In this chapter, we again look at applying SNR-based and MI-based waveform design to an M-target ID problem. We first derive expressions for the optimum waveform spectrum under both design metrics, and show how both depend on a function called the spectral variance [2]. The formulation for the SNR-based waveform is different from the one used in [6,7] and uses the unifying spectral variance concept. We then compute a probability-weighted effective spectral variance over the target hypotheses and substitute it into the spectral variance quantity needed for waveform design. Using simulation, we evaluate the efficacy of these waveform strategies in comparison to a wideband pulse. We also consider an iterative procedure where the hypothesis probabilities are updated after each transmission. The updated probabilities can then be used to adapt subsequent waveforms. In the multiple-transmission scheme, sequential hypothesis testing is used to control when the transmissions may cease. It is shown that the number of transmission can be considerably reduced by choosing one of the optimized and adaptable waveform strategies. Furthermore, the relative performance of the waveforms is different under multiple transmissions than it is for a single transmission.
Iterative Technique for System Identification with Adaptive Signal Design, Page 1 of 2
< Previous page Next page > /docserver/preview/fulltext/books/ra/sbra023e/SBRA023E_ch61-1.gif /docserver/preview/fulltext/books/ra/sbra023e/SBRA023E_ch61-2.gif