IMPLEMENTATION OF TEAN-SLEEP FOR WIRELESS SENSOR NETWORKS Jesus Jaquez, David Valencia, Manikanden Balakrishnan, Eric E. Johnson and Hong Huang New Mexico State University Las Cruces, NM August 19, 2009 ABSTRACT Wireless Sensor Network have many potential applications, including environmental monitoring and threat surveillance, among many others. Sensor nodes are limited in processing capabilities, radio range, storage but most importantly in energy. When a node’s energy is exhausted, the node can no longer provide sensor data to the network, nor can it forward data from other nodes. This can result in a network partition. One way a node may extend its lifetime is by the use of sleep. In many of the applications of WSNs the nodes are idle most of the time until an event occurs and information needs to be transmitted. A node is able to save energy while idle by sleeping. Topology and Energy Adaptive, Non-synchronous (TEAN) sleep provides the nodes in a network a mechanism to save energy by sleeping while also keeping a connected network. In this paper, we present an implementation of TEAN- sleep in tinyOS on a 50 node test bed of Crossbow TelosB motes. Preliminary test results show significant improvement in network lifetime when TEAN-sleep is engaged, with insignif- icant degradation of connectivity. I. I NTRODUCTION Wireless Sensor Networks consist of sensor nodes which have the capability to communicate, process information, and sense the area around them. These devices work together in order to perform such tasks as environmental and disaster monitoring [1]. In many applications nodes are idle for a huge percentage of their lifetime. Many approaches have been used in order to try to extend the operational lifetime of a network. These include topology control and energy harvesting, among many others. In this paper we provide an implementation of the TEAN- sleep algorithm [2] in tiny Operating System (tinyOS) [3]. This implementation is tested on Crossbow Telosb motes [4]. The paper gives a review of TEAN-sleep, details regarding its implementation, performance analysis, and finally conclusions. A. Related Work and Original Contributions Conserving energy and extending the lifetime of sensor nodes is an important topic in the area of WSNs. Many pro- posals have been made in order to address possible solutions to these problems. Researchers have investigated the use of sleep scheduling as in [5] and some which pair sleep scheduling with the reduction of redundant sensing by nodes which cover the same area [6]. In [7] locally available information is used in order to make decisions about sleeping. Each node calculates a drowsiness factor, an estimate of a sensors current energy. It then will compute its Shout Time Delay. If the node has a small Shout Time Delay then it is more likely to be able to sleep. Original Contributions The original work [2] aimed at introducing the framework, and simulations were done under idealistic channel conditions. The predicted alpha-performance of connectivity based sleep coordination is heavily dependent on the underlying link behavior, and thus analysis under real wireless channel environments is required to quantify TEAN-sleep performance. In that perspective, the following are original contributions of this paper: • Update of Sleep Eligibility Condition • Implementation and verification of TEAN-sleep algo- rithm • Link establishment and maintenance • Design and implementation of the routing layer II. TEANSLEEP TEAN-sleep was first introduced in [2]. The premise of TEAN-sleep is to allow nodes in a network, who are idle, to sleep, given that they meet certain energy and connectivity requirements, which are described in the following sections. TEAN-sleep has the following characteristics [2]: • Topology adaptive: WSNs are densely deployed for ro- bustness against frequent node and link failures, which implies only a subset of nodes is required to establish a connection backbone for reliable information supply. The rest of the nodes could sleep to conserve energy without impacting network connectivity. TEAN-sleep ap- plies a sleep condition that ensures network connectivity is maintained and only the redundant nodes are switched off. Nodes adapt their sleep decisions based on locally computed neighborhood topology. • Energy adaptive: When preparing to sleep, a node will calculate the time it can go to sleep based of its energy level relative to the average energy of its 1-hop neighbors. This will cause energy starved nodes to sleep for longer amounts of time. Sleep Duration, T S ,is calculated as: T S = 1 - P A P A T A (1) Equation 1 defines the amount of time that a node will be able to sleep. In this is equation P A is the probability that a