For future configurations, we study the relation between the abatement of the noise sources and the Signal to Noise Ratio (SNR) for coalescing binaries.
Our aim is not the proposition of a new design, but an indication of where in the bandwidth or for which noise source, a noise reduction would be most efficient. We take VIRGO as the reference for our considerations, solely applicable to the inspiralling phase of a coalescing binary. Thus, only neutron stars and small black holes of few solar masses are encompassed by our analysis. The contributions to the SNR given by final merge and quasi-normal ringing are neglected. It is identified that i) the reduction in the mirror thermal noise band provides the highest gain for the SNR, when the VIRGO bandwidth is divided according to the dominant noises; ii) it exists a specific frequency at which lies the potential largest increment in the SNR, and that the enlargement of the bandwidth, where the noise is reduced, produces a shift of such optimal frequency to higher values; iii) the abatement of the pendulum thermal noise provides the largest, but modest, gain, when noise sources are considered separately. Our recent astrophysical analysis on event rates for neutron stars leads to a detection rate of one every 148 or 125 years for VIRGO and LIGO, respectively, while a recently proposed and improved, but still conservative, VIRGO configuration would provide an increase to 1.5 events per year. Instead, a bi-monthly event rate, similar to advanced LIGO, requires a 16 times gain. We analyse the 3D (pendulum, mirror, shot noises) parameter space showing how such gain could be achieved.