Disclosed are techniques for wireless communication. In an aspect, a network node receives a reference signal comprising a plurality of time and frequency resources, the plurality of time and frequency resources spanning a plurality of disjoint bandwidth segments, the plurality of disjoint bandwidth segments being time aligned or having a known time offset with respect to each other, and determines a time of arrival (ToA) of the reference signal based on jointly processing the plurality of time and frequency resources of the reference signal across the plurality of disjoint bandwidth segments.