Recursive Least Squares for Real-Time Implementation [Lecture Notes]

Recursive least squares (RLS) is a technique used for minimizing a quadratic cost function, where the minimizer is updated at each step as new data become available. RLS is more computationally efficient than batch least squares, and it is extensively used for system identification and adaptive control. This article derives RLS and emphasizes its real-time implementation in terms of the availability of the data as well as the time needed for the computation.