27 Février – Thesis defense - Mahdi Saleh

10 h15 Amphi Jean-Paul Dom - Laboratory IMS (building A31) / Talence campus

Contributions to high range resolution radar waveforms: design of complete processing chains of various intra-pulse modulated stepped-frequency waveforms.

In various radar systems, a great deal of interest has been paid to selecting a waveform and designing a whole processing chain from the transmitter to the receiver to obtain the high range resolution profile (HRRP). For the last decades, radar designers have focused their attentions on different waveforms such as the pulse compression waveforms and the stepped frequency (SF) waveform:
On the one hand, three different types of wide-band pulse compression waveforms have been proposed: the linear frequency modulation (LFM) waveform, the phase coded
(PC) waveform, and the non-linear frequency modulation (NLFM) waveform. They are very popular but the sampling frequency at the receiver is usually large. This hence requires
an expensive analog-to-digital convertor (ADC). In addition, the PC and NLFM waveforms may be preferable in some high range resolution applications since they lead to peak sidelobe ratio (PSLR) and integrated sidelobe ratio (ISLR) better than the ones obtained with the LFM waveform.
On the other hand, when dealing with SF waveforms, a small sampling frequency can be considered, making it possible to use a cheap ADC.
Pulse compression and SF waveforms can be combined to take advantage of both. Although the standard combination of PC or NLFM with SF leads to the exploitation of a cheap ADC, the performance of the PC waveform or NLFM waveform in terms of PSLR and ISLR cannot be attained. As the PSLR and the ISLR have a great influence on the probability of detection and probability of false alarm, our purpose in the PhD dissertation is to present two new processing chains, from the transmitter to the receiver:
1)    In the first approach, the spectrum of a wide-band pulse compression pulse is split into a predetermined number of portions. Then, the time-domain transformed versions of these various portions are transmitted. At the receiver, the received echoes can be either processed with a modified FD algorithm or a novel timewaveform reconstruction (TWR) algorithm. A comparative study is carried out between the different processing chains, from the transmitter to the receiver, that can be designed. Our simulations show that the performance in terms of PSLR and ISLR obtained with the TWR algorithm is better than that of the modified FD algorithm for a certain number of portions. This comes at the expense of an additional
computational cost. Moreover, whatever the pulse compression used, the approach we present outperforms the standard SF waveforms in terms of PSLR and ISLR.
2)    In the second approach, we suggest approximating the wide-band NLFM by a piecewise linear waveform and then using it in a SF framework. Thus, a variable chirp rate SF-LFM waveform is proposed where SF is combined with a train of LFM pulses having different chirp rates with different durations and bandwidths.
The parameters of the proposed waveform are derived from the wide-band NLFM waveform. Then, their selection is done by considering a multi-objective optimization issue taking into account the PSLR, the ISLR and the range resolution.
The latter is addressed by using a genetic algorithm. Depending on the weights used in the multi-objective criterion and the number of LFM pulses that is considered, the performance of the resulting waveforms vary.
An appendix is finally provided in which additional works are presented dealing with model comparison based on Jeffreys divergence.

Event localization