In WSNs, localization is the process of determining the physical position of sensor nodes. Importance of localization arises from the need to tag the sensed data and its associate events with their location of occurrence. Location information of sensor nodes can be found by using GPS (Global Positioning System), but installing GPS in every node results in high cost and power consumption. Thus only some sensors in the network know their position using GPS. Localization algorithms utilize the position of anchors to estimate the location of other sensor nodes which does not know its position in the network.
In wireless sensor networks, an unknown node estimates the distance to the anchors, and then utilizes the estimated distance to compute its position by multilateration. The localization accuracy by multilateration depends on the distance estimation accuracy and the relative positions of anchors. Since the range-free scheme is based on the connectivity information such as hop count, the distance can always be estimated with error. Actually the unknown nodes can estimate its position accurately even with three reliable anchors. However when selecting the reliable anchors, each unknown node should also consider the relative positions. Hence we study the relationship between the localization accuracy by considering the distance estimation and relative positions of anchors. From the study we find that there are two conditions need to be satisfied to improve the localization accuracy. To increase the possibility of all unknown nodes to select the reliable anchors based on the two conditions, anchor node placement needs to be considered. Hence, we propose an optimal anchor placement algorithm and reliable anchor selection scheme. Simulation results show that the proposed methods improves the localization performance better than the existing methods.
