Randomized Initialization of a Wireless Multihop Network

TL;DR

This paper presents a fully distributed randomized protocol for assigning unique identifiers to nodes in a wireless multihop network, considering random node placement, limited communication range, and no prior topology knowledge.

Contribution

It introduces a novel protocol that achieves address autoconfiguration in a challenging wireless setting with minimal assumptions.

Findings

Protocol runs in expected time O(n^{3/2} log^2 n)

Transmitting range affects network connectivity and properties

Addresses are uniquely assigned without prior topology knowledge

Abstract

Address autoconfiguration is an important mechanism required to set the IP address of a node automatically in a wireless network. The address autoconfiguration, also known as initialization or naming, consists to give a unique identifier ranging from 1 to $n$ for a set of $n$ indistinguishable nodes. We consider a wireless network where $n$ nodes (processors) are randomly thrown in a square $X$, uniformly and independently. We assume that the network is synchronous and two nodes are able to communicate if they are within distance at most of $r$ of each other ($r$ is the transmitting/receiving range). The model of this paper concerns nodes without the collision detection ability: if two or more neighbors of a processor $u$ transmit concurrently at the same time, then $u$ would not receive either messages. We suppose also that nodes know neither the topology of the network nor the number of nodes in the network. Moreover, they start indistinguishable, anonymous and unnamed. Under this extremal scenario, we design and analyze a fully distributed protocol to achieve the initialization task for a wireless multihop network of $n$ nodes uniformly scattered in a square $X$. We show how the transmitting range of the deployed stations can affect the typical characteristics such as the degrees and the diameter of the network. By allowing the nodes to transmit at a range $r= \sqrt{\frac{(1+\ell) \ln{n} \SIZE}{\pi n}}$ (slightly greater than the one required to have a connected network), we show how to design a randomized protocol running in expected time $O(n^{3/2} \log^2{n})$ in order to assign a unique number ranging from 1 to $n$ to each of the $n$ participating nodes.