Range-based localization in wireless networks using decision trees

Node localization is an essential component of many wireless networks. It can be used to improve routing and enhance security. Localization can be divided into range-free and range-based algorithms. Range-based algorithms use measurements to estimate the distance between nodes. Range-free algorithms are based on proximity sensing between nodes. Range-based algorithms are more accurate but also more complex. However, in applications such as target tracking, localization accuracy is important. In this paper, we propose a new range-based algorithm which is based on decision tree classification, a well known technique in data mining. This algorithm is compared with those based on linear least squares (LLS) and weighted linear least squares based on singular value decomposition (WLS-SVD). It is shown that the proposed algorithm performs better than these algorithms even when the anchor geometric distribution about an unlocalized node is poor.