The use of wireless sensor networks is essential for the implementation of information and control technologies in precision agriculture. In this thesis, we address the challenges associated with the design of such a network system. We present our design of the network stack for a wireless sensor network used for a precision agriculture application where sensors periodically collect environmental data from spatially distributed locations in the farm-field. The physical (PHY) layer in our network allows multiple power modes in both receive and transmit operations for the purpose of achieving energy savings. We design our medium access control (MAC) layer which uses these multiple power modes to save energy during the wake-up synchronization phase. The network layer is designed to custom fit the needs of the application, namely, reliable collection of data and minimization of the energy consumption. The design of various protocol layers involves a cross-layer design strategy. We present analytical models and simulation studies to compare the energy consumption of our MAC protocol with that of the popular duty-cycle based S-MAC protocol and show that our protocol has better energy efficiency as well as latency in a periodic data collection application operating over a multi-hop network of sensor nodes.

We also study the problem of sensor node localization for a hybrid wireless sensor network, with nodes located both underground (sensor nodes) and above-ground (satellite nodes). We consider two types of ranging measurements (received signal strength and time of arrival) from unmodulated signals transmitted between neighboring sensor nodes and between satellite nodes and sensor nodes. The problems are formulated with the goal of parameter estimation of the joint distribution of the received signal strength and time of arrival of the received signals. First, we arrive at power fading models for various communication scenarios in our network to model the received signal strength in terms of the propagation distance and hence, the participating nodes' location coordinates. We account for the various signal degradation effects such as fading, reflection, transmission, and interference between two signals arriving along different paths. With the same goal, we derive statistical models for the measured time of arrival with the parameters governed by the sensor nodes' location coordinates. The probability distribution of the detected time of arrival of a signal is derived based on rigorous statistical analysis. Then, we formulate maximum likelihood optimization problems to estimate the nodes' location coordinates using the derived statistical models. The results are validated through the implementation of the proposed sensor localization approach in Python using the SciPy optimization package. We also present a sensitivity analysis of the estimates with respect to the soil complex permittivity and magnetic permeability.

The contributions of this work are threefold. We present the system design of a wireless sensor network for use in a large scale deployable periodic data collection application. Next, we develop a thorough performance evaluation of the energy efficiency, throughput and latency of the system and compare with a traditional duty cycle based approach. Finally, we formulate maximum likelihood estimation based frameworks involving received signal power as well as latency measurements to solve the problem of sensor node localization based on relatively cheaper received signal strength measurements and more accurate time of arrival measurements for nodes deployed in multiple physical media (air and soil), and accounting for multi-path effects, signal loss and delays, and Gaussian and Rician fading.