Interval-scaling for multitarget localization

摘要

In this paper we illustrate the potential of the generic I-SCAL framework for localization. Much like algebraic Multidimensional Scaling, which was originally utilized in other fields of science until it was identified as suitable for localization, I-SCAL is a SMACOF optimization-based generic framework which, to the best of our knowledge is here, for the first time, employed to solve the localization problem. To do so we propose to modify the rectangular objects employed in the standard I-SCAL framework with circular ones, resulting in faster and better performing algorithm in standard localization scenario. In addition it is shown that the computational complexity be further reduced by means of a vector extrapolation stage added in the optimization stage. The application of the proposed algorithm to the two standard localization scenarios here considered shows that the I-SCAL algorithm outperforms the SMACOF algorithm.