Demo Abstract: Magneto-inductive tracking of underground animals

1 Overview and Motivation Existing sensor network deployments for wildlife tracking (e.g. ZebraNet [1]) have concentrated on monitoring animal behaviour above-ground. However, a wide variety of animals create underground tunnels for shelter and protection whilst the animal is asleep. The extent and internal architecture of the underground structure varies considerably amongst species. For example, badgers excavate wide ranging underground tunnel systems which a number of animals inhabit in a community [6]. In addition, their tunnel structure is something which is often only determined by the destructive extreme of excavation [6]. As fossorial animals spend a large proportion of their lifetimes underground, this means that zoologists only have a partial view of their behaviour and habits. There is thus a need for a system which can localize animals whilst they are underground, in a non-invasive and automatic way. Localization techniques based on using radio transmissions (whether RSSI, time-of-flight or angle of arrival) all suffer from severe attenuation of the electric field component of the electromagnetic wave by the soil. This also precludes the use of technologies such as GPS. As some animals can burrow up to a depth of 3 m, even high powered radio approaches are unsuitable. Localization through dead reckoning [2] or inertial tracking has major issues with drift which can result in large cumulative errors, reducing the confidence in drawing significant biological conclusions from location information. Although electric fields are strongly affected by the presence of soil and water, magnetic fields on the other hand are subject to negligible levels of attenuation. This makes them an ideal modality for tracking animals underground. 2 Magnetic Localization The use of magnetic fields for localization was first proposed by Raab et al.[5], but to date, the focus has been on high-precision operation over a small volume (e.g. 3 m radius). Moreover, existing approaches have not considered operation through soil. The devices used in these systems are not designed for long-term low power operation and typically have lifetimes in the range of a few hours, not the months required for an animal tracking system. Thus, cus-