SENSOR LOCALIZATION USING ACOUSTIC DOPPLER SHIFT WITH A MOBILE ACCESS POINT Richard J. Kozick Bucknell University Department of Electrical Engineering Lewisburg, Pennsylvania 17837 Brian M. Sadler Army Research Laboratory 2800 Powder Mill Road Adelphi, Maryland 20783 ABSTRACT The nodes in a sensor network are often deployed in random locations. However, most applications require that the location and/or orientation of the nodes be known, so post-deployment algorithms for self-localization of the sen- sor nodes are important. We consider using a mobile ac- cess point (AP) for sensor node localization in a randomly deployed sensor network. We focus on the particular case in which the sensors measure the acoustic Doppler shift in a tone that is emitted from the mobile AP. In addition to the acoustic emission, the mobile AP broadcasts a radio signal that contains AP position, velocity, timing, and pa- rameters of the acoustic signal. We demonstrate that at- mospheric turbulence has a significant impact on the ac- curacy of sensor localization, degrading performance by as much as two orders of magnitude relative to an ideal, plane- wave propagation model. We present Cramer-Rao bounds (CRBs) for sensor localization accuracy and compare the performance of algorithms to the CRBs under “cloudy” and “sunny” weather conditions. Keywords: Sensor networks, sensor localization, location uncertainty, Cram´ er-Rao bound, atmospheric turbulence, at- mospheric scattering. 1. INTRODUCTION The nodes in a sensor network are often deployed in ran- dom locations, but most applications require that the lo- cation and/or orientation of the nodes be known. One ap- proach for post-deployment localization of the sensor nodes uses beacons that are either external to the network (such as GPS) or deployed within the network. In this paper, we consider using a mobile access point (AP) for sensor node localization in a randomly deployed sensor network. Deploying fixed beacons within the network enables a solution based on message passing between the nodes, e.g., see Moses et al. [1]. These beacons may be radio or another modality, such as acoustic, that takes advantage of the sen- sor capability. This approach assumes the communications network is pre-established, and requires sufficient density of beacons to be deployed, which raises the complexity of at least some of the nodes. Cevher and McClellan [2] proposed an approach to sensor localization that exploits an acoustic source that is external to the network and moves along a straight line path. The source trajectory is not known to the sensor nodes, so the source is not necessarily cooperating with the nodes. Each node tracks the angle of arrival (AOA) of the source and transmits the AOA tracks to a central in- formation processor, then the central processor determines the node locations and orientations. This approach also as- sumes that the communications network is pre-established. The use of a mobile AP for both communications and beaconing provides several advantages. The mobile AP can be used to localize many sensor nodes simultaneously in a broadcast mode, without requiring communications or syn- chronization between the nodes. The localization algorithms require only that the nodes receive the AP broadcast, and the AP can broadcast from several different beaconing positions as it moves. As a particular example, consider a network of acoustic sensors (microphones), where each node contains a microphone, a signal processor, and a narrowband radio. A helicopter that flies over the sensor network can serve as the mobile AP, and we assume that the AP has an accurate estimate of its own location and motion. The AP simulta- neously broadcasts a radio message and an acoustic signal. The radio message contains timing, AP location and motion, and information about the acoustic signal. The sensor nodes receive this multi-modal transmission (radio and acoustic) and process it to self-localize. Acoustic experiments with a helicopter as the mobile AP are reported in [3]. At least three different quantities may be estimated at the sensor node for self-localization: Doppler shift, time delay (TDE), and angle of arrival (AOA). Although a sin- gle transmission modality would suffice, the multi-modal transmission has several advantages. If the acoustic broad- cast is used alone, then the sensor nodes must communicate with each other since they do not know the AP timing, loca- tion / heading, and acoustic signal parameters. If the radio