Flexible Power Scheduling for Sensor Networks Barbara Hohlt Computer Science Division University of California, Berkeley Berkeley, CA, USA 94720-1776 [email protected] Lance Doherty Electrical Engineering Division University of California, Berkeley Berkeley, CA, USA 94720-1770 [email protected] Eric Brewer Computer Science Division University of California, Berkeley Berkeley, CA, USA 94720-1776 [email protected] ABSTRACT We propose a distributed on-demand power-management protocol for collecting data in sensor networks. The protocol aims to reduce power consumption while supporting fluctuating demand in the network and provide local routing information and synchronicity without global control. Energy savings are achieved by powering down nodes during idle times identified through dynamic scheduling. We present a real implementation on wireless sensor nodes based on a novel, two-level architecture. We evaluate our approach through measurements and simulation, and show how the protocol allows adaptive scheduling and enables a smooth trade-off between energy savings and latency. An example current measurement shows an energy savings of 83% on an intermediate node. Categories and Subject Descriptors C.2 [Computer Communication Networks]: Network Architecture and Design, Network Protocols, Network Operations, Distributed Systems; C.3 [Special-Purpose and Application-Based Systems]: Real-time and embedded systems; D.4.4 [Operating Systems]: Communications Management General Terms Algorithms, Management, Measurement, Design, Economics, Experimentation Keywords Sensor Networks, Communication Scheduling, Power Management 1. INTRODUCTION The combination of technological advances in integrated circuitry, MEMS, communication and energy storage has driven the development of low-cost, low-power sensor nodes [14,2]. Networking many nodes through radio communication allows for data collection via multi-hop routing, but the practical limits on available power and the lack of global control present challenges. Constraints imposed by limited energy stores on individual nodes require careful selection of tasks, and as communication is the most costly task in terms of energy, it must be used particularly sparingly. In general, sensor nodes comprising the network have the ability to sense the environment, control actuators, make simple computations, and communicate data either to other nodes or to a centralized observer. We will consider networks consisting of the Crossbow MICA sensor nodes [6] running the UC Berkeley TinyOS operating system [13]. Power consumption limits the utility of sensor networks. Replacing batteries every week in building networks is a laborious task and replacing them in a less friendly environment may not be possible. Researchers agree, above all functions, radio communication dominates the power consumed in wireless sensor networks [2,8,25,19]. At the communication distances typical in sensor networks, listening for information on the radio channel is of a cost similar to transmission of data [21], so unnecessary radio operation must be pared to increase node lifetimes. In addition, the energy cost for a node in idle mode is approximately the same as in receive mode. Therefore, protocols that assume receive and idle power are of little consequence are not efficient for sensor networks. Idle listening, the time spent listening while waiting to receive packets, is a significant cost. Stemm et al. [27] observed that idle listening dominated the energy costs of network interfaces in hand-held devices. It became clear that to reduce power consumption in radios, the radio must be turned off during idle times. Mangione-Smith and Ghang [18] proposed such a scheme for an energy-efficient MAC layer for one-hop mobile devices. This paper presents Flexible Power Scheduling (FPS), a distributed power management protocol for sensor networks that reduces radio power consumption while supporting fluctuating demand in the network. A novel, two-level architecture combines coarse grain scheduling at the routing layer to plan radio on/off times and fine grain medium access control to provide channel access. The protocol provides local communication schedules for a multi-hop sensor network and acts only on locally acquired information. We detail a distributed algorithm that exploits a tree based topology in combination with an adaptive slotted communication schedule to route packets, synchronize with neighbors, and schedule radio on/off times. The assignment and modification of schedules is based on a supply and demand algorithm that allows for implicit and explicit deletion of nodes from the network without global control or re-initialization. Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. IPSN’04, April 26–27, 2004, Berkeley, California, USA. Copyright 2004 ACM 1-58113-846-6/04/0004…$5.00. 215205We use sense-to-gateway data collection as the application driver for our protocol. Sense-to-gateway applications represent a large class of wireless sensor network applications and have a natural tree topology that can be exploited for flexible power scheduling. These applications collect data from the environment and forward the data to a base station where it can be stored in a central database for evaluation. Such applications include equipment tracking, building-wide energy monitoring, habitat monitoring [17], conference room reservations [5], art museum monitoring [24], and automatic lawn sprinklers [7]. The communication is primarily one way: from the data-collecting node to the base station. Our protocol may be extended to two-way communication, but in this paper we will focus primarily on network to gateway communication. Our protocol does not support arbitrary many-to-many communication such as would be required in event tracking applications. The remainder of the paper is organized as follows: Section 2 describes sensor network issues, Section 3 introduces our flexible power scheduling protocol, Section 4 makes the algorithm precise, Section 5 describes an implementation on UC Berkeley nodes and