The waveform agility in tracking originated from [
15], where the Kalman filter was used for single target tracking. In addition, the Cramér–Rao lower bound (CRLB) on the measurement error covariance was derived from the Fisher information of the transmitted waveform. Therefore, the transmitted waveform and the tracking filter were connected. References [
16,
17] extended this work to the clutter environment; however, the linear observation model was still utilized. The impacts of different waveforms on tracking performance were studied in [
19,
21,
22,
32], but the dynamic waveform configuration was not considered. The problem of the joint beam and waveform scheduling was investigated in [
5,
23]. One-step-ahead and two-step-ahead algorithms were proposed to select the waveform and sample interval. To optimize the detection threshold and the transmitted waveform jointly, reference [
13] choses the cumulative probability of track loss and the state covariance as the cost function. In [
14,
24], the fractional Fourier transform was exploited to rotate the ambiguity function (AF) of the transmitted waveform. Therefore, the waveform library was established. Additionally, the interacting multiple model (IMM) algorithm was used for the maneuvering target tracking. References [
25,
26,
29] formed the frequency-modulated (FM) waveform library and used the particle filter (PF) to deal with the nonlinear measurement. References [
6,
7] added the LFM library utilizing the IMM probabilistic data association filter (IMM-PDAF) and achieved urban terrain tracking in high clutter. Reference [
9] presented a novel Kalman filter, which was embedded into the extended kernel recursive least squares Kalman filter (Ex-KRLS-KF) algorithm to further improve the tracking performance. The proposed algorithm improved the tracking performance effectively compared to the state-of-the-art algorithms. Reference [
31] proposed a new diffusion sign subband adaptive filtering algorithm with an individual weighting factor (IWF-DSSAF) for distributed estimation in the impulsive noise environment, which achieved better convergence performance than their counterparts. Reference [
35] concerned the application of Huber-based robust unscented Kalman filter (HRUKF) in a nonlinear system. An adaptive strategy was proposed to improve filtering performance with non-Gaussian measurement noise, which had a better performance than the traditional ones. In reference [
27], an adaptive kernel Kalman filter-based belief propagation algorithm is presented to tracking targets in the case of clutter and false alarms. The proposed algorithm has a better tracking performance and lower computation cost compared with other algorithms. Reference [
28] enhances the performance of the interactive multiple model-integrated probabilistic data association algorithm (IMM-IPDA) by the fixed lag smoothing algorithm. Compared with the other recent algorithms in the literature, the algorithm is better in the root-mean-square error (RMSE), true track rate (TTR), and mode probabilities. Reference [
8] introduces a novel tracking algorithm, which can significantly reduce the estimation error during non-maneuvering periods. Therefore, it is very suitable for tracking low maneuvering targets.
The rest of this paper is organized as follows. The state and measurement model, the waveform model, and the clutter model are described in Sect.
2. In Sect.
3, the MPDAF is integrated with the SCKF to structure the MPDA-SCKF. A waveform library based on the FRFT is established, and a direct waveform selection method is presented in Sect.
4. Extensive simulation results to verify the effectiveness of the proposed algorithm are shown in Sect.
5. Section
6 contains the conclusions.