Calibrating Multi-Channel RSS Observations for Localization using Gaussian Process

This letter proposes to use a Gaussian process regression model to compensate for frequency dependent shadowing effects and multipath in received signal strength (RSS) observations. Parametric multi-channel RSS calibration models are introduced and characterized by the inter-channel bias and the scale factor terms. With the proposed calibration model, multi-channel RSS observations can be more effectively combined for localization over large space. Field tests with BLE4.2 wireless radio broadcasting at channels 0, 12, 39 have been conducted over a 9,600 m2 outdoor area. Test results with sufficient and insufficient multi-channel RSS observations both confirm improved positioning performance by using the proposed model.