Tiered sensor network architectures are gaining currency. In contrast with flat networks of impoverished nodes (the hitherto common assumption in sensor networking), such systems offer the promise of migrating computational load from sensing nodes to higher capability ‘master’ nodes. We argue that for certain data fusion-based services this means that compute intensive algorithms, often shunned as impractical for sensor networks, are in fact a viable possibility. Using localization as an example, we show how accurate results may be obtained by leveraging this capability without the use of specialized hardware or high configuration detail; both of which are standard approaches to the problem when computation is at a premium. Specifically, we propose a mathematical optimization-based framework for localization based on proximity constraints. Most variants of localization can be cast into this framework depending on the kinds of input available (e.g. ranging). We show accurate results, and exploit a technique from distributed optimization to divide the problem into pieces suitable for computation at the master-level nodes. We conclude with remarks on the general implications of this example for tiered systems, with pointers on how it is likely to be applicable to other problems such as power-aware routing.