Probabilistic Addressing in Wireless Networks

The lack of permanent network infrastructure and often unplanned deployments in many multihop wireless communication scenarios restrict nodes to determine their own addresses based on the underlying connectivity in the network. However, due to unreliable connectivity and rapidly changing link qualities in wireless networks, establishing uniform addressing and stable point-to-point routing is challenging. In this paper, we present Statistical Vector Routing (SVR), a virtual coordinates based addressing and routing mechanism that efficiently deals with dynamic communication links in wireless networks. It assigns stable probabilistic addresses to nodes without the need to pessimistically estimate links over longer periods of time. The routing metric predicts the current location of a node in its address distribution. Our prototype implementation over real testbeds indicates that SVR, when compared to current approaches, achieves 3-7 times more stable addressing, reduces the magnitude of change in addresses by 2-10 times, and minimizes the hop distance and transmissions in the network by 10-15%.