Empirical approach to network sizing for connectivity in wireless sensor networks with realistic radio propagation models

Choosing the appropriate network size to guarantee connectivity in a WSN deployment is a challenging and important question. Classic techniques to answer this question are not up to the challenge because they rarely consider realistic radio models. This work proposes a methodology to evaluate the performance of network size estimation techniques in terms of connectivity efficiency under realistic radio scenarios. This study is carried out using Atarraya, a simulation tool for wireless sensor networks, considering three classical estimation techniques and a radio model based on the specifications of the ZigBee radio from off-the-shelf WaspMote nodes from Libelium. The results show that the hexagon-based optimal grid technique provides the most efficient estimate, offering a high connectivity level with the lowest estimated number of nodes for a given proximity radius parameter, followed by the circle packing and the triangle-based grid distribution. In addition, the results show that packet error rates of 10% could still produce highly connected topologies.