A GLR algorithm for multiple consecutive measurement bias estimation

This paper proposes an algorithm for handling multiple consecutive measurement biases that appear and disappear frequently, typically encountered in GNSS in small low flying UAVs due to range errors and signal blocking. The method is derived as an extension to the well known GLR algorithm and is based on a corrected innovation sequence for detection and an identification stage based on least square estimation. A recursive (RLS) and a non-recursive (LS) solution is proposed in the identification stage. Results in a GNSS position error example show that the proposed algorithms are significantly better than the original algorithm in terms of estimation precision when biases appear and disappear frequently.