensitively.

Presented herein is an improved routing metrics algorithm for WMNs that uses page rank (PR) theory and linear regression. To solve the problems mentioned above, PR theory and multiple linear regression (MLR) methods are introduced in metric calculation to generate metrics closer to real values by conditionally referring sibling metrics to a common parent or candidate.

The method may be run in a distributed fashion and/or it may be centralized. The dependency between the A->D metric and the B->D and C->D, or even B->C, may be observed by a learning machine. In that case the controller may push the parameters for the PR algorithm to the device so as to use only the metrics that are cross influential.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.