Model to integration of RFID into Wireless Sensor ... - IEEE Xplore

2008 11th IEEE International Conference on Computational Science and Engineering

Model to integration of RFID into Wireless Sensor Network for tracking and monitoring animals Daniel Patrick Pereira1,3 Wanderson Roger Azevedo Dias1 Marcus de Lima Braga1 Raimundo da Silva Barreto1 Carlos Maur´ıcio S. Figueiredo2 Virg´ınia Brilhante1 1

Departamento de Ciência da Computaça˜ o - Universidade Federal do Amazonas - Brazil {dpp, roger, mlb, rbarreto, virginia}@dcc.ufam.edu.br 2

Departamento de Ciência da Computaça˜ o - FUCAPI - Brazil [email protected] 3

Genius Instituto de Tecnologia - Brazil [email protected]

Abstract

larly, in this vision of Pervasive Computing, the technologies of Wireless Sensor Networks (WSN) [1] and RadioFrequency Identification (RFID) [3] play an important role. The former has a great applicability on wide out-door environments such as object tracking and habitat monitoring, while the later is commonly applied to in-door areas such as industrial production processes. However, some applications can be improved by the integration of both technologies.

The pervasive computing concept aims the development of solutions for better perception of events in interest areas. Since 90’s, scientific efforts are being applied on the development of Wireless Sensor Networks (WSNs) technologies and applications such as environmental monitoring, surveillance, and security. More recently, industrial and commercial applications demanded advances on object identification and then RFID technology has emerged. Clearly, the RFID technology, initially applied to indoor solutions, can be combined with common outdoor WSN solutions in order to provide a better monitoring system. This fact is shown in this work by proposing an integrated architecture to be applied on animal monitoring applications.

1. Introduction

In particular, Wireless Sensor Networks (WSNs) form a very active research area. These networks consist of lowcost compact devices with sensing and processing capabilities that are connected among themselves by a wireless medium to perform distributed and cooperative sensing tasks. Additionally to the scale, these sensor networks must be able to operate under very dynamic conditions, must adapt themselves to the environment and need to operate, in most cases, in an unattended mode (i.e., without external interference and control). These features lead these networks to operate under autonomous principles. All these characteristics make the WSN a powerful platform for processing data collected from wide environments.

Recent technological advances have allowed the development of small computational devices with characteristics such as low-cost production, low energy consumption and wireless communication. Thus, in a near future, we can expect to leave in a totally connected environment with constant monitoring and information exchange. Particu-

The development of efficient architectures and algorithms to be applied to WSN solutions depend on the application requirements. A starting point in this task refers to the phenomena to be monitored, which impacts on the sensors to be applied and the data needed by the application. Common monitoring application can use ordinary sensors, such as vibration and temperature, to detect some event oc-

keywords: Wireless Sensor Network, RFID, Adhoc Sensor, RFID Reader, Protocols, Tracking animal, telemetry.

978-0-7695-3193-9/08 $25.00 © 2008 IEEE DOI 10.1109/CSE.2008.25

125

currence, but they cannot identify objects individually. In this intent, research has advanced regarding the integration of WSNs and RFID [5, 13, 2].

2. Related Works

The Radio-Frequency Identification (RFID) is a wireless technology for object or living beings unique identification through tags. In fact, this technology has started to be applied for bar-code substitution in industrial and commercial scenarios. Some advantages are the solution flexibility, the detection of out-of-sight targets and related information storage in the tags. Basically, RFID architectures are comprised of RFID tags and readers. Tags are small chips with radio and antenna for wireless communication. These tags can be passive, which depends on the reader interrogation and which uses the received energy to respond back, or active, which is more independent due to am independent power source. The passive tags have the advantage of longer lifetimes but lower communication ranges, while the active tags can reach longer distances and store dynamic data. RFID reader are responsible for identification acquiring and further storage and processing of collected data from tags.

Tracking animal is not innovation, it is very important to know about the environment where the animals live to preservation and maintenance the natural resources. It is meaningful to know where the animals shelter, feed, disperse and others routines for one better exploitation of the area without harming the animals. The literature has distinct solutions to tracking and monitoring animals in its habitat. The Argos System [4] was developed to tracking animals in its habitat in large areas. The system relies on satellites, which are at around 850 km above the Earth surface. It is able to geometrically establish the location of the transmitter, velocity, light intensity, activity daily pattern, cardiac rhythm and others. The accuracy of the determined location depends on various factors: vegetation cover, land use, rainy day, and slope were different. Another method to tracking animal is through cellular technology. Nowadays, the infrastructure allow tracking animals throughout a large area. The advantages of cellular technology area the steadily increasing coverage of the network around the world and the continual advances in speed of data transfer through cellular networks. Disadvantages are the technological challenges of reducing size, the challenge of reducing power consumption, and the use of different cellular architectures in different countries [6]. Sikka et al. [8] describe a large heterogeneous sensor network on a working Australia farm to explore sensor network applications to tracking and monitoring animals. Because of the wireless sensor constraints, they were inspired to develop their own devices with improved radio range, solar power capacity, mechanical and electrical robustness, and with unique combinations os sensors [7]. A couple of wireless sensor was augmented with actuator to stimulate the animals such as vibration, sound and also a lowlevel electric sock. The data being broadcast in the WSN is picked up by the relay and sent to the server, where the users can see the animal information.

2.1. Tracking animals

An application of interest refers to animal monitoring. This is the case of Sauim Project (Technological Development - CNPq 554071/2006-1). This project aims the habitat and area occupation monitoring of the extinction-threatened Saguinos-Bicolor monkeys in Manaus (capital of the Amazon state) urban areas. In such applications, individual animal identification is primordial and RFID technologies can be very useful. In fact, existing applications use radiofrequency telemetry to detect the animals. However, data collection is performed by local presence of the observer who is equipped with a radio signal detector. This characteristics turns the monitoring process very complex and expensive. In this context, we propose to integrate both technologies by extending the radio frequency identification to an open area with wireless sensor network. The motivation of the application is related to animal detection and identification in their habitat. However, according to animal behavior, we have integrated different architectures.

2.2. Integrated technologies Lei Zhang et al. [13] describe a deep analysis of RFID and WSN, three forms of new system architecture that combines the two technologies. The first one, mix RFID tags and sensor nodes in the same environment. A station gather information from tags and sensor nodes then transmit it to local host PC or remote LAN. That is can be a disadvantage in an open environment if needing the electric net for the station. The second architecture in [13] propose a new smart node, which makes use of kinds of sensors, to detect in-

This work is organized as follows. Section 2 discuss the related work regarding animal monitoring applications and WSN-RFID integration. Section 3 presents the proposed architectures and their design aspects. Section 4 formally describes the architecture operation and their validation. Section 5 presents final considerations and points to future work. Finally, Section 6 presents the acknowledgements.

126

3.1. Long range tags

terested physical scenario, reading RFID tags, and radio transceiver which transporting sensed data. Although this architecture is applied in a indoor environment, the requirement for new architecture served as basis for our two architecture. The last one proposed in [13] is replace the RFID active and semi-active tags to Mica mote [9]. The active tag is similar to the Mica mote, but they are not exactly sensor network nodes because they communicate in centralized mode and can’t cooperate with each other through formed ad-hoc network. Nevertheless, replace all active RFID tags to sensor nodes can be expensive in a large area. Therefore, in our architectures considered viable distribution among active RFID tags and sensor nodes able to read RFID tags, this architecture is described in Section 3.1. McKelvin et al. [5] describe the system architecture that employs an automate method for recording, retrieving, and tracking inventory to improve the efficiency and accuracy in inventory management and tracking. The commercial radio frequency identification system used in the prototype was the ALR9780 [11]. During the experiments the maximum reading range in open space for passive tags used was 7 feet. Chen et al. [2] describe a application framework for group tour guiding services based on RFIDs and wireless sensor network. The system provide tracking the locations leader, maintaining the guiding path to each leader, showing guiding paths for lost members, and helping leaders call their members. The architecture includes: sensor node is attached to a direction board for displaying simple guiding direction, each group member simply carries a ticket with a passive RFID tag, and some nodes are designed as help centers, each connected to a laptop and a RFID reader, where the member group can get information. The prototype used sensor nodes MICAz [9] and SQLServer. The papers exposed here, employed in an open environment could become impracticable its applications. The system architecture in our work differ for considering applications in open environment and for considering the behavior of the mobile object.

The first system architecture is better suited when there is not a central mobile object among the objects. Therefore, the mobile objects disperse and become impracticable to identify the central mobile object and to subordinate objects that will be close. In this way, we have a low density where the monitored objects remain away from others and they become not appropriate for practical use of passive RFID tags. Nevertheless, the active RFID tags can be read at distances of one hundred meters or more and they may have other sensors (temperature, pressure, among others). This architecture is composed for active RFID tags and sensor nodes with the ability to read the content of active RFID tags. Each sensor node will be fixed around the environment where the objects will be tracking and monitoring. In this system only active RFID tags are mobiles. The sensor node with RFID Reader contains three parts: sensing part which makes use of kinds of sensor (temperature, pressure), reading part which gathers data from RFID tags, and radio transceiver part which sends data to the sink node. Hence, the new sensor node has to manage this new device. The Figure1 illustrate the architecture.

Figure 1. Long range tags. Advantage in this architecture is in substitute sensor node to active RFID tags. They can incorporate sensors into their design, and they are cheaper than sensor nodes. Therefore, monitoring an area with active RFID tags could be cheaper. It is important to point out that the specific area to be monitored can expand easily. For instance, we can provide a solution if the animals being monitoring are moving to another area. Therefore, we should distribute new sensor nodes in order to expand our monitored area, while in the RFID system we should connect new cables to RFID readers. We describe the architecture components as follow:

3. System architecture Integration of RFID into Wireless Sensor Network has the advantage of gathering the best features among technologies by making a heterogenous network. From this spectrum of best features, a cheaper network can be developed in the proposal environment. This section present two system architectures to tracking and monitoring animals in its habitat. The main component to integrate both technologies is the special node. Beyond the common functionalities such as self-organization and transfer data, the sensor node now is able to gather data from RFID tags.

Sensor node with RFID Reader (fixed) - Special Node This sensor node has a RFID Reader that allow

127

gathering data from tags and transmit through WSN to the sink node; Active RFID tag (mobile) WSN usually fixes sensor node in the mobile object to track and monitor. However, in this system architecture, only RFID tags are fixed in the mobile objects. In addition to tracking and monitoring, some active RFID tags can collect the temperature data and transmit to reader in real-time. The Sensor node with RFID Reader (fixed) has two read distances. One of them is the read distance to RFID Reader to gather data from tags and the other one responsible to form ad-hoc network. The data gathered are send to a specific sink node. Moreover, the sensors operate in a selfconfiguring, self-network, self-diagnosing, self-healing, decentralized manner that maintains the best connectivity and communicate messages via multi-hop.

Figure 2. Short range tags. passive tags in a group, if it fail, all passive RFID tags will be alone until find another special node. We describe the architecture components as follow: Sensor node with RFID Reader (mobile) - Special node This sensor node should be fixed in the central mobile objects;

3.2. Short range tags Different from the first architecture, this architecture is better used where is possible to recognize a central mobile object among the others. Based on this context, we through passive RFID tags can distribute them to take advantage in this situation. When we recognize a central mobile object, that object is a powerful candidate to have a special node with the ability to read RFID tags. On the other hand, the subordinate objects will have passive RFID tags. Therefore, the special node is responsible to gather data from passive RFID tags and send to a sensor node (fixed) in the net. The special node has to be closer from the passive RFID tag, because it can be read only at very short distances, typically a few feet at most. Nevertheless, an environment with a high density where the objects usually stay in groups, we can distribute especial nodes among them to gather data. This architecture become perfect where the animals move in group and we find the leader to get the special node (e.g., female). Similar the architecture in the Section 3.1, this architecture is composed for sensor node with the ability to read RFID tags (special nodes), sensor nodes (fixed) and passive RFID tags. The suitable difference between this architecture and the previous one is that sensor nodes (fixed) don’t have the ability to read RFID tags, while the special nodes carries through this task. The Figure 2 illustrate in detail the architecture. Advantage in this architecture is monitoring with lesser sensor node than the first architecture, substituting some sensor node to passive RFID tags. However, its performance is not inferior, because among passive RFID tags has a special sensor to gathering data and sending to others nodes in the net. Despite the special node read data from

Sensor node (fixed) This sensor nodes are distributed in the area; Passive RFID tag (mobile) This passive RFID tags are fixed in the subordinate objects.

4. System Description The architectures presented in Section 3 expose two scenery. In this sceneries we have a common sensor node with RFID reader. The special node in the long range architecture read active RFID tag, while the other one passive RFID tag. This section present how the special node manage that Reader. A sensor node has typically the following states: sensing, compute, routing and communication. However, when incorporate a RFID Reader, emerges a new state relative the reader. It is important observer that the typical states are not influenced. In the Figure 3 is presented a modeling through pushdown automata (PDA). Each state has the following functionalities. IDLE The sensor node is with all operations turn-off. The sensor node remain in stand-by until the Timer wakeup; RFID The RFID Reader perform a read in the environment, whether find a tag moves to PROC state or comes back to IDLE state; PROC Compute data from RFID Reader and radio sensor node;

128

• E, is a finite set of states; E = {IDLE, SEN SIN G, RF ID, P ROC, T XRX, ROU T E} • Σ, a finite set called the alphabet; Σ = {i(T IM E IN ), o(T IM E OU T ), e(READ), p(P ROCESS), t(T X), r(RX), s(SET T IN G), n(RESET T IN G), u(ROU T IN G)} • Γ, a finite set called the stack alphabet, where P represent the collected data Γ = {P } • I, a finite set of start states; I = {IDLE} • F , a finite set of final states; F = {IDLE} • δ, transition function, is a partial function of E × (Σ {λ}) × (Γ {λ}) to D, where D is constituted of the finite subgroups of E × Γ∗

Figure 3. States in the special node

Now we present the partial function as follow: TXRX This state is divided in two stages: (i) transmit data from RFID Reader or radio sensor node; (ii) receive data from adjacent sensor nodes to transmit data to sink node;

δ(IDLE, time in, λ) = {[RF ID, λ], [T XRX, λ], [SEN SIN G, λ]} δ(IDLE, setting, λ) = {[ROU T E, λ]} δ(RF ID, read, λ) = {[RF ID, P ]} δ(RF ID, process, P ) = {[P roc, P ]} δ(RF ID, time out, λ) = {[IDLE, λ]} δ(P ROC, process, P ) = {[P roc, P ]} δ(P ROC, tx, P ) = {[T XRX, P ]} δ(SEN SIN G, timeo ut, λ) = {[IDLE, λ]} δ(SEN SIN G, read, λ) = {[SEN SIN G, P ]} δ(SEN SIN G, process, P ) = {[P ROC, P ]} δ(T XRX, rx, λ) = {[T XRX, P ]} δ(T XRX, λ, P ) = {[T XRX, λ]} δ(T XRX, tx, P ) = {[T XRX, λ]} δ(T XRX, process, P ) = {[P roc, P ]} δ(T XRX, resetting, λ) = {[Route, λ]} δ(T XRX, time out, λ) = {[IDLE, λ]} δ(T XRX, λ, λ) = {[IDLE, λ]} δ(ROU T E, routing, λ) = {[Route, λ]} δ(ROU T E, rx, P ) = {[T XRX, P ]} δ(ROU T E, tx, P ) = {[T XRX, P ]} δ(ROU T E, time out, λ) = {[IDLE, λ]} δ(ROU T E, setting, λ) = {[IDLE, λ]}

ROUTE The sensor node perform the self-configuring and self-network. This task is performed when the sensor node is on at the first time or when a transmission or reception failure; SENSING The sensor node perform a sensing in the environment, where could be monitoring: temperature, pressure, altitude, luminosity and others. Due to memory necessity to store the gathered data and consequently transmit them, we chose the Nondeterministic pushdown automata (NPDA). A λ argument specifies that the value of the component should be neither consulted nor acted upon by transition. When λ occurs as an argument in the stack position of the transition function, the transition is applicable whenever the current state and input symbol match those in the transition regardless of the status of stack. The stack top may contain any symbol or the stack may be empty. The symbol λ may also occur in the new stack position of transition. The execution such a transition does not push a symbol onto the stack. If the input position is λ, the transition does not process an input symbol. Thus, transition pops and pushes the stack symbol without altering the state or the input. A formal definition of NPDA is a sextuple describe as follow [12] [10]:

4.1. Automata validation Automata validation can be perform using the machine configuration as follow. The machine configuration processes the input in a left-to-right manner. The definition is: [e, w, p], where e, is the start state; w, is the word to be

M = (E, Σ, Γ, δ, I, F ), where:

129

computed and p, is the stack. A computation that accepts a string is called successful, it means the sensor node move from IDLE state to another state (RFID, SENSING, PROC, TXRX, ROUTE) and comeback to start state, IDLE. A word is accepted, when:

B) The word ”IEPTPTTT” is not accepted, because there are more transmission than reception. Our model doesn’t allow transmit more information than receive, that is a energy constraint. In this moment the stack was entire processed, but the work wasn’t. [IDLE, IEP T P T T T, λ] [RF ID, EP T P T T T, λ] [RF ID, P T P T T T, P ] [P roc, T P T T T, P ] [T XRX, P T T T, P ] [T XRX, P T T T, λ] [P roc, T T T, λ]

• a word is entire processed; • halts in a final state; • the stack must be empty. 1. Accepted words A) The word ”IEEPPTT” is computed and accepted, because the RFID state read two tags, after that move to PROCESS state and finally move to TXRX state. In this moment the word was entire processed, the stack is empty and it was in the final state. [IDLE, IEEP P T T, λ] [RF ID, EEP P T T, λ] [RF ID, EP P T T, P ] [RF ID, P P T T, P P ] [P roc, P T T, P P ] [P roc, T T, P P ] [T XRX, T, P P ] [T XRX, λ, P ] [T XRX, λ, λ] [IDLE, λ, λ]

5. Conclusion and Future Work This paper presents a proposal of two heterogeneous architectures for integration between technologies WSN and RFID. For this study was presented with a model of Nondeterministic pushdown automata (NPDA), which made their validations, through its states and transitions. This model represents the set of states sensor node and add on RFID transceiver. Then were defined: the language, alphabets and their evidence of recognition of words through its configuration instantly. This work is inserted in a larger project aimed at the implementation of Wireless Networks Sensors and georeference database management system (DBMS) in monitoring and specimens of terrestrial fauna. Our object of study are the populations of species of primates Saguinus bicolor, popularly known as sauim-de-coleira. The experiments will be directly on the habitat of animals (Amazonian forest), where we can test the proposed architectures in a scenario with great density of trees and a warm and humid climate, thus providing a major challenge to technology proposal. The next step of this work includes the proposal of models that can better represent the behaviors of the system presented in this paper. We also aim to develop extension to the current model, want to add communicating among automata, so that they can better represent the system as a whole and from the scene to check the properties of the system: liveness, boundedness, reachability among others.

B) The word ”IRRTT” is computed and accepted, because the TXRX state gathering data from radio sensor, after that in the same state transmit the data. This word present how a sensor node receive data from adjacent node and transmit to another one. [IDLE, IRRT T, λ] [T XRX, RRT T, λ] [T XRX, RT T, P ] [T XRX, T T, P P ] [T XRX, T, P ] [T XRX, λ, λ] [IDLE, λ, λ] 2. Unaccepted words A) The word ”IP” is not accepted, because the stack is empty, and the next symbol is to move to the PROCESS state. Our model doesn’t allow move to PROCESS state when the stack is empty, it means that doesn’t have data to compute or transmit. [IDLE, IP, λ] [RF ID, P, λ]

6. Acknowledgements The authors would like to thank the support received from the Brazilian Council for Scientific and Technological Development (CNPq) by its partial financial support through project 554071/2006-1.

130

References [1] K. Chakrabarty and S. S. Iyengar. Scalable Infrastructure forDistributed Sensor Networks. 2005. [2] P. Y. Chen, W. T. Chen, C. H. Wu, Y.-C. Tseng, and C.-F. Huang. A group tour guide system with rfids and wireless sensor networks. ISPN’07 - Association for Computing Machinery, pages 561–562, Abril 2007. [3] V. D. Hunt, A. Puglia, and M. Puglia. RFID - A guide to Radio Frequency Identification. 2007. [4] J. E. Mantovani, J. E. Santos, and J. S. R. Pires. Rastreamento via satélite de animais: Performance do sistema argos em diferentes situaço˜ es. Anais XI SBSR, Belo Horizonte, Brasil, 05-10 abril 2003, INPE, pages 589–594, 2004. [5] M. L. McKelvin, M. L. Williams, and N. M. Berry. Integrated radio frequency identification and wireless sensor network architecture for automated inventory management and tracking applications. TAPIA’05 - Association for Computing Machinery, pages 44–47, Outubro 2005. [6] W. D. Robinson, T. Fiez, H. Liu, K. Mayaram, and Z. Wang. Tracking migratory animals with cellular technology. 2000. [7] P. Sikka, P. Corke, and L. Overs. Wireless sensor devices for animal tracking and control. In First IEEE Workshop on Embedded Networked Sensors in 29th Conference on Local Computer Networks, pages 446–454, 2004. [8] P. Sikka, P. Corke, P. Valenci, C. Crossman, D. Swain, and G. Bishop-Hurley. Wireless adhoc sensor and actuator network on the farm. IPSN’06, pages 492–499, Abril 2006. [9] C. Solutions. MICA2 Mote. Dispon´ıvel em http://blog.xbow.com/xblog/mica2 mote/index.html. ´ Ultima visita no dia 20 de Novembro, 2007. [10] T. A. Sudkamp. Languagens and Machines. Pearson, 2006. [11] A. technology incorporated. Alien technology alr9780 RFID Reader. Dispon´ıvel em http://www.alientechnology.com. ´ Ultima visita no dia 20 de Novembro, 2007. [12] N. Vieira. Introduça˜ o aos Fundamentos da Computaça˜ o Linguagens e Máquinas. 2006. [13] L. Zhang and Z. Wang. Integration of rfid into wireless sensor networks: Architectures, opportunities and challenging problems. IEEE 2006, Agosto 2006.

131