Algorithm 1 Pseudo-code of the MHT-IDP algorithm.

Input: the measurement data z(k). Output: the best tracks $T_t(k),t=1,2,\cdots ,N.$ 1: Set k=1. 2: for i=1→length (z(k)) 3:  calculate the adaptive threshold τi(k) with (14) 4:  if amplitude ai(k) > τi(k) 5:    calculate the adaptive detection probability $P_d^i(k)$ with (19) and calculate the clutter density and ${\stackrel{⏜}{\lambda }}_m^i(k)$ with (22). 6:    calculate the adaptive score function $\Delta L_m^i(k)$ with (26) and acquire the changing alarm number $N_{fa}^i(k)$ with (29). 7:  end if 8: end for 9: calculate the track score Lt(k), t = 1,2,...,N. with (24). 10: calculate the adaptive SPRT threshold with (31)and (32) 11: for t = 1→N 12:  if Lt(k) > T1(k), confirm the track, end if 13:  else if T2(k) ≤ Lt(k) ≤ T1(k), continue to test track, end if 14:  else if Lt(k) < T2(k), delete the track, end if 15: end for 16: cluster the tracks, form the global hypothesizes and N-best pruning the tracks.

17: Set k = k+1, return the predicted data $\widehat{\mathrm{z}}(k+1)$ to step 2.