To gain benefit from netted sensors, it is essential to correctly integrate each sensor into a global frame of reference. The process is termed sensor registration. It involves minimizing the effects of the errors from the imprecisely known relative positions or orientations of the various sensors, the biased sensor measurements, the offset sensor clocks, and so on. In the paper, a model of a typical sensor registration problem in a 2-sensor scenario is presented for illustration. Four kinds of registration errors, i.e., calibration errors (offsets), attitude (orientation) errors, position errors, and timing errors are mainly discussed in literatures, and literatures are categorized in the paper by its purpose that which one of the four kinds of errors, or which combination of the decoupled errors among them is aimed to solve. The registration errors can be viewed as constant and they can also vary or drift slowly. To address them, different kinds of algorithms have been developed. The algorithms for sensor registration can be divided into three types referred to as Type 1, Type 2, and Type 3 respectively. Algorithms of Type 1, mainly including batch processing methods and meta-heuristic optimization ones, are applied to cases that the registration errors are constant over time. Algorithms of Type 2 provide online estimation of sensor biases, but it is decoupled from the target estimation. The assumption for constant errors is relaxed, and the approaches are extended to cope with both constant and varying errors. The algorithms of Type 3 attempt to simultaneously solve for target variables and sensor systematic biases, and most of them achieve this via augmented state Kalman filter with the augmented state vector combining the target states and sensor uncertainty. References are accordingly organized by type, discussed, and annotated. Related issues of sensor registration are concerned including the non-unique detection probability, the asynchronous sensor measurements, the observability of registration problem, the relative registration and absolute registration, and the performance bounds of the estimator of the registration errors. The developed techniques for solving the problems are summarized respectively. The paper is concluded with a brief summary and some thoughts on the future trends and research directions. [ABSTRACT FROM AUTHOR]