The Einstein Telescope is a conceived third generation gravitational-wave detector
that is envisioned to be an order of magnitude more sensitive than advanced LIGO,
Virgo and Kagra, which would be able to detect gravitational-wave signals from the
coalescence of compact objects with waveforms starting as low as 1Hz. With this level
of sensitivity, we expect to detect sources at cosmological distances. In this paper
we introduce an improved method for the generation of mock data and analyse it with
a new low latency compact binary search pipeline called gstlal. We present the results
from this analysis with a focus on low frequency analysis of binary neutron stars.
Despite compact binary coalescence signals lasting hours in the Einstein Telescope
sensitivity band when starting at 5 Hz, we show that we are able to discern various
overlapping signals from one another. We also determine the detection efficiency for
each of the analysis runs conducted and and show a proof of concept method for estimating
the number signals as a function of redshift. Finally, we show that our ability to
recover the signal parameters has improved by an order of magnitude when compared
to the results of the first mock data and science challenge. For binary neutron stars
we are able to recover the total mass and chirp mass to within 0.5% and 0.05%, respectively.