Sectorized Localization

Overview

Knowledge of location information has been recognized as a fundamental primitive for nodes in a sensor network. Several applications and routing algorithms assume availability of location information for their functioning. While an easy solution is to equip every node with a GPS, this may not be practically feasible today because of three primary reasons,

• GPS needs line of sight to atleast four satellites,
• Cost of GPS chips are significant comapared to cost of motes, and
• GPS consumes a huge amount of power compared to rest of components.

The goal of Ad-hoc localization is to make location information available to all nodes. It assumes that,

• a ``small'' (about 10%) fraction of sensor nodes have GPS or similar devices which notify them of their position; such nodes are called anchors,
• there is no control over the positions of the anchors, and
• sensors are equipped with cheap ranging mechanisms, so that they can range to each other.

Traditional ad-hoc localization has focused on range-only schemes, i.e. sensors can obtain only range information to their neighbors. Range-only scheme based on acoustic ranging have been deviced and implemented. One draw-back of such schemes is that they require high deployment densities. Expensive ranging technologies such as laser promise both accurate range and bearing information and would require much smaller dense deployments of nodes. However, mounting such technologies on sensors may be very impractical. This project is an effort to seek a middle ground, where nodes obtain imprecise bearing information (30-60$^\circ$ sectors) and try to use this information for localization. We term this as range-sector localization. Our preliminary simulations suggest that range-sector localization can achieve acceptable performance for much lower deployment densities than range-only schemes. Our focus is on developing such devices and the corresponding software.

People

• Krishna Chintalapudi
• Gaurav Sukhatme
• Ramesh Govindan