Most of the earlier approaches for frequency‐hopping (FH) signals sorting with a single sensor do not adapt to synchronous network, whereas the multiple‐sensor‐based algorithms request the number of sensors must be more than that of signals. Since the number of sensors is limited in many practical applications, it is important to sort synchronous or asynchronous FH networks with as little as possible sensors. This study introduces the underdetermined blind source separation (UBSS) algorithm to solve this problem. Considering the time–frequency (TF) sparsity of FH signals, the problem is formulated as one of UBSS based on sparse TF representation. First, an improved k ‐means clustering algorithm is developed to estimate the mixing matrix, according to the TF properties of FH signals. Secondly, to separate more signals overlapped in the TF domain using given sensors, an improved subspace‐based algorithm utilising the information of the comparative power is proposed. In the proposed method, the sparsity condition of the sources in the TF domain is relaxed, and the number of FH signals that exist at any TF point simultaneously is allowed to equal that of the sensors. Simulations demonstrate that the proposed method can separate FH signals efficiently and outperforms the previous methods.

This paper addresses the problem of online hop timing detection and frequency estimation of multiple frequency-hopping (FH) signals with antenna arrays. The problem is deemed as a dynamic one, as no information about the hop timing, pattern, or rate is known in advance, and the hop rate may change during the observation time. The technique of particle filtering is introduced to solve this dynamic problem, and real-time frequency and direction of arrival estimates of the FH signals can be obtained directly, while the hop timing is detected online according to the temporal autoregressive moving average process. The problem of network sorting is also addressed in this paper. Numerical examples are carried out to show the performance of the proposed method.

This paper addresses the problem of direction-of-arrival (DOA) estimation of multiple wideband coherent chirp signals, and a new method is proposed. The new method is based on signal component analysis of the array output covariance, instead of the complicated time-frequency analysis used in previous literatures, and thus is more compact and effectively avoids possible signal energy loss during the hyper-processes. Moreover, the a priori information of signal number is no longer a necessity for DOA estimation in the new method. Simulation results demonstrate the performance superiority of the new method over previous ones.