Real-time estimation of wireless ground-to-air communication parameters

To enable effective communication-based mission planning, a Scaled Unscented Kalman Filter (SUKF) is implemented on an unmanned aircraft to estimate the communication performance of ground nodes in real time, based solely on signal strength measurements. Our communication model extends Rappaport's log-distance model to account for directional antenna patterns. The filter is tuned on simulated data for fast convergence, and then tested through 12 flight experiments where the aircraft communicates with several types of ground nodes over IEEE 802.11b channels. State estimates settle within 15 minutes of a 45-minute flight, with the final eistmate's RMS deviation from the observed signal achieving between 4 and 9 dB. This represents a 20 to 80% improvement over a tuned filter based on the traditional log-distance model that ignores directional effects.

I. Introduction

Unmanned aircraft are able to provide useful services to ground-based units through wireless ad hoc communication, enhancing the unit's operational capability. In any such multiagent system, much emphasis must be placed on planning and organizing communication among the agents [1]. In order to be effective in cooperating with other nodes, advanced planning for the aircraft must be based on accurate models of the communication environment. This issue arises in several applications, including dynamic task allocation [2] and information sharing in either dense or sparse sensor networks [3].