This paragraph from page 4 is really informative:

"Radial velocity surveys, accurate to 3-5 m/s (Butler et al. 1996), have been so far a unique tool for planet discovery. However, we anticipate that high-precision astrometry, both from ground (Mariotti et al. 1998; Booth et al. 1999; Colavita et al. 1999) and in space (Danner & Unwin 1999; Roser 1999; Gilmore et al. 2000), will be among the preferred means for helping all regions of the parameter space Doppler techniques cannot reach. Astrometry has a significant advantage over the radial velocity technique because it measures two rather than one projection of an orbit and thus describes the full three-dimensional geometry. Astrometry, removing the degeneracy on the inclination angle, provides unambiguous mass estimates and directly determines coplanarity for systems of planets. Astrometric techniques can be used to search for planets around young and bright stars (earlier than F), and late M dwarfs. These objects cannot be searched by radial velocity, either because of their spectral properties (absence of relevant spectral lines) or because of intrinsic instability of the stellar atmospheres (active chromospheres, spots, significant rotation). Furthermore, astrometric sensitivity increases for planets with longer periods, thus complementing radial velocity searches, which favor short-period planets. Finally, radial velocity detection limits are currently of about a Saturn mass within 1 AU. As we will see, astrometry with SIM's exquisite sensitivity pushes detection two orders of magnitude lower, down to Earth masses."

This paper has some equations for astrometric wobble. The following equations for doppler wobble are taken from this website. One simply has to combine them with the astrometric wobble equations to figure out the astrometric wobble of the Sun caused by Earth (what a pain)...

"Spectra of the star are taken regularly to determine the doppler shifts of its' spectral lines. This is directly related to the line of sight velocity of the star, from which it is possible to determine its' period and the amplitude of its' velocity change.

The period of the star will be the same as the period of the planet, and can be related to the size of the planets' orbital semi-major axis by Keplers 3rd Law (assuming we know the mass of the star).

r3=GMP2/4pi2 (1)

where; r=semi-major axis, G=gravitational constant, M=mass of star and P=period of orbit

The planets' velocity can also be calculated by assuming that at any given instant the planets' orbital path is approximately circular, applying the equation for circular motion;

Vp=sqrt(GM/r) (2)

We can then substitute for r from Kepler 3rd Law to give;

Vp=cube root(2piGM/P) (3)

The star and the planet are constantly exchanging momentum, but in order to keep the system stable the exchange must be equal and opposite. In other words, the system conserves momentum, and we thus have;

MpVp=MV (4)

where; Mp=mass of planet, Vp=velocity of planet, M=mass of star and V=velocity of star.

Therefore;

Mp=MV/Vp (5)

The mass of the star is inferred from its' spectral type, its' velocity is measured from its' doppler shift, and the planets' velocity is measured as explained above.
This last equation only holds in that form for systems that are perfectly edge on (ie the orbital plane lies along our line of sight to the system). This is because, as the inclination changes the measured velocity of the star drops until the plane of the system is at right angles to our line of sight. At this point the measured doppler velocity is zero. So the measured velocity for the star is actually Vsin(i), where i is the inclination angle of the system. What is measured as V is the minimum
velocity that the star could have, and equation 5 thus becomes;

Mpsin(i)=MVsin(i)/Vp (6)

Mpsin(i) is, therefore, the minimum mass of the planet.

The final ingredient is the eccentricity of the planets' orbit. This is calculated by fitting a sinusoidal curve to a plot of Vsin(i) against time. A circular orbit will produce a perfectly symmetrical sinusoidal curve, but an elliptical orbit will produce a curve which is skewed and stretched, and the exact shape of the curve can be directly related to the eccentricity of the ellipse. "

End quote.