Power-Efficient Data Propagation Protocols for Wireless Sensor ...

SIMULATION http://sim.sagepub.com

Power-Efficient Data Propagation Protocols for Wireless Sensor Networks Azzedine Boukerche, Ioannis Chatzigiannakis and Sotiris Nikoletseas SIMULATION 2005; 81; 399 DOI: 10.1177/0037549705056220 The online version of this article can be found at: http://sim.sagepub.com/cgi/content/abstract/81/6/399

Published by: http://www.sagepublications.com

On behalf of:

Society for Modeling and Simulation International (SCS)

Additional services and information for SIMULATION can be found at: Email Alerts: http://sim.sagepub.com/cgi/alerts Subscriptions: http://sim.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav

Downloaded from http://sim.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008 © 2005 Simulation Councils Inc.. All rights reserved. Not for commercial use or unauthorized distribution.

Power-Efficient Data Propagation Protocols for Wireless Sensor Networks Azzedine Boukerche SITE University of Ottawa, Canada [email protected] Ioannis Chatzigiannakis Sotiris Nikoletseas Computer Technology Institute and Department of Computer Engineering and Informatics University of Patras 26500 Patras, Greece Wireless sensor networks are composed of a vast number of ultra-small, fully autonomous computing, communication, and sensing devices, with very restricted energy and computing capabilities, that cooperate to accomplish a large sensing task. Such networks can be very useful in practice. The authors propose extended versions of two data propagation protocols: the Sleep-Awake Probabilistic Forwarding (SW-PFR) protocol and the Hierarchical Threshold-Sensitive Energy-Efficient Network (H-TEEN) protocol. These nontrivial extensions aim at improving the performance of the original protocols by introducing sleep-awake periods in the PFR case to save energy and introducing a hierarchy of clustering in the TEEN case to better cope with large network areas. The authors implemented the two protocols and performed an extensive comparison via simulation of various important measures of their performance with a focus on energy consumption. Data propagation under this approach exhibits high fault tolerance and increases network lifetime. Keywords: Wireless sensor networks, data propagation, power awareness, distributed protocols, performance evaluation

1. Introduction

surveillance, targeting, nuclear, biological and chemical attack detection), (b) environmental applications (e.g., fire detection, flood detection, precision agriculture), (c) health applications (e.g., telemonitoring of human physiological data), and (d) home applications (e.g., smart environments and home automation). For an excellent survey of wireless sensor networks, see Akyildiz, Sankarasubramaniam, and Cayirci [1]. Note, however, that the efficient and robust realization of such large, highly dynamic, complex, nonconventional networking environments is a challenging algorithmic and technological task. Features, including the huge number of sensor devices involved, the severe power, computational and memory limitations, their dense deployment, and frequent failures, pose new design and implementation aspects that are essentially different not only with respect to distributed computing and systems approaches but also to ad hoc networking techniques.

Recent dramatic developments in micro-electromechanical (MEMS) systems, wireless communications, and digital electronics have already led to the development of small-sized, low-power, low-cost sensor devices. Such extremely small devices integrate sensing, data processing, and communication capabilities. Large numbers of sensor nodes can be deployed in areas of interest (such as inaccessible terrain or disaster places) and use self-organization and collaborative methods to form a sensor network. Their wide range of applications is based on the possible use of various sensor types (i.e., thermal, visual, seismic, acoustic, radar, magnetic, etc.) to monitor a wide variety of conditions (e.g., temperature, object presence and movement, humidity, pressure, noise levels, etc.). Thus, sensor networks can be used for continuous sensing, event detection, location sensing, and micro-sensing. Hence, sensor networks have important applications, including (a) military (e.g., forces and equipment monitoring, battlefield

SIMULATION, Vol. 81, Issue 6, June 2005 399-411 © 2005 The Society for Modeling and Simulation International DOI: 10.1177/0037549705056220

1.1 Contribution | | | |

We study the problem of multiple event detection and propagation—that is, the local sensing of a series of crucial events and the energy-efficient propagation of data reporting the realization of these events to a (fixed or


Boukerche, Chatzigiannakis, and Nikoletseas

mobile) control center. The control center could, in fact, be some human authorities responsible for taking action upon the realization of the crucial event. We use the term sink for this control center. We note that this problem generalizes the single-event propagation problem (with regard to [26]) and poses new challenges for designing power-efficient data propagation protocols. The protocols we present here can also be used for the more general problem of data propagation in sensor networks (see [7]). Under these more general and realistic (in terms of motivation by practice applications) modeling assumptions, we have implemented and experimentally evaluated two information propagation protocols: (1) the Sleep-Awake Probabilistic Forwarding (SW-PFR) protocol, which avoids flooding by favoring in a probabilistic way certain “close to optimal” data transmissions and also allows particles to alternate between sleeping and awake modes to save energy, and (2) the hierarchical version of the Threshold-Sensitive Energy-Efficient Network (H-TEEN) protocol, where particles self-organize into clusters and build a tree of transmissions, propagating data only to their parent (cluster head) in this tree. We note that we had to carefully design the protocols to work under the new network models. Using simulation, we propose, implement, and evaluate a modified version of the Threshold-Sensitive EnergyEfficient Network (TEEN) protocol introduced in Manjeshwar and Agrawal [8]. In particular, we extended the concept of clustering used in the original TEEN protocol to that of hierarchical clustering. This possibility was discussed in the original paper, but no such protocol was designed. We show that in large-area networks and when the number of layers in the hierarchy is small, TEEN tends to consume a lot of energy because of long-distance transmissions. On the other hand, when the number of layers increases, the transmissions become shorter, but there is a significant overhead in the setup phase as well as the operation of the network. In light of the above, we chose to implement a four-layer hierarchical clustering. The main idea behind the hierarchical clustering is to enable the TEEN protocol to operate in networks that occupy large areas. As opposed to Chatzigiannakis, Kinalis, and Nikoletseas [9] (where a 2-D lattice deployment of wireless sensor devices has been used), we extend the network model here to the general case of particle deployment according to a random, uniform distribution. We further extend the Probabilistic Forwarding (PFR) protocol and the network model by enabling the wireless sensor particles to periodically enter a power save mode, which we call the sleeping mode. Finally, we implement PFR so as to efficiently handle not only a single event, as described in Chatzigiannakis et al. [2, 3, 10], but to propagate multiple events. The extensive simulations we have performed show that both protocols are successful. In the setting we considered, SW-PFR seems more robust in the case of high rates of event generation and large networks (considering the area they occupy), while H-TEEN seems to perform better in smaller networks. We also concluded that the values of

certain parameters, such as the radius of transmission, are highly critical for the performance of the SW-PFR protocol, given the network area and the number of the particles spread, and we investigate the precise effect of such parameters on performance. In light of the above, we also propose and discuss a hybrid protocol combining the two protocols and an intermediate randomization switching phase between them. This hybrid protocol can be parameterized toward achieving certain desired goals and trade-offs. 1.2 Previous and Related Work In the past few years, wireless sensor networks have attracted a lot of attention from researchers at all levels of the system hierarchy, from the physical layer and communication protocols up to the application layer. A family of negotiation-based information dissemination protocols suitable for wireless sensor networks is presented in Heinzelman, Kulik, and Balakrishnan [11]. Sensor Protocols for Information via Negotiation (SPIN) focus on the efficient dissemination of individual sensor observations to all the sensors in a network. However, in contrast to classic flooding, in SPIN, sensors negotiate with each other about the data they possess using meta-data names. These negotiations ensure that nodes only transmit data when necessary, reducing the energy consumption for useless transmissions. A data dissemination paradigm called directed diffusion for sensor networks is presented in Intanagonwiwat, Govindan, and Estrin [7], where data generated by sensor nodes are named by attribute-value pairs. An observer requests data by sending interests for named data; data matching the interest are then “drawn” down toward that node by selecting a single path or through multiple paths by using a low-latency tree. Intanagonwiwat et al. [12] present an alternative approach that constructs a greedy incremental tree that is more energy efficient and improves path sharing. Furthermore, this work is related to previous research of Chatzigiannakis, Nikoletseas, and Spirakis [5], where new local detection and propagation protocols are proposed that are very energy and time efficient, as shown by a rigorous average case analysis performed in these works under certain simplifying assumptions. In particular, the Local Target Protocol (LTP) uses local optimization to select the next-hop neighbor. For a detailed discussion on data propagation protocols in wireless sensor networks, see also Boukerche and Nikoletseas [13]. For a comparative study of the multi-pathdelivery PFR protocol to single-path delivery (the LTP), see Chatzigiannakis et al. [2, 3] and Chatzigiannakis, Nikoletseas, and Spirakis [5]. For other important aspects of energy efficiency (such as energy balance), see Boukerche and Nikoletseas [14] and Efthymiou, Rolim, and Nikoletseas [15]. For a discussion of challenges in wireless sensor networks research, see Spirakis [16].

400 SIMULATION Volume 81, Number 6


DATA PROPAGATION PROTOCOLS FOR WIRELESS SENSOR NETWORKS

2. The Model Sensor networks are composed of a vast number of ultrasmall homogeneous sensors, which we here call particles. Each particle is a fully autonomous computing and communication device, characterized mainly by its available power supply (battery) and the energy cost of computation and transmission of data. Such particles (in our model here) cannot move. We adopt here (as a starting point) a 2-D (plane) framework: a wireless sensor network (a set of particles) is spread in an area (for a graphical presentation, see Fig. 1). Note that a 2-D setting is also used by other researchers [7, 8, 11, 12, 17, 18]. DEFINITION 1. Let n be the number of wireless sensor devices, and let d (usually measured in numbers of particles/m2 ) be the density of particles in the area. There is a single point in the network area, which we call the sink S, and it represents a control center where data should be propagated to. Furthermore, we assume that there is a setup phase of the wireless sensor network, during which the smart cloud is dropped in the terrain of interest; when using special control messages (which are very short, cheap, and transmitted only once), each sensor particle is provided with the direction of S. By assuming that each smart dust particle has individually a sense of direction and by using these control messages, each particle is aware of the general location of S. We here note that these modeling assumptions can be relaxed by alternatives, such as introducing a preprocessing phase where sensors use localization techniques by executing an underlying protocol f to decide on fictitious virtual coordinates [18] and thus be able to estimate directions and distances of transmissions. Each particle is equipped with a set of monitors (sensors) for light, pressure, humidity, temperature, and so on. Each particle has a broadcast (digital radio) beacon mode, which can be also a directed transmission of angle α around a certain line (possibly using some special kind of antenna; see Fig. 2). The transmission range (which we denote by R) can vary, while the transmission angle (let it be α) is fixed and cannot change throughout the operation of the network (since this would require a modification or movement of the antenna used). Note that the protocols we study in this work can operate even under the broadcast communication mode (i.e., α = 2π). Each particle can be in one of four different modes at any given time regarding the energy consumption. These modes are as follows: (a) transmission of a message, (b) reception of a message, (c) sensing of events, and (d) sleeping. During the sleeping mode, particle ceases any communication with the environment, and thus it is unable to receive any message. In our model, we assume that the energy consumption of a sleeping particle is negligible, but it needs a certain amount of energy to return to the sensing state. Following Heinzelman, Chandrakasan, and Balakrishnan [17], for the case of transmitting and receiving a mes-

Control Center Sensor nodes

Sensor field

Figure 1. A smart dust cloud

o

S

-o R

p' beacon circle

Figure 2. Directed transmission of angle α

sage, we assume the following simple model in which the radio dissipates Eelec to run the transmitter and receiver circuitry and amp for the transmit amplifier to achieve acceptable SNR (signal-to-noise ratio). We also assume an r 2 energy consumption due to channel transmission at distance r. Thus, to transmit a k-bit message at distance r in our model, the radio expends ET (k, r) = ET −elec (k) + ET −amp (k, r), ET (k, r) = Eelec · k + amp · k · r 2 , and to receive this message, the radio expends ER (k) = ER−elec (k), ER (k, r) = Eelec · k, where ET −elec , ER−elec stand for the energy consumed by the transmitter’s and receiver’s electronics, respectively. Concluding, there are four different kinds of energy dissipation, which are as follows: • • • •

ET : energy dissipation for transmission ER : energy dissipation for receiving Eidle : energy dissipation for idle state Epowerup : energy dissipation for returning from sleeping state

For the idle state, we assume that the energy consumed for the circuitry is constant for each time unit and equals Eelec (the time unit is 1 second). On the other hand, the Volume 81, Number 6


SIMULATION 401


power-up energy equals three times the amount of energy consumed in a time unit during the idle state, that is, 3·Eelec . In our simulations, we explicitly measure the above energy costs by adjusting trans , recv , and Eidle to match as close as possible the specifications of the mica platform [19] (see section 3). The particles in the H-TEEN protocol do not exploit the ability to get to the sleeping mode; thus, in that case, the energy dissipation is characterized by the rest of the three types of energy consumption. We also assume that each particle, in protocol H-TEEN, does not spend any amount of energy to listen to what the other nodes of its cluster send to the cluster head. This assumption is also made for TEEN in Manjeshwar and Agrawal [8], where transmissions from particles to cluster heads are done according to a Time Division Multiple Access (TDMA) schedule broadcast by each cluster. We believe that our model depicts accurately enough the technological specifications of real wireless sensor systems. Similar models are being used by other researchers to study sensor networks (see [8, 17]). The above assumptions suggest a strong model that, however, does not trivialize the problem; we believe that even assuming such a model, the design of power-efficient protocols is still a challenging task. In contrast to Intanagonwiwat, Govindan, and Estrin [7] and Karp [20], our model is weaker in the sense that no geolocation abilities are assumed (e.g., a Global Positioning System [GPS] device) for the particles, leading to more generic and thus stronger results. In Hollar [21], a thorough comparative study and description of wireless sensor systems is given, from the technological point of view. 3. Technological Specifications of Wireless Sensor Devices New technology is changing the nature of sensors and the way they interface with data acquisition and control systems. Researchers have developed an open-source hardware and software platform that combines sensing, communications, and computing into a complete architecture. The first commercial generation of this platform was dubbed the Rene Mote, and several thousand of these sensors have been deployed at commercial and research institutions worldwide to promote the development and application of wireless sensor networks. The platforms development community is based on the open-source model, which has become well known with the increasingly popular Linux operating system. Most development work is done in the public domain, and it includes the hardware design and software source code. Users of the technology contribute their developments back to the community so that the base of code and hardware design grows rapidly. Currently, a number of research institutions in the United States are working on centimeterscale (and even smaller) distributed sensor networks [22, 23].

3.1 Hardware Design of Wireless Sensors The basic mica hardware uses a fraction of a watt of power and consists of commercial components a square inch in size. The hardware design consists of a small, lowpower radio and processor board (known as a mote processor/radio, or MPR, board) and one or more sensor boards (known as a mote sensor, or MTS, board). The combination of the two types of boards forms a networkable wireless sensor. The MPR board includes a processor, radio, A/D converter, and battery. The processor is an ATMEL ATMEGA, but there are other processors that would meet the power and cost targets. The processor runs at 4 MHz and has 128 KB of flash memory and 4 KB of SDRAM. In a given network, thousands of sensors could be continuously reporting data, creating heavy data flow. Thus, the overall system is memory constrained, but this characteristic is a common design challenge in any wireless sensor network. The MPR modules contain various sensor interfaces, which are available through a small 51-pin connector that links the MPR and MTS modules. The interface includes an 8-channel, 10-bit A/D converter; a serial UART port; and an I2C serial port. This allows the MPR module to connect to a variety of MTS sensor modules, including MTS modules that use analog sensors as well as digital smart sensors. The MPR module has a guaranteed unique, hard-coded 64-bit address. The processors, radio, and a typical sensor load consume about 100 mW in active mode. This figure should be compared with the 30-µA draw when all components are in sleep mode. The MTS sensor boards currently include light/temperature, two-axis acceleration, and magnetic sensors and 420-mA transmitters. The wireless transmission is at a 4-Kbps rate, and the transmission range may vary. Researchers are also developing a GPS board and a multisensor board that incorporates a small speaker and light, temperature, magnetic, acceleration, and acoustic (microphone) sensing devices. The mica developers community welcomes additional sensor board designs. 3.2 Software and the TinyOS A considerable portion of the challenge faced by the developers of mica devices is in the software embedded in the sensors. The software runs the hardware and network, making sensor measurements, routing measurement data, and controlling power dissipation. In effect, it is the key ingredient that makes the wireless sensor network produce useful information. To this end, a lot of effort has gone into the design of a software environment that supports wireless sensors. The result is a very small operating system named TinyOS, or Tiny micro-threading Operating System, which allows the networking, power management, and sensor measurement details to be abstracted from the core application development. The operating system also creates a standard method




CPU Speed Memory

Power Supply Power Consumption Processor Current Draw Radio Current Draw

Output Device I/O Port Network

4 MHz ROM: 128Kb FLASH SDRAM: 4Kb EEPROM: 4Kb 2X AA batteries 0.75 mW 5.5 mA (active current) < 20 µΑ (sleep mode) 12 mA (transmit current) 1.8 mA (receive current) < 1 µΑ (sleep mode) 3 LEDs Expansion Connected (51 pin) Serial port (Proprietary 16-pin) Wireless 4 Kbits/sec at 916 MHz (ISM band) Radio range depends on antennae configuration

Figure 3. MPR300CB specifications

Software Footprint Transmission Cost Inactive State Peak Load Typical CPU Usage Events Propagate Thru Stack

3.4 Kb 1 µϑ/Bit < 25 µΑ 20 mA < 50 % < 40 µΣ

Figure 4. TinyOS key facts

of developing applications and extending the hardware. Although tiny, this operating system is quite efficient, as shown by the small stack-handling time. 4. The Problem Assume the realization of a series of K crucial events Ei , with each event being sensed by a single particle pi (i = 1, 2, . . . , K). Then the multiple-event propagation problem P is the following: How can each particle pi (i = 1, 2, . . . , K), via cooperation with the rest of the grain particles, efficiently (mainly with respect to energy) propagate information info(Ei ) reporting realization of event Ei to the sink S? We remark that this problem is a generalization of the single-event propagation problem, which is more difficult to cope with. Certainly, because of the dense deployment of sensor particles close to each other, communication between two particles is much more energy efficient than direct transmission to the sink. Furthermore, short-range hop-by-hop transmissions can effectively overcome some of the signal propagation effects in long-distance transmissions and may help to smoothly adjust propagation around obstacles. Finally, the low energy transmission in multihop communication may enhance security, protecting from the undesired discovery of the data propagation operation.

On the other hand, long-range transmissions require the participation of few particles and therefore reduce the overhead on particle resources and provide better network response times. Furthermore, long-range communication permits the deployment of clustering and other efficient techniques, developed for ad hoc wireless networks. In particular, a clustering scheme enables cluster heads to reduce the amount of transmitted data by aggregating information. The above suggest that many diverse approaches exist to the solution of the multiple-event propagation problem P. In any case, the objective is to propagate most, if not all, of the events Ei to the sink efficiently. Note that, because of the multiplicity of events sensed and the small energy supplies, it is not obvious at all that all events sensed manage to be reported to the sink. Efficiency means minimizing the energy consumption in the sensor network, either by minimizing the number of transmissions and also the duplicate messages, observed in multihop delivery, or by reducing data size (by aggregations techniques) and the number of transmissions (by clustering) in long-range communication. Note that an appropriate Medium Access Control (MAC) protocol is required in either case to handle collisions and avoid message retransmissions [9, 29]. Furthermore, an interesting aspect of the problem under investigation is the lifetime of particles since it affects the ability of the network to propagate data to the sink because available routes are reduced as more particles consume their energy resources and “die” (see also [15]). 5. The Sleep-Awake Probabilistic Forwarding Protocol The basic idea of the protocol lies in probabilistically favoring transmissions toward the sink within a thin zone of particles around the line connecting the particle sensing the event E and the sink (see Fig. 5 for graphical representation). Although data propagation along this line is optimal with respect to energy and time cost, such propagation is not always feasible. This is true because, even if initially this direct line was appropriately occupied by sensors, certain sensors on this line might become inactive, either permanently (because their energy has been exhausted) or temporarily (because these sensors might enter a sleeping mode to save energy). The protocol evolves in two phases: Phase 1: The “Front” Creation Phase. During this initial phase, a sufficiently large front is built to ensure that the data propagation process survives for an appropriately large period of time. This front is created by using a “flooding” mechanism for a configurable number of steps. According to this mechanism, the header of each message includes a counter called β. This counter is set to a predefined value by the source particle, when the latter generates the message relative to the sensed event. Following this initialization, each particle, upon receiving a pertinent message containing a positive β counter, reduces its value by 1 and Volume 81, Number 6


SIMULATION 403


E p2

E

p1

2

1

Particles that particiapate in forwarding path

S S

Figure 6. Angle φ and closeness to optimal line

Figure 5. Thin zone of particles around the line connecting the particle sensing the event E and the sink S

deterministically forwards the message toward the sink. To do so, each particle uses directed “angle” transmission to broadcast data to all of its neighbors that lie in the direction of the sink. When the β counter becomes zero, the particle proceeds to the second phase of the SW-PFR protocol. Ultimately, the beta counter determines the length of the first phase of the SW-PFR protocol. Phase 2: The Probabilistic Forwarding Phase. In this second phase, data propagation is done in a probabilistic manner. Each particle calculates a probability of participation in the propagation process. The closer a particle is to the optimal transmission line, connecting the source node E detecting an event and the sink S, the higher its probability to forward data pertinent to that particular sense event. The “forwarding probability” IPf wd is chosen to be IPf wd =

φ , π

where φ is the angle defined by (a) the line connecting the particle performing the random choice and the sensor that initially sensed the event and (b) the line connecting this node to the sink. Remark that, indeed, a bigger angle φ suggests a sensor position closer to the direct line between E and S. Figure 6 displays this graphically. Clearly, when φ = π, then the sensor lies on this line. Thus, we get that φ1 > φ2 implies that for the corresponding particles, p1 , p2 , p1 is closer to the E-S line than p2 ; thus, IPf wd (p1 ) = φ1 φ2 π > π = IPf wd (p2 ). Angle φ calculation. Under appropriate (and realistic) modeling assumptions for sensor particles, angle φ can be locally calculated running a simple subprotocol (see [10] for detailed discussion). Such modeling assumptions include the ability of sensor particles (a) to estimate the direction of a received transmission (e.g., by direction-sensing antennae), (b) to estimate the distance from a nearby particle that did the transmission (e.g., via signal attenuation estimation techniques), (c) to have a common a common

coordinates system, and (d) to know the direction toward the sink (this is possibly done during a setup phase). The above assumptions suggest a strong model that, however, does not trivialize the problem; we believe that even assuming such a model, the design of our protocol is still a challenging task. Furthermore, some of our assumptions are realistic (e.g., it is indeed possible to apply smart antennas in tiny sensors; see [25]) or may become realistic in the near future. Furthermore, some modeling assumptions can be relaxed by alternatives, such as introducing a preprocessing phase in which sensors use localization techniques by executing an underlying protocol f to decide on fictitious virtual coordinates [18] and thus be able to estimate directions and distances of transmissions. Notice that GPS information is not needed for the PFR protocol. Also, there is no need to know the global structure of the network (i.e., the size of the network, the positions of other particles, etc.). The SW-PFR protocol proposed in this article is actually a nontrivial modification of the PFR protocol proposed in Chatzigiannakis et al. [10]. There are two major additions made to that protocol. First, the current version of PFR, SW-PFR, targets sensor networks with multipleevent generations. In particular, at any given time, there could be more than one event being propagated toward the sink. To avoid repeated transmissions and infinite loops, each particle is provided with a limited “cache memory.” In this cache, the particle registers the event IDs for each distinct event it has “heard of.” Each event ID’s uniqueness is guaranteed, by choosing it to be a concatenation of the source particle ID and the time stamp of the sensed event. Upon the receipt of a message, a particle checks whether the pertinent event is enlisted in its cache. If that event is not in the particle’s cache, it is registered, and then the particle proceeds to the proper actions defined by the PFR protocol. However, if the event was already seen, the message is dropped, and no further action is taken. Presumably, a relatively small amount of memory (e.g., up to 2 MB) would be adequate for such purpose. Note that in the future, the particle cache could enforce a policy of limited lifetime for each of its contents, thus




reducing the space requirements to a minimum. Data aggregation also poses a challenge for further study and efficiency assessment. As a second modification, this version of PFR, SW-PFR, encompasses an intriguing sleep-awake scheme. According to this scheme, each particle goes through alternating periods of “sleeping” and “awake.” During a sleeping period, the particles cease any communication with the environment; thus, the power consumption is assumed to be minimal and practically insignificant, whereas when a particle is awake, it consumes the regular amount of energy. In addition, a special energy amount is considered to be spent during the transition from “sleep” to “awake.” We assume that the sleeping/awake time periods alternate stochastically independently in each particle and have durations s, w, respectively. DEFINITION 2. The energy-saving specification is en = s (a typical value for en may be 0.5). s+w 6. A Hierarchical Threshold-Sensitive Energy Efficient Network Protocol TEEN was introduced in Manjeshwar and Agrawal [8] and is essentially an interesting modification of the fundamental Low-Energy Adaptive Clustering Hierarchy (LEACH) protocol [17] for reactive sensor networks. By reactive sensor networks, we mean networks in which particles immediately react only to sudden and drastic changes in the value of a sensed attribute. The basic concepts in TEEN are the clustering of particles and the use of thresholds to decide whether a particle should transmit data to the sink. TEEN uses particle self-organization into clusters, originally proposed in LEACH, to reduce transmissions. We have extended the concept of clustering to a hierarchical clustering, a proposal also made in the original paper. A tree of transmissions is built, where each particle transmits data only to its parent (cluster head) in this tree. To be more specific, TEEN’s operation is divided into rounds, and every round is further divided into a short setup phase, in which the organization in clusters occurs, and a relatively long steady phase, which is the phase in which normal network operation occurs. Setup phase. In this phase, each particle decides on whether it should become a cluster head. This decision is based on a fixed probability IPc (in our simulations, IPc = 0.05) and on whether it has been a cluster head in the last 1/IPc rounds. Specifically, a particle n picks a random number from 0 to 1 and compares it to a threshold T (n), which is calculated in every round as IPc if n ∈ G 1 T (n) = 1−IPc ·(r mod IPc ) otherwise 0 Here, r is the current round and G is the set of particles that have not been elected as cluster heads in the last 1/IPc

rounds. As we can see, this threshold is growing as rounds pass, and so we can ensure that every node alive will become a cluster head at some point within 1/IPc rounds. The same process begins again after 1/IPc rounds. After a particle decides to become a cluster head, it broadcasts an advertisement message to the entire network. The other particles hear the advertisements from all the cluster heads and choose the cluster head they belong to based on the strength of the signal they received. When a particle decides on which cluster it wants to belong to, it transmits a response back to the cluster head of the corresponding cluster. This is probably done using some kind of Carrier Sense Multiple Access MAC protocol to avoid collisions. The cluster head at some point decides that it has heard all of the responses from the particles in its cluster using a timeout and then sets up a TDMA schedule for transmissions. It then broadcasts this schedule to the particles in its cluster. At this time, the next level of hierarchy can be built. We extended the original TEEN paper with the following scheme for hierarchical clustering. The nodes that have become cluster heads decide on whether they should pass on the next level of hierarchy with the same probability IPh (in our simulations, IPh = 0.1) and on whether the number of cluster-head advertisements in the previous level of hierarchy received is less than eight. If a particle decides on moving to the next level of hierarchy, it broadcasts an advertisement message again. Only cluster heads of the previous level decide to which cluster head of the next level they belong to, and this process is repeated for the next levels. In our simulation, we have used four levels of hierarchy. Steady phase. After all the levels of clustering have been set up, the actual data transmissions from particles can begin. The transmissions from particles to cluster heads are done according to the TDMA schedule of each cluster. To reduce interference between different neighboring clusters, any cluster head picks randomly a code from a certain list, and so particles in each cluster use different CDMA codes. Finally, there are two thresholds, the “hard” and the “soft” thresholds, which determine when a particle must transmit data to the base station since particles sense the field continuously. The first one is a value beyond which the particle enters an alarm state. The second one is a small change in the value of the sensed attribute that took place in the field. By the first time the hard threshold is reached, the particle stores the sensed value and transmits it to the base station. If the next sample differs from the stored value by at least the soft threshold, this information should be transmitted back to the base station. An example of these thresholds is when the measured attribute is temperature; we set the hard threshold to be 30deg and the soft threshold any change over 0.5deg . In our simulations, we used the notion of an event, meaning a situation in which particles must report to the base station, rather than just random Volume 81, Number 6


SIMULATION 405


values of the sensed attributes. So, when an event occurs, it is implied that these thresholds are outreached. Some preliminary discussion on the protocol presented here was included in Nikoletseas et al. [26]. 7. Implementation Details For the purpose of the comparative study of the previously described protocols, we have designed a new simulator, named simDust, that was implemented in Linux using C++ and the LEDA [27] algorithmic and data structures library. Another interesting feature of our simulator is the ability to experiment with very large networks. In fact, the complexity of extending existing network simulators and their (in cases of large instances) time-consuming execution were two major reasons for creating this simulator. simDust enables the protocol designer to implement the protocol using just C++ and avoids complicated procedures that involve the use of more than one programming language. In addition, simDust generates all the necessary statistics at the end of each simulation based on a big variety of metrics that are implemented (such as delivery percentage, energy consumption, delivery delay, longevity, etc.). The key points in simDust’s implementation are the following: Operation in rounds: A basic concept used in the simulator is that its operation is divided into discrete rounds, both in SW-PFR and H-TEEN. One round represents a time interval in which a particle can transmit or receive a message and process it according to the protocol that is being simulated. MAC layer assumptions: simDust leaves transmission collisions to be handled by lower MAC-layer protocols and does not take them into account. It is our intention to consider them in next versions of this simulator. Some MAC protocols for avoiding collisions can be found in Chatzigiannakis, Kinalis, and Nikoletseas [9] andYe, Heidemann, and Estrin [24]. Energy assumptions: We have included an energy dissipation scheme for both protocols implemented. In particular, we have assumed that a particle consumes a standard amount of energy Eelec per round while being awake. Furthermore, in each transmission, energy fading is proportional to the square of the distance. For each receive, a node is credited with an amount of energy that practically reflects the power needed to run the transceiver circuit— namely, Eelec . Finally, a particle can switch over to the sleep state to save energy. No energy consumption virtually takes place while the particle remains in the sleep mode since it keeps its transceiver and its sensors shut down. Size of messages: Regarding the communication cost in terms of the bits transmitted per message, we assume that information messages require 1 Kbyte, plus a 40-bit header, containing a 32-bit identifier for the sender particle and an 8-bit code that determines the message type.

SW-PFR specific assumptions: We assume that the protocol’s execution is preceded by an initialization phase, in which each particle discovers all the neighboring particles that are both within transmission range and within the cycle sector demarcated by a predefined angle value. The transmission range of each particle’s transmitter is set to be fixed and preferably of a value significantly smaller than the network diameter. The sensed events that can trigger data propagation in the network can occur at any given round. In addition, in the process of embedding a sleepawake scheme in the SW-PFR protocol, we have used two configurable variables that effectively define the ratio between the duration of the sleep and awake periods. Finally, the required memory where each particle stores the events it has previously seen is considered to be of infinite capacity for reasons of keeping the implementation complexity considerably low. H-TEEN specific assumptions: Our hierarchical implementation uses four levels of hierarchy. We used a relatively large interval of 24 rounds between successive cluster setup rounds. Finally, we have calculated the size of the transmission schedule that is broadcast by each cluster head as a header of 40 bits followed by 32 bits (size of particle ID) times the number of particles included in that particular cluster. 8. Efficiency Metrics On each execution of the experiment, let K be the total number of crucial events (E1 , E2 , . . . , EK ) and k the number of events that were successfully reported to the sink S. Then, we define the success rate as follows: DEFINITION 3. The success rate, IPs , is the fraction of the number of events successfully propagated to the sink over the total number of events, that is, IPs = Kk . Another crucial efficiency measure is the average available energy of each particle in the network over time: DEFINITION 4. Let Ei be the available energy for particle n E i. Then, Eavg = in i is the average energy per particle in the wireless sensor network, where n is the number of the total particles dropped. Finally, we consider the number of alive particles as a measure of efficiency and the network survivability in each case. As in the case of energy, the more particles that are alive, the better. A source of valuable information is also the particular manner in which particles die over time. DEFINITION 5. Let hA (for “active”) be the number of “active” sensor particles participating in the sensor network. We also give another definition concerning the way the events are generated in the network. This parameter critically influences the efficiency measures defined above.




1

1

PFR 500x500

PFR 0.05 0.9

0.9

PFR 1500x1500

PFR 0.8 0.8

H-TEEN 0.8

0.6 0.5 0.4 0.3

H-TEEN 500x500

0.7

Success Rate

0.7

Success Rate

0.8

H-TEEN 0.05

H-TEEN 1500x1500

0.6 0.5 0.4 0.3 0.2

0.2

0.1

0.1 0

0

0

0

200

400

600

800

1000

1200

1400

1600

Network Size

Figure 7. Success rate for various network sizes for injection rates 0.05 and 0.8

DEFINITION 6. Let Is be the injection rate, measured as the probability of occurrence of a crucial event during a round. 9. Simulation Results We evaluate the performance of the two protocols by conducting a comparative experimental study. In our experiments, we generate a variety of sensor fields. The field size ranges from 200 by 200 m to 1500 by 1500 m. We note that these network size studies are significantly larger than those usually investigated in the relevant research and enable a study of the scalability of the network. In these fields, we drop n ∈ [500, 3000] particles randomly uniformly distributed on the smart dust plane. We start by examining the success rate of the protocols with respect to the network size for two different injection rates, Is (0.05 and 0.8). We focus on two extreme values to investigate the divergent protocol behavior in corresponding extreme settings. In Figure 7, the success rate of the protocols is depicted as the network size increases. It is clear that for small dimensions (200 × 200 m, 500 × 500 m), both protocols achieve high performance (≥ 0.85), while performance drops as network size increases. However, SW-PFR’s success rate remains quite high (i.e., ≥ 0.7), even for very large networks (i.e., 1500 × 1500 m), while SW-PFR seems to behave well in both the case of low and high event generation rates. On the contrary, H-TEEN seems to perform poorly in the case of frequent events since its success rate drops below 0.3 in that case. We continue by examining the success rate of the protocols with respect to the Is for two different network sizes (Fig. 8). Initially, for low injection rates, in small networks (500 × 500 m), both protocols behave almost optimally, achieving a success rate close to 1, and decrease as the injection rate increases. However, for a larger network size,

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

Event Rate

Figure 8. Success rate for various injection rates and network area 500 × 500, 1500 × 1500

the impact of the injection rate seems to be more significant. In particular, H-TEEN’s success rate drops from 85% to almost 30%, while PFR, even though its initial success rate is about 70%, seems to be less affected by the increase in injection rate. The apparent dependence of TEEN protocol’s performance from the injection rate is due to its clustering scheme. As injection rate increases, a cluster head is responsible for delivering more events, and thus it consumes more energy during its leadership. If the injection rate becomes too high, cluster heads are more likely to exhaust their energy supplies before a new cluster head is elected. When a cluster head “dies,” the cluster ceases to function, and all events that occur on that cluster are lost until a new cluster head is elected. Furthermore, large network sizes worsen this phenomenon because more energy is required from the cluster head to propagate an event to the sink (since the energy spent in one hop is of the order of the square of the transmission distance). On the other hand, the SW-PFR protocol is mostly unaffected by high injection rates but influenced by larger network sizes. This is due to its multihop nature since more hops are required for the propagation of an event. We move on by examining the average energy consumed per particle in time. We remind the reader that each particle dropped to the smart dust plane has 1 Joule of energy at its disposal. Figure 9 depicts the average amount of energy that each particle has available in each round, for network areas of (500 × 500 m) and (1000 × 1000 m), where the injection rate is Is = 0.05. Note that in both cases, the success rate, IPs , is more than 80%. For the case of the small network, H-TEEN consumes less energy than SW-PFR, while on the larger network, SW-PFR seems less expensive. To explain this phenomenon, we have to consider the nature of the event propagation of the two protocols. The H-TEEN protocol uses a constant number of hops to Volume 81, Number 6


SIMULATION 407


2600

1

PFR 500x500

PFR 1000x1000 0.9

PFR 1000x1000

2350

H-TEEN 500x500

H-TEEN 500x500 H-TEEN 1000x1000

0.8

0.7

Alive Particles

Average Energy per Particle (J)

PFR 500x500

H-TEEN 1000x1000

2100

1850

1600

0.6

1350

0.5

1100 0

1000

2000

3000

4000

5000

6000

7000

8000

Simulation Rounds

0

3000

6000

9000

12000

15000

Simulation Rounds

Figure 9. Average energy per particle for Is = 0.05 and network size 500 × 500, 1000 × 1000

Figure 10. Number of alive particles for Is = 0.05 and network size 500 × 500, 1000 × 1000

propagate the messages, having its cluster heads transmitting directly to the next level of the hierarchy, while SWPFR uses short-ranged multihop transmissions to deliver the message to the sink. When the network area is fairly small, the transmissions of cluster heads cost less than the multiple multihop transmissions of SW-PFR. However, when the network area increases, the cluster heads have also to increase its transmitting power by an order of R 2 , as the radio range R increases, in contrast to SWPFR, where the multihop transmissions increase linearly. So, for the case of large-area networks, SW-PFR seems more appropriate than H-TEEN. Finally, we examine the way the number of alive particles varies with time. In Figure 10, the number of alive particles is presented for network sizes 500 by 500 m and 1000 by 1000 m, for injection rate Is = 0.05. Note that more particles are dropped in a 1000-by-1000 m network, so that both protocols deal with more or less similar densities. We notice that for the H-TEEN protocol for both network sizes, the number of alive nodes decreases at a constant rate, while for the SW-PFR protocol, there is a sudden decrease in the number of alive particles. This observation depicts the property of the H-TEEN protocol to evenly distribute the energy dissipation to all the particles in the network. On the contrary, the SW-PFR protocol stresses more the particles that are placed closer to the sink, so at a point in time, these particles start to “die” rapidly. On the other hand, in the larger network area (1000 × 1000 m), the particles in H-TEEN protocol “die” more rapidly than those of SW-PFR. This was expected because in H-TEEN, protocol particles are forced to transmit in larger distances than those in SW-PFR, so they consume more energy and “die” faster. We should also notice the same behavior of SW-PFR as in Figure 10—from a point in time and on, particles start to die faster than before, and this is because SW-PFR stresses more the particles that lie

near the sink, forcing them to propagate the majority of the messages that occur in the smart dust network. For the SW-PFR protocol, we adjusted some extra parameters such as the radio transmission range (R) and sleep-awake ratio (en) to achieve good performance for a fixed network density, while the values of the parameters defined for the H-TEEN protocol are set as described in section 7. The choice of values for the SW-PFR protocol is crucial for its performance (thus, SW-PFR seems more “programmable”), while the H-TEEN protocol seems to benefit less from adjusting its parameters. In Figure 11, we give an example of how the success rate of the protocol can vary for different values of the radio range (R) in a 500-by-500 m network with 1500 particles. We notice that there is a strict area of the radio range where the protocol behaves well, while outside this area, its performance drops quickly. Thus, we provide an indication of how to exactly adjust the radio range to get a high success rate (i.e., for a success rate at least 0.8, R should be between 35 and 60 m in the particular setting). Some first simulation results were presented in Nikoletseas et al. [26]. 10. Toward a Hybrid Protocol Overall, for the H-TEEN protocol, the simulation suggests that in large-area networks and when the number of layers in the hierarchy is small, TEEN tends to consume a lot of energy because of long-distance transmissions. On the other hand, when the number of layers increases, the transmissions become shorter, but there is a significant overhead in the setup phase as well as the operation of the network. Thus, H-TEEN exhibits a certain trade-off behavior with respect to the number—let it be c, the numbers of layers used in the hierarchy. Here, we have implemented and evaluated the (rather intermediate) value c = 4. We plan to further investigate, both by analytical and experimental means, this trade-off.




We note that the details of this hybrid protocol’s design (i.e., how long each of the three phases should be, what is the range of the parameters [angle, radius] in the randomization phase, etc.) are quite complicated, so we plan to study them and implement and evaluate the hybrid protocol in a separate work.

1 0.9 0.8

Success Rate

0.7 0.6 0.5 0.4

11. Using Varying Transmission Range

0.3 0.2 0.1 0 0

10

20

30

40

50

60

70

80

90

100

110

Transmission Range

Figure 11. Success rate of the Sleep-Awake Probabilistic Forwarding (SW-PFR) protocol for various values of radio range

Furthermore, the main findings of this work are the following: 1. SW-PFR seems to be more energy efficient in networks covering large geographical areas. In such networks, H-TEEN seems to be expensive. 2. On the other hand, H-TEEN tends to be more efficient in small-area networks and also tends to somehow evenly distribute energy consumption among the particles locating the event. In light of the above two findings, we propose a possible hybrid combination of the protocols in large-area networks. This hybrid protocol should have three distinct phases (we list below phase 2 last in order, to better discuss how this intermediate phase works). Phase 1: Initially, and for a sufficiently large part of the network area, SW-PFR is used to propagate data. Phase 3: When data get appropriately close to the sink, H-TEEN is employed. Phase 2: Since H-TEEN’s energy consumption characteristics heavily depend on the spatial distribution of event occurrence, and since H-TEEN performs better in the case of rather uniform event placement, we introduce an intermediate phase between phase 1 (SW-PFR) and phase 3 (HTEEN). This phase may use a “load-balancing” technique, such as randomization, to rather uniformly further propagate data, to avoid worst-case behavior for the H-TEEN protocol that would arise if data transmissions reach the H-TEEN employment area around the same, more or less, cluster heads. This randomization can be implemented in the following way: the particles in the “phase 2” application area select the next particle to propagate data by randomly and uniformly choosing the angle of transmission and the transmission radius, in some range.

In addition to choosing between long or short transmissions, certain additional trade-offs are introduced by choosing between fixed or varying transmission range. In particular, we wish to focus on the following important properties: (a) Obstacle avoidance: This may be achieved by increasing the transmission range when an obstacle is encountered. (b) Fault tolerance: Increasing range may reach active sensors when the current range does not succeed, either because of faulty or “sleeping” sensors close to the sensor that is currently transmitting or in the case of very low network densities. (c) Network longevity: An interesting aspect of the problem under investigation is the lifetime of particles since it affects the ability of the network to propagate data to the sink because available routes are reduced as more particles consume their energy resources and “die.” Varying the transmission range may bypass the sensors lying close to the sink, which tend to be overused in the case of fixedrange transmissions, since all data pass through them in this case. The same holds also in the case of a geographical concentration of event generation.

Based on the above, we now propose the Variable Transmission Range Protocol (VTRP), where each particle p that has received info(E) from p (via, possibly, other particles) does the following: Phase 1: The search phase. It uses a periodic low-energy broadcast of a beacon to discover a particle nearer to S than itself. Among the particles returned, p selects a unique particle p that is “best” with respect to progress toward the sink. More specifically, the particle pE that, among all particles found, achieves the bigger progress on the p S line should be selected (see Fig. 2). Phase 2: The direct transmission phase. Then, p sends info(E) to p and sends a success message to p (i.e., to the particle that it originally received the information from). Phase 3: The transmission range variation phase. If the search phase fails to discover a particle nearer to S, p enters the transmission range variation phase. More specifically, each particle maintains a local counter τ, with initial value τ = 0. Every time the search phase fails, this counter is increased by 1. Thus, τ is an indication of the number of failures to locate an active particle. Based on τ, the particle modifies its transmission range R according to a change function F(τ). Here, we consider four different functions for varying the transmission range: Volume 81, Number 6


SIMULATION 409


(a) Constant progress. This choice is more suitable in the case where the network is composed of a large number of particles; thus, a small increment of the transmission range will probably suffice to locate an active particle. Based on this assumption, the change function is defined as follows: F (τ) = Rnew = Rinit + c · τ , where c is a constant set to a small value (i.e., c = 10). This is considered as the “basis” VTRP and is denoted as VTRPc . (b) Multiplicative progress. In this case, the transmission range of the particle is increased more drastically. We call this variation of our protocol VTRPm . F (τ) = Rnew = Rinit + Rinit · m · τ , where m is a constant set to a small value (i.e., m = 3). This drastic change has a bigger probability of finding an active particle, but it leads to higher energy consumption. (c) Power progress. In this case, the transmission range of the particle is increased even faster using the following scheme: √ (τ+1)

F (τ) = Rnew = Rinit + Rinit

.

We call this protocol VTRPp . (d) Random progress. When the density of the network is not known in advance, we use randomization to avoid bad behavior due to the worst-case input distributions for each choice above (i.e., small modifications to the transmission range in VTRPc in case of low densities and big modifications resulting from VTRPp in high-particle densities). We call this variation VTRPr and is defined as follows: F(0) = Rinit , F (τ) = F (τ − 1) + Rinit · r , where r ∈ (0, 8], a random value.

The analysis of the properties of the VTRP protocol (efficiency, correctness, fault tolerance, etc.) is a quite complicated task. In Antoniou et al. [28], we presented some preliminary results (based on simulation). We intend to further study these properties (extending the simulation findings and carrying onto a rigorous analysis) in future work. 12. Conclusions We presented in this work two extended versions of the protocols PFR and TEEN (i.e., SW-PFR and H-TEEN) for information propagation in sensor networks. We have implemented the new protocols and conducted an extensive comparative experimental study on networks of large size to validate their performance and investigate their scalability. Our results basically show that the SW-PFR protocol achieves high success rates under specific conditions of network density and radio range (R), and it is resistant to

frequent events and operates efficiently in large-area networks. On the other hand, the H-TEEN protocol achieves high success rates in small-area networks, but there is a deterioration of its performance when the size of the network or the injection rate increases. We plan to study different network shapes, various distributions used to drop the sensors in the area of interest, and the fault-tolerance of the protocols. Finally, we plan to provide performance comparisons with other protocols mentioned in the related work section, as well as implement and evaluate hybrid approaches that combine the SW-PFR and H-TEEN protocols in a parameterized way, exploiting their relative advantages. We note that the latter task (the hybrid protocol) is highly complicated, so we plan to do it in a separate work. 13. Acknowledgments This work has been partially supported by NSERC, OIT/Distinguished ResearcherAward, CFI and Canada Research Chair Programs, and by the IST Programme of the European Union under contract numbers IST-2004-001907 (DELIS) and the Programme PYTHAGORAS under the European Social Fund (ESF) and Operational Program for Educational and Vocational Training II (EPEAEK II). A preliminary version of this work has appeared in Antoniou et al. [28] and Nikoletseas et al. [26]. 14. References [1] Akyildiz, I. F., W. Su, Y. Sankarasubramaniam, and E. Cayirci. 2002. Wireless sensor networks: A survey. Journal of Computer Networks 38: 393-422. [2] Chatzigiannakis, I., T. Dimitriou, M. Mavronicolas, S. Nikoletseas, and P. Spirakis. 2003. A comparative study of protocols for efficient data propagation in smart dust networks. In 9th International Conference on Parallel and Distributed Computing (EUROPAR 2003), Lecture Notes in Computer Science vol. 2790, 1003-16. New York: Springer-Verlag. [3] Chatzigiannakis, I., T. Dimitriou, M. Mavronicolas, S. Nikoletseas, and P. Spirakis. 2003. A comparative study of protocols for efficient data propagation in smart dust networks. Journal of Parallel Processing Letters 13 (4): 615-27. [4] Chatzigiannakis, I., and S. Nikoletseas. 2003. A sleep-awake protocol for information propagation in smart dust networks. In 3rd International Workshop on Mobile, Ad-hoc and Sensor Networks (WMAN 2003), p. 225. [5] Chatzigiannakis, I., S. Nikoletseas, and P. Spirakis. 2002. Smart dust protocols for local detection and propagation. In 2nd ACM International Annual Workshop on Principles of Mobile Computing (POMC 2002), pp. 9-16. [6] Chatzigiannakis, I., S. Nikoletseas, and P. Spirakis. 2005. Efficient and robust protocols for local detection and propagation in smart dust networks. Journal of Mobile Networks and Applications 10 (1): 133-49. [7] Intanagonwiwat, C., R. Govindan, and D. Estrin. 2000. Directed diffusion: A scalable and robust communication paradigm for sensor networks. In 6th ACM/IEEE Annual International Conference on Mobile Computing (MOBICOM 2000), pp. 56-67. [8] Manjeshwar, A., and D. P. Agrawal. 2002. TEEN: A routing protocol for enhanced efficiency in wireless sensor networks. In 2nd International Workshop on Parallel and Distributed Computing Issues in Wireless Networks and Mobile Computing (WPIM 2002), p. 195b.




[9] Chatzigiannakis, I., A. Kinalis, and S. Nikoletseas. 2004. Wireless sensor networks protocols for efficient collision avoidance in multipath data propagation. In ACM Workshop on Performance Evaluation of Wireless Ad Hoc, Sensor, and Ubiquitous Networks (PEWASUN 2004), pp. 8-16. [10] Chatzigiannakis, I., T. Dimitriou, S. Nikoletseas, and P. Spirakis. 2004. A probabilistic forwarding protocol for efficient data propagation in sensor networks. In 5th European Wireless Conference on Mobile and Wireless Systems beyond 3G (EW 2004), pp. 34450. [11] Heinzelman, W. R., J. Kulik, and H. Balakrishnan. 1999. Adaptive protocols for information dissemination in wireless sensor networks. In 5th ACM/IEEE Annual International Conference on Mobile Computing (MOBICOM 1999), pp. 174-85. [12] Intanagonwiwat, C., D. Estrin, R. Govindan, and J. Heidemann. 2001. Impact of network density on data aggregation in wireless sensor networks. Tech. Rep. 01-750, University of Southern California, Computer Science Department. [13] Boukerche, A., and S. Nikoletseas. 2004. Protocols for data propagation in wireless sensor networks: A survey. In Wireless communications systems and networks, edited by M. Guizani. New York: Kluwer Academic. [14] Boukerche, A., and S. Nikoletseas. 2004. Energy efficient algorithms in wireless sensor networks. In Wireless Networks, edited by A. Safwat. New York: Springer-Verlag. In print. [15] Efthymiou, C., J. Rolim, and S. Nikoletseas. 2004. Energy balanced data propagation in wireless sensor networks. In 4th International Workshop on Mobile, Ad-hoc and Sensor Networks (WMAN 2004), p. 225. [16] Spirakis, P. 2004. Algorithmic and foundational aspects of sensor systems. In 1st International Workshop on Algorithmic Aspects of Wireless Sensor Networks (ALGOSENSORS 2004), Lecture Notes in Computer Science vol. 3121, 3-8. New York: Springer-Verlag. [17] Heinzelman, W. R., A. Chandrakasan, and H. Balakrishnan. 2000. Energy-efficient communication protocol for wireless microsensor networks. In 33rd IEEE Hawaii International Conference on System Sciences (HICSS 2000), p. 8020. [18] Rao, A., S. Ratnasamy, C. Papadimitriou, S. Shenker, and I. Stoica. 2003. Geographic routing without location information. In 9th ACM/IEEE Annual International Conference on Mobile Computing (MOBICOM 2003), San Diego, pp. 96-108. [19] Crossbow Technology, Inc. n.d. Mica motes. http://www.xbow.com/ Products/productsdetails.aspx?sid=71 [20] Karp, B. 2000. Geographic routing for wireless networks. Ph.D. diss., Harvard University, Cambridge, MA.

[21] Hollar, S. E. A. 2000. Cots dust. MSc thesis, EngineeringMechanical Engineering, University of California, Berkeley. [22] Berkeley Wireless Research Center, http://bwrc.eecs.berkeley.edu [23] Wireless integrated sensor networks, http:/www.janet.ucla.edu/ WINS/ [24] Ye, W., J. Heidemann, and D. Estrin. 2002. An energy-efficient MAC protocol for wireless sensor networks. In 21st Annual IEEE International Conference on Computer Communications and Networking (INFOCOM 2002), pp. 947-57. [25] Dimitriou, T., and A. Kalis. 2004. Efficient delivery of information in sensor networks using smart antennas. In 1st International Workshop on Algorithmic Aspects of Wireless Sensor Networks (ALGOSENSORS 2004), Lecture Notes in Computer Science vol. 3121, 109-22. New York: Springer-Verlag. [26] Nikoletseas, S., I. Chatzigiannakis, H. Euthimiou, A. Kinalis, T. Antoniou, and G. Mylonas. 2004. Energy efficient protocols for sensing multiple events in smart dust networks. In 37th Annual Simulation Symposium (ANSS 2004), pp. 15-24. [27] Mehlhorn, K., and S. Näher. 1999. LEDA: A platform for combinatorial and geometric computing. Cambridge, UK: Cambridge University Press. [28] Antoniou, T., A. Boukerche, I. Chatzigiannakis, G. Mylonas, and S. Nikoletseas. 2004. A new energy efficient and fault-tolerant protocol for data propagation in smart dust networks using varying transmission range. In 37th Annual Simulation Symposium (ANSS 2004), pp. 43-52.

Azzedine Boukerche is a full professor and holds a Canada Research Chair position in wireless and mobile networking at the University of Ottawa. He is the founding director of PARADISE Research Laboratory at Ottawa University. Ioannis Chatzigiannakis is a researcher at the Computer Technology Institute in Patras, Greece, and a senior researcher in the Department of Computer Engineering and Informatics, University of Patras, Patras, Greece. Sotiris Nikoletseas is a senior researcher at the Computer Technology Institute in Patras, Greece, and a lecturer professor in the Department of Computer Engineering and Informatics, University of Patras, Patras, Greece.

Volume 81, Number 6


SIMULATION 411

Power-Efficient Data Propagation Protocols for Wireless Sensor ...

Power-Efficient Data Propagation Protocols for Wireless Sensor ...

Suggest Documents

Probabilistic Data Propagation in Wireless Sensor

Sink Mobility Protocols for Data Collection in Wireless Sensor ... - CEID

Scalable Data Collection Protocols for Wireless Sensor ... - CiteSeerX

Wireless Sensor Network Protocols - SASE

Routing Protocols for Wireless Sensor Networks

Fundamental Protocols for Wireless Sensor Networks - CiteSeerX

Protocols and Architectures for Wireless Sensor Netwoks

Secure protocols for data propagation and group

Efficient data propagation strategies in wireless sensor ... - CiteSeerX

Wireless Sensor Networks : Technology, Protocols, and Applications

Hierarchical Routing Protocols in Wireless Sensor ...

Routing Protocols in Wireless Sensor Networks - MDPI

energy efficient mac protocols for wireless sensor network

Wireless Sensor Networks Protocols for Efficient ... - Semantic Scholar

A Survey on Transport Protocols for Wireless Multimedia Sensor ...

Analyzing Routing Protocols for Underwater Wireless Sensor ... - IJCNIS

Atypical Hierarchical Routing Protocols for Wireless Sensor Networks ...

A Survey of Transport Protocols for Wireless Sensor ... - CiteSeerX

QoS-aware MAC protocols for wireless sensor networks: A ... - CiteSeerX

A Survey of Transport Layer Protocols for Wireless Sensor Networks

energy efficient mac protocols for wireless sensor ... - Semantic Scholar

Energy Efficient Protocols for Wireless Sensor Networks: A Survey and ...

Energy Efficient Protocols for Wireless Sensor Networks: A ... - CiteSeerX

Multimedia Data Transport for Wireless Sensor Networks