Building context-aware services in pervasive computing environments have enabled the wide development of wireless local area network-based indoor positioning systems. In fingerprint localization, radio frequency (RF) signal strengths from access points (APs) are annotated with location labels to build the map of RF fingerprints. However, the newly received signal strength (RSS) variation due to device heterogeneity, which may cause RSS pattern mismatch, could jeopardize positioning accuracy. Solutions based on extra manual calibrations of RSSs for new, individual devices could address the problem. However, they are laborious and unpractical for real-world deployment. In this paper, an indoor positioning algorithm that utilizes two homogeneous features of different devices is proposed to solve the problem of device heterogeneity in fingerprint localization. The features of RSS order and linear dependency between RSSs measured by different devices are extensively investigated. The experimental results show that the proposed positioning algorithm solves the device heterogeneity problem without requiring extra manual calibration for diverse devices.