Scheduling to Minimize Energy and Flow Time in Broadcast Scheduling

TL;DR

This paper studies energy-efficient broadcast scheduling by dynamically scaling transmission speed to minimize combined flow time and energy, providing scalable algorithms under resource augmentation.

Contribution

It introduces the first study of energy minimization in broadcast scheduling with convex power functions and develops scalable algorithms using resource augmentation.

Findings

Established an $oldsymbol{ ilde{O}(1)}$-approximate algorithm under resource augmentation.

Provided a scalable non-clairvoyant algorithm for weighted flow time plus energy.

Extended the model to general convex power functions.

Abstract

In this paper we initiate the study of minimizing power consumption in the broadcast scheduling model. In this setting there is a wireless transmitter. Over time requests arrive at the transmitter for pages of information. Multiple requests may be for the same page. When a page is transmitted, all requests for that page receive the transmission simulteneously. The speed the transmitter sends data at can be dynamically scaled to conserve energy. We consider the problem of minimizing flow time plus energy, the most popular scheduling metric considered in the standard scheduling model when the scheduler is energy aware. We will assume that the power consumed is modeled by an arbitrary convex function. For this problem there is a $\Omega(n)$ lower bound. Due to the lower bound, we consider the resource augmentation model of Gupta \etal \cite{GuptaKP10}. Using resource augmentation, we give a scalable algorithm. Our result also gives a scalable non-clairvoyant algorithm for minimizing weighted flow time plus energy in the standard scheduling model.