1: \begin{abstract}

2: 

3: The cold dark matter (CDM) structure formation model predicts

4: that about 5-10 \% of a typical galactic 

5: halo of mass $\sim 10^{12} \ms$

6: is in substructures with masses $\lesssim 10^8 \ms$.

7: To directly detect such substructures, we propose to 

8: observe dust continuum emission from a strongly lensed 

9: QSO-host galaxy using a large submillimeter 

10: interferometer array with a high angular 

11: resolution of $\sim 0.01$arcsec such as the planned 

12: Atacama Large Submillimeter Array (ALMA). 

13: To assess their observational feasibility, 

14: we numerically simulate millilensing of 

15: an extended circular 

16: source by a CDM substructure modeled as a

17: tidally truncated singular isothermal sphere

18: (SIS) embedded in a typical QSO-galaxy lens system, B1422+231, 

19: modeled as a singular isothermal ellipsoid (SIE) with an 

20: external constant shear and a constant convergence. 

21: Assuming an angular resolution of $0.01$arcsec, we find that the 

22: angular positions of $\sim 10^8 \ms$ substructures at several kpc

23: from the center of the macrolens halo can be 

24: directly measured if the 

25: size of the dust continuum emission region and

26: the gradient of the surface brightness  

27: at the position of the perturber

28: are sufficiently large.

29: From the astrometric shift on a scale of a few times $10~$mas 

30: of an image perturbed by a subhalo 

31: with respect to an unperturbed macrolensed image, 

32: we can break the degeneracy between subhalo mass and distance   

33: provided that macrolensing parameters are determined 

34: from positions and fluxes of multiple images.

35: \end{abstract}

36: