I. Introduction
Pulsed sensing systems such as radars play a key role in surveillance, detection and tracking of targets. Such a system typically operates by transmitting a pulse and performing a matched filtering operation on the incoming delayed and Doppler-shifted pulse echoes. This process of transmitting and receiving incoming pulses is repeated a number of times (slow-time) within a defined coherent processing interval. The collected data is then often stacked together in a matrix and a Fourier transform across the slow-time domain is implemented to construct a range-Doppler map. A range-Doppler map can be used for target detection and tracking as it provides estimates of both range and velocity. Modern radars are further typically equipped with electronically steering arrays and are able to digitally steer a beam instantaneously. While emitting pulses at a specific direction a radar may desire to skip over a few pulses and rather transmit these towards a different angle for various purposes, such as tracking, and then come back to the main angle to resume standard operation. Alternatively, a radar may split the number of pulses available within a coherent processing interval between several directions and only transmit all the pulses at the same angle if some unusual activity does show up. This type of continuous beam switching, with the full array, introduces empty incoherent data gaps in the slow-time domain and a range-Doppler profile has to be composed with fewer pulses, resulting in limited integration gain and lower Doppler bin resolution. Lack of data within the dwell period may also occur due to other causes such as high duty cycle and hardware limitations.