Wireless Sensor Networks

Cow with sensing collarIn 2008, two papers were published documenting the challenges and progress made concerning Wireless Sensor Networks (WSNs) within the context of Condition Based Monitoring (CBM) of cattle. The system tested comprised of antenna collars attached to each cow in a herd which transmit to a base station (or multiple) for concatenation. In general, CBM detects aspects of the physiology or behaviour of an animal and reports abnormal conditions at an early stage which will allow the farmer to take an appropriate action. Considering frequency availability on a global scale, the 2.4 GHz ISM radio band was chosen as the carrier frequency.

Wireless Sensor Networks for Beef and Dairy Herd Management – This paper aims to identify some issues in the design and operations of wireless sensor networks for CBM of cattle.

WSN Modelling the Attenuation of Radio Signals by Bovines – This paper focuses on predicting the available signal power in any given location in a farm environment and using the information to allow base stations to be sited for minimum signal attenuation and hence maximum collar battery life.

The problems of the system and major findings of the two papers are summarised below.

WSN Base StationMost RF sensor technologies are aimed at mass markets that have characteristics that differ from those in agriculture. In agriculture, however, power is limited, devices will be constantly mobile, transmission may have to penetrate through heavy rain and, although the background radio noise may be less than in an urban environment, there may be hundreds if not thousands of nodes within a network. The weight of a device attached to an animal is also constrained. The weight limit for a neck based device for use on cattle should not exceed 1 kg due to pressure and irritation concerns. This severely limits the battery power available and robust environmental power harvesting technologies are not yet available. Constraints imposed both by the engineering limitations (power, weight, radio frequency) and also by the animal behaviour and operations have been investigated.

The CBM technology has been based on a collar mounted wireless and sensor platform. The collar can act as a hub for other sensors elsewhere on the animal and when the collar is ready to transmit data it will attempt to communicate with a base station. Failure to communicate to the base station will result in the collar sleeping for a pre-determined period of time. Once successful the collar will send data to the station then sleep for an extended period of time to conserve battery. Antenna diversity is also implemented to achieve multiple signal propagation paths.

The connectivity of CBM is assumed to be sporadic, where most of the collars are out of coverage range periodically. Connectivity is mainly influenced by the farm size, animal movement and environment. Along with these (usually) fixed restrains the collar can only successfully transmit a packet if the antenna is facing the base station, since the signal cannot penetrate the animal’s body and is positioned on one side of the animal’s neck.

The penetration depth of 2.4 GHz signals was determined to be less than 2.5 cm through the animal’s body and, although this will change with different frequencies, the cow’s neck is approximately 0.25 m across and as such it is not possible for any radio signal to reliably penetrate.

An antenna diversity scheme was examined, in which two antennae (placed at top left and top right of the collar) can be used to optimise the collar radio coverage. The locations of the antennae are carefully considered so that signal propagates efficiently outward with fewer impediments from the other cattle in the proximity. An experiment conducted with two antennae, one facing the base station and one facing away, showed significant increase in signal strength from antenna facing base station, therefore the transmission path is strongly influenced by the orientation of the animal.

It was concluded that the operational time allocated to the communications portion of collar should be around 20 seconds per day to maintain an operational battery lifetime of five years at 20-30 mA current drain during transmission. These calculations were made assuming the communication platform uses 10% of the power budget of the collar. Due to the very small data load per cow per day, 20 seconds was predicted to be enough time to complete transmission. Time intervals between transmission attempts can be set to 20 mins while still maintaining the 5 year battery life, with the possibility of reducing time windows to allow for more reliable transmissions, although battery life will suffer.

Since space is not critical at the base station, a high gain (6dBi) omni-directional antenna can be employed. To determine when and where was best to place the base station in order to predictably collect data from animals a study was conducted on forty beef cows. Although distance from watering troughs varied substantially throughout the day and between days depending on factors such as weather conditions, it was concluded that, given a base station range of around 30 m, the cows will be within range for a sufficient period to enable data transmission.

WSN CAD CowIn order to understand how and where to implement the antennae on the collar and the base station, a predictive, simulation based approach was used to model the system and determine how best to approach the issues associated with radio transmission. These processes and findings can be followed in detail in WSN Modelling the Attenuation of Radio Signals by Bovines.

By modifying a freely available 3-D computer aided design (CAD) model of a cow, oriented in different poses, it was possible to conclude that the best signal was transmitted above the cow. Further modelling involving an example barn and an example pasture both complete with 20 model cows. Simulations of base stations situated at heights of 1.5 m, 3 m and 5 m showed increased signal strength with increased height. This was assumed to be in part due to shadowing caused by fences and other cows blocking signals, although raising the height did not improve all situations as cows with antennae placed close to the ground, for example feeding or resting, consistently gave poor signals. In those cases, several receiving antennae distributed throughout the area should increase the chance of good line of sight.

Overall the system shows great potential and since we used a standard commercially available patch antenna for the collar, and do not assume a specific type of antenna for the base station, these results are widely applicable. Cows were within range of base stations even in grazing situations for several minutes per day and thus long enough to download data daily although robust protocols still need to be developed for non-milking animals. Further work is required to enable protocols to be optimised for cattle monitoring and to allow interoperability.