The Energy Complexity of Broadcast

TL;DR

This paper explores the energy complexity of broadcast in multi-hop radio networks, establishing lower bounds, near-optimal algorithms, and tradeoffs between energy and time efficiency.

Contribution

It links energy complexity to leader election time bounds and provides algorithms that nearly match these bounds, highlighting the tradeoff between energy and time.

Findings

Energy complexity lower bounds are derived from leader election bounds.

Algorithms achieving near-optimal energy usage are proposed.

A tradeoff between energy and time complexity is demonstrated.

Abstract

Energy is often the most constrained resource in networks of battery-powered devices, and as devices become smaller, they spend a larger fraction of their energy on communication (transceiver usage) not computation. As an imperfect proxy for true energy usage, we define energy complexity to be the number of time slots a device transmits/listens; idle time and computation are free. In this paper we investigate the energy complexity of fundamental communication primitives such as broadcast in multi-hop radio networks. We consider models with collision detection (CD) and without (No-CD), as well as both randomized and deterministic algorithms. Some take-away messages from this work include: 1. The energy complexity of broadcast in a multi-hop network is intimately connected to the time complexity of leader election in a single-hop (clique) network. Many existing lower bounds on time complexity immediately transfer to energy complexity. For example, in the CD and No-CD models, we need $\Omega(\log n)$ and $\Omega(\log^2 n)$ energy, respectively. 2. The energy lower bounds above can almost be achieved, given sufficient ($\Omega(n)$) time. In the CD and No-CD models we can solve broadcast using $O(\frac{\log n\log\log n}{\log\log\log n})$ energy and $O(\log^3 n)$ energy, respectively. 3. The complexity measures of Energy and Time are in conflict, and it is an open problem whether both can be minimized simultaneously. We give a tradeoff showing it is possible to be nearly optimal in both measures simultaneously. For any constant $\epsilon>0$, broadcast can be solved in $O(D^{1+\epsilon}\log^{O(1/\epsilon)} n)$ time with $O(\log^{O(1/\epsilon)} n)$ energy, where $D$ is the diameter of the network.