Selected impact

Developing centimeter-level satellite positioning

A Bayesian satellite-positioning algorithm developed at MERL and implemented in C demonstrated less than 3 cm positioning error over 24 hours in experiments on customer data.

Challenge

Carrier-phase GNSS measurements support precise positioning, but the receiver must resolve an unknown number of complete signal cycles. These integer ambiguities can change when a cycle slip occurs. The estimator needs to handle those changes while tracking the receiver’s motion.

Approach and contribution

At Mitsubishi Electric Research Laboratories, I designed and analyzed a Bayesian satellite-positioning algorithm and implemented it in C. The implementation was evaluated in experiments on customer data.

Results

Experiments on customer data demonstrated positioning error below 3 cm over a 24-hour period.

Integer Ambiguity Resolution by Mixture Kalman Filter for Improved GNSS Precision describes a related method for integer ambiguity resolution and cycle-slip handling. The paper reports numerical tests and an experimental evaluation distinct from the customer-data result above.