Analysis of the current state of the artThe techniques normally used for generating pulses in low-cost monolithic technology (CMOS or BiCMOS) are different from those usually used in hybrid technologies. The latter mainly use edge combination methods generated from SRD diodes via quarter-wave lines or stubs (Jeongwoo Han et al., 2004). In a low-cost context, these techniques are not possible, as they cannot be integrated into the desired technologies. We shall present here only the following techniques:Baseband pulse transposition -Baseband pulse filtering -Synthesis by pulse combination.This analysis is performed using circuits that have been built and measured and that can be found in the literature. Before commencing this analysis, it is necessary to point out that the figure of merit often used in pulse generation is the energy consumed by the pulse. This criteria is subject to a caveat, as it often fails to take into account the power consumed between two pulses, which means it is only relevant if the system is combined with power management (sometimes difficult to achieve, depending on the topology of the circuit). This is why we shall be presenting, wherever possible, an estimate of the total energy consumed by the pulse (Ec t =P DC /D S ) that take into account the average consumption between two pulses. The energy consumed per pulse will then be used to quantify the energy consumed during the pulse alone (), and if this is not given, it will be approximated as follows: E c =P DC *. Moreover, in order to be completely suitable, the energy consumed by the total pulse must be compared to the energy of the pulse produced (E p ) by way of the energy efficiency per pulse (n=E p /Ec t ). As this energy is rarely given, it will be estimated by assuming that the pulse is a sinusoid modulated by a gate of width and amplitude V p (E p =.V p 2 /2R 0 ).