Randomized Communication in Radio Networks

TL;DR

This paper studies randomized communication protocols in radio networks, focusing on how network topology and node knowledge affect the efficiency and design of distributed algorithms for message exchange.

Contribution

It analyzes algorithmic challenges in randomized protocols for radio networks with various topologies and node knowledge scenarios, providing new insights into their design and performance.

Findings

Protocols depend heavily on network topology and node knowledge.

Single-hop networks use protocols like Aloha and exponential backoff.

Multi-hop networks require routing and are robust to unknown neighbors.

Abstract

A communication network is called a radio network if its nodes exchange messages in the following restricted way. First, a send operation performed by a node delivers copies of the same message to all directly reachable nodes. Secondly, a node can successfully receive an incoming message only if exactly one of its neighbors sent a message in that step. It is this semantics of how ports at nodes send and receive messages that defines the networks rather than the fact that only radio waves are used as a medium of communication; but if that is the case then just a single frequency is used. We discuss algorithmic aspects of exchanging information in such networks, concentrating on distributed randomized protocols. Specific problems and solutions depend a lot on the topology of the underlying reachability graph and how much the nodes know about it. In single-hop networks each pair of nodes can communicate directly. This kind of networks is also known as the multiple access channel. Popular broadcasting protocols used on such channels are Aloha and the exponential backoff. Multi-hop networks may have arbitrary topology and packets need to be routed hopping through a sequence of adjacent nodes. Distributed protocols run by such networks are usually robust enough not to expect the nodes to know their neighbors. These ad-hoc networks and protocols model the situation when nodes are mobile and do not rely on a fixed infrastructure.