Real-Time Receiver Clock Jump Detection for Code Absolute Positioning with Kalman Filter

In global navigation satellite systems (GNSS) navigation the receiver and satellite clocks play a key role. The receivers are usually equipped with inaccurate quartz clocks, which experiment large drift relative to system time and consequently offset growing very fast; receiver manufactures bound the magnitude of the receiver clock offset to prevent it becomes too large and the actual bounding procedures vary from one manufacturer to another. The most common approach consists of introducing discrete jumps when the offset exceeds a threshold (usually 1 ms). This method is common in low-cost GNSS receivers and influences several applications as differential positioning, cycle-slip detection, precise point positioning technique, absolute positioning with Kalman filter. In this work some techniques to detect and account for millisecond clock jump, suitable for code positioning of a single receiver with Kalman filter, are proposed. Two deterministic algorithms to detect receiver clock jumps are shown: in measurement and parameter domain. The technique in measurement domain uses current pseudorange measurements compared with pseudorange and Doppler measurements at previous epoch; the technique in parameter domain compares current and previous least squares estimations of receiver clock bias, considering the clock drift. Two different approaches are described to account for the clock jumps, once detected, a deterministic one, consisting of fixing the pseudorange discontinuities, and a statistic one, consisting of suitably varying the Kalman filter settings. A static GNSS data set is processed with and without the proposed algorithms to demonstrate their efficiency.