Range-free and range-based localization of wireless sensor networks

Wireless sensor networks consist of many wireless sensor nodes that enable the collection of sensor data from the physical world. A key requirement to interpreting the data is to determine the locations of the sensor nodes. The localization techniques developed can be divided into two categories: range-free and range-based..;Range-free localization usually assumes isotropic networks where the hop count between two nodes is proportional to their distance. However, anisotropic networks are more realistic due to the presence of various anisotropic factors in practice, e.g. irregular radio propagation, low sensor density, anisotropic terrain condition, and obstacles which can detour the shortest path between two nodes. The previous anisotropy-tolerating solutions focused on only one anisotropic factor - the obstacles. We will propose a pattern-driven localization scheme to tolerate multiple anisotropic factors.;Range-based localization assumes that the inter-node distances can be accurately measured by special ranging hardware. There are two important issues: (1) the ranging noise which affects the localization accuracy ; (2) the collinearity of critical node sets which may produce unanticipated flip ambiguities and harm the localization robustness. However, the previous research does not fully address these two issues, especially for patch merging, a powerful tool to localize sparse sensor networks. We will present our inflexible body merging algorithm to address these two issues for both patch merging and multilateration. Our algorithm can also improve the percentage of localizable nodes by nearly two times in sparse networks as compared with state-of-the-art work.;Another critical issue for range-based localization is the existence of outliers in raw data (i.e. distance measurements and anchor positions) which strongly deviate from their true values. These outliers can severely degrade localization accuracy and need to be rejected. Previous studies have two inadequacies of (1) focusing on adding an outlier rejection ability to multilateration but neglecting patch merging; (2) rejecting only the outlier distances but neglecting outlier anchors which are more difficult to remove, because outlier anchors may collude by declaring positions in the same coordinate frame. We will present an algorithm to reject both outlier distances and colluding outlier anchors, in both dense networks and sparse networks.