In this manuscript, an innovative method for the detection and the
estimation of multiple targets in a radar system employing orthogonal
frequency division multiplexing is illustrated. The core of this method
is represented by a novel algorithm for the detection of multiple
superimposed two-dimensional complex tones in the presence of noise and
the for estimation of their parameters. This algorithm is based on a
maximum likelihood approach and combines a single tone estimator with a
serial cancellation procedure. Our numerical results lead to the
conclusion that the developed method can achieve a substantially better
accuracy-complexity trade-off than various related techniques in the
presence of closely spaced targets.