Otiy: Loactors tracking nodes

TL;DR

Otiy is a node-centric location service for wireless mesh networks that predicts node locations based on mobility patterns, reducing location update costs and delays by using a cyclic agenda of predicted locations.

Contribution

Otiy introduces a novel mobility-aware location service that leverages cyclic mobility patterns to improve location prediction accuracy and efficiency in wireless mesh networks.

Findings

Approximately 75% of nodes have at least 50% accurate location predictions over a year.

Location prediction accuracy increases by 10% when nodes consider their proximity to locators.

The approach reduces location delays and bandwidth consumption compared to traditional identifier-based services.

Abstract

We propose Otiy, a node-centric location service that limits the impact of location updates generate by mobile nodes in IEEE802.11-based wireless mesh networks. Existing location services use node identifiers to determine the locator (aka anchor) that is responsible for keeping track of a node's location. Such a strategy can be inefficient because: (i) identifiers give no clue on the node's mobility and (ii) locators can be far from the source/destination shortest path, which increases both location delays and bandwidth consumption. To solve these issues, Otiy introduces a new strategy that identifies nodes to play the role of locators based on the likelihood of a destination to be close to these nodes- i.e., locators are identified depending on the mobility pattern of nodes. Otiy relies on the cyclic mobility patterns of nodes and creates a slotted agenda composed of a set of predicted locations, defined according to the past and present patterns of mobility. Correspondent nodes fetch this agenda only once and use it as a reference for identifying which locators are responsible for the node at different points in time. Over a period of about one year, the weekly proportion of nodes having at least 50% of exact location predictions is in average about 75%. This proportion increases by 10% when nodes also consider their closeness to the locator from only what they know about the network.