Broadcasting in an Unreliable SINR Model

Pascal Felber, João Leitão
2017 21 Leibniz International Proceedings in Informatics Schloss Dagstuhl-Leibniz-Zentrum für Informatik   unpublished
We investigate distributed algorithms for broadcasting in unreliable wireless networks. Our basic setting is the signal to noise and interference ratio (SINR) model, which captures the physical key characteristics of wireless communication. We consider a dynamic variant of this model in which an adversary can adaptively control the model parameters for each individual transmission. Moreover, we assume that the network devices have no information about the geometry or the topology of the network
more » ... and do neither know the exact model parameters nor do they have any control over them. Our model is intended to capture the inherently unstable and unreliable nature of real wireless transmission, where signal quality and reception depends on many different aspects that are often hard to measure or predict. We show that with moderate adaptations, the broadcast algorithm of Daum et al. [DISC 13] also works in such an adversarial, much more dynamic setting. The algorithm allows to broadcast a single message in a network of size n in time O(D·polylog(n+R)), where D is the diameter and R describes the granularity of the communication graph. 1 Introduction In the signal to noise and interference (SINR) model (a.k.a. the physical interference model), a message is received if and only if the ratio between the signal strength at the receiving node and the combined strength of the background noise and any interfering signals is above a given threshold. By now, the SINR model has become the standard communication model to study wireless network algorithms. In the distributed algorithms literature, different variants of the basic SINR model have been studied, based on the properties of the underlying geometric space, how much geometric information the network devices have and how much they know about the model parameters or the network topology. One of the most general variants of the SINR model has been termed the ad hoc SINR model by Daum et al. in [4], where they study the problem of broadcasting a message to all nodes of a wireless network. In [4], it is assumed that the network nodes have no information about the geometry or the topology of the network, prohibiting algorithms that utilize advance knowledge about network topology and layout or the way that signals propagate in space. The distances between the nodes are assumed to form a general growth-bounded metric space. Furthermore, the nodes use uniform transmission powers, they have only approximate