Sensor-based measurements are intrinsically prone to errors. One can distinguish two types of errors: systematic and random noise errors. Calibration is the process of validating and/or adjusting the accuracy of a measuring instrument so that the systematic error is calculated and the bias is corrected. Distributed localized in-field calibration of sensor devices in an ad-hoc wireless network is of crucial importance since manual calibration is not feasible and cost-effective. In order to address calibration for an arbitrary system of sensors and actuators and an arbitrary model of systematic errors, we formulate calibration as nonlinear function minimization problem. We start by identifying the necessary and sufficient conditions to solve a specific instance of calibration in a given network. We solve the minimization problem using combination of polak-ribiere conjugate gradient method and singular value decomposition (SVD) algorithms in conjunction with binary search procedure. In addition, we have developed four-phase localized calibration procedure that minimizes the amount of required multi-hop limited-size packet communication and storage...