Turbo MRC-SMDS: Low-complexity Cooperative Localization from Hybrid Information

We introduce a complex-domain1 reformulation of the super multidimensional scaling (SMDS) wireless localization framework, obtaining from it an entirely new method to extract accurate location information from hybrid (angles and distances) information. Specifically, under this reformulation, the SMDS edge kernel is complex-valued and its block structure exposes clear relationships between anchor-to-anchor, anchor-to-target and target-to-target information dependencies. Exploiting these features, several new SMDS algorithms are designed which not only eliminate the need for eigen-decompositions in favor of much simpler vector multiplication operations similar to maximum ratio combining, but also are suited to particular data erasure structures emerging from typical and practical conditions faced by wireless localization systems. It is shown that these new algorithms offer different complexity/performance improvements, culminating with a new iterative design which is both faster and more accurate than the original SMDS method.