Innovative Algorithms to Improve Long Range RTK Reliability and Availability

Resolving the GPS carrier-phase integer ambiguities reliably and quickly has been a cornerstone of state-ofthe-art high precision GPS positioning techniques. The challenge of doing so in real time and over long baselines is much greater due to errors introduced by the spatial decorrelation of differential atmospheric biases and orbital errors. This paper presents innovative algorithms, which have been developed to deal with these error sources and thereby significantly improve long range Real-Time Kinematic (RTK) fix reliability and availability. The overall architecture of the new ambiguity resolution algorithm is based upon a modified LAMBDA method. The unique and innovative techniques used include; 1) an extended functional model to estimate the atmospheric biases and integration of real-time orbit corrections from WADGPS systems, 2) a modified Kalman filter with optimal detection and estimation of cycle slips and code outliers, and 3) partial search and fix strategies embedded in the Kalman filter parameter estimation. The partial search/fix techniques are used to assist in validating the integer ambiguities by excluding selected integer ambiguity values to help discriminate between the best and second best candidates. It is shown how these techniques significantly reduces the time to first RTK fix without sacrificing reliability, especially for long-range baselines and challenging signal environments. Development of the algorithms has been completed including extensive offline testing with recorded data and hosting of the real-time version in the target GPS receiver. Both offline and real-time results are presented using extensive recorded data collected over baselines ranging up to 75 km in length. The reliability of integer ambiguity is shown to be from 99.6% to 99.99% probability of being correct. Ambiguity search times are typically one or two 1Hz epochs for baselines less than 20 km.