System design choices in smartautonomous networked irrigation systems

Wireless Sensor Networks are often deployed in great numbers spanning large,sometimes hard to reach and hostile, areas with the aim of monitoring environmentalconditions through the use of different sensors. Due to decreasing costs of ownership(e.g. non-proprietary protocols), recent advances in processor, radio, and memorytechnologies and the engineering of increasingly smaller sensing devices, theavailability and area of application for wireless sensor networks have steadily beenincreasing.Sigma Technology Development Stockholm AB raised the question as to whether awireless sensor network, running an open-source operating system and communicatingover IPv6, could be used in the field of smart autonomous irrigation? The company alsorequired a proof-of-concept system for demonstration purposes and to identify if thedesign choices made were suitable for an actual implementation.There are numerous of design decisions that have to be made when constructing anirrigation system: the back-end set-up, which irrigation algorithms to use, what hardwareto choose and how to communicate? This thesis therefore focuses on the overallsystem design of a wireless sensor network in the field of irrigation and highlights thetrade-offs being made and their pros and cons.Two improvements related to the existing technology and the proof-of-concept systemare presented in this thesis. Firstly, the recommendation to use clustered self-healingrouting despite claimed power consumption issues. Secondly, a new technique tominimize power consumption, by dynamically changing the sleep interval on the sensornodes with the help of weather data. Furthermore, the proof-of-concept system isconstructed and analysed to assess whether the system design choices made are validfor a real-life deployment.