Time-delay interferometry for LISA

LISA (Laser Interferometer Space Antenna) is a mission to detect and study low-frequency cosmic gravitational radiation through its influence on the phases or frequencies of laser beams exchanged between three remote spacecraft. We previously showed how, with lasers of identical frequencies on stationary spacecraft, the measurement of twelve time series of Doppler shifts could be combined to cancel exactly the phase noise of the lasers and the Doppler fluctuations due to noninertial motions of the six optical benches, while preserving gravitational wave signals. Here we generalize those results on gravitational wave detection with time-delay interferometry to the expected LISA instrument. The six lasers have different center frequencies (in the nominal LISA configuration these center frequencies may well differ by several hundred megahertz) and the distances between spacecraft pairs will change with time (these slowly varying orbital Doppler shifts are expected to be up to tens of megahertz). We develop time-delay data combinations which, as previously, preserve gravitational waves and exactly cancel the leading noise source (phase fluctuations of the six lasers); these data combinations then imply transfer functions for the remaining system noises. Using these, we plot frequency and phase power spectra for modeled system noises in the unequal Michelson combination X and the symmetric Sagnac combination zeta. Although optical bench noise can no longer be cancelled exactly, with the current LISA specifications it is suppressed to negligible levels. It is known that the presently anticipated laser center frequency differences and the orbital Doppler drifts introduce another source of phase noise, arising from the onboard oscillators required to track the photodetector fringes. For the presently planned mission, our analysis indeed demonstrates that noise from current-generation ultrastable oscillators would, if uncorrected, dominate the LISA noise budget. To meet the LISA sensitivity goals either achievable improvements in oscillator stability must be combined with much stricter requirements on the allowed laser center frequency differences and on the allowed Doppler shifts from orbital drifts or, as has been previously suggested, additional calibrating interspacecraft data must be taken, by modulating the laser beams and considerably increasing system complexity. We generalize prior schemes for obtaining the required oscillator instability calibration data to the case of six proof masses, six lasers, and three onboard oscillators. For this realistic configuration we derive appropriate time-delayed combinations of the calibrating data to correct each of the laser-noise-free data combinations.