Decoupling the permanent identifier of a node from the node's topology-dependent address is a promising approach toward completely scalable self-organizing networks. A group of proposals that have adopted such an approach use the same structure to: address nodes, perform routing, and implement location service. In this way, the consistency of the routing protocol relies on the coherent sharing of the addressing space among all nodes in the network. Such proposals use a logical tree-like structure where routes in this space correspond to routes in the physical level. The advantage of tree-like spaces is that it allows for simple address assignment and management. Nevertheless, it has low route selection flexibility, which results in low routing performance and poor resilience to failures. In this paper, we propose to increase the number of paths using incomplete hypercubes. The design of more complex structures, like multi-dimensional Cartesian spaces, improves the resilience and routing performance due to the flexibility in route selection. We present a framework for using hypercubes to implement indirect routing. This framework allows to give a solution adapted to the dynamics of the network, providing a proactive and reactive routing protocols, our major contributions. We show that, contrary to traditional approaches, our proposal supports more dynamic networks and is more robust to node failures.