Analysis of quantities influencing the performance of time synchronization based on linear regression in low cost WSNs

Time synchronization is a key feature of Wireless Sensor Networks (WSNs). In low cost WSNs this is a very challenging issue since the general employment of a large number of nodes, characterized by low power and limited energy resources. Therefore, in such contexts, the adoption of synchronization algorithms is preferable to the adoption of GPS-based solution because of its high energy consumption, high cost and poor performance in indoor environment. An even increasing number of papers dealing with the design and the implementation of protocol-based techniques for synchronization can be found in literature.These synchronization algorithms can be divided into three main categories: one way messaging, two ways messaging and consensus based synchronization. Among the others, those based on the one way messaging and, in particular, on the adoption of regressive algorithms are widely used in many application contexts. Focusing the attention on this class of synchronization algorithms, this paper proposes a deep performance analysis aimed at highlighting the sensitivity of the regression-based algorithms to some factors that influence the accuracy in typical low cost applications, such as the finite resolution of the local timing clock, the presence of clock drift, clock high frequency noise and low frequency noise, the presence of latencies due to the radio devices, the presence of latencies due to the microcontroller device. The main goal is to evaluate the effect of each one of these factors of influence on the overall synchronization performance. To these aims, suitable analyses both in simulation environment and on real nodes have been carried out.