Repositorio Institucional de Documentos

Politecnico di Torino WIRELESS SENSORS NETWORKS Final Project Fco. Javier Osta Ardevines 08/03/2011 Fco. Javier Osta Wireless Sensors Networks 2 INDICE Introduction. ........................................................................................................ 3 What is a wireless sensor network (WSN)?........................................................ 4 Integration of wireless sensor networks in environmental monitoring cyber infrastructure. ................................................................................................ 16 Wireless Sensor Networks for Fire Rescue Applications .................................. 29 Monitoring structures ........................................................................................ 39 Automotive ....................................................................................................... 40 Agriculture and ranching. ................................................................................. 41 Bibliography. ..................................................................................................... 42 Conclusion........................................................................................................ 43 Fco. Javier Osta Wireless Sensors Networks 3 Introduction Today the sensors can be found in large number of systems and electronic devices. Most of these sensors lack the ability to process and analyze the data they detected, limiting itself to function like a transducer that performs the measurement of one or more environment variables and sends that information to a central processor. However, the researchers predict the arrival of a new generation of sensors, equipped with its own intelligence, able to organize themselves and to interface wirelessly with other fellows. Arises so-called Wireles Sensor Networks, WSN, consisting of ad-hoc macronetworks consist of many individual sensors that exchange information with each other wirelessly and through a protocol of preestablished communication. In recent years, several research laboratories, and especially multinationals such as Intel, have bet heavily on this technology. It predicts that these networks will lead a technological revolution similar to the internet. There is talk of global surveillance network on the planet, capable of recording the habits of the people, make a monitoring about people and goods, monitor traffic, etc. Even if it will have to wait a few more years, it have arisen multiple initiatives and research projects interest and practical applicability. This paper will discuss some of these experiments that demonstrate the potential of this technology. Fco. Javier Osta Wireless Sensors Networks 4 What is a wireless sensor network (WSN)? Wireless sensor networks (WSN), are based on devices low cost and consumption (nodes) that are able to obtain information from their environment, process locally, and communicate via wireless links to a central coordinating node. The nodes act as elements of the communications infrastructure to forward messages received from nodes farther toward the focal point. The wireless sensor network consists of numerous distributed devices spatially, using sensor to monitor various conditions at different points between including temperature, sound, vibration, pressure, motion or pollutants. Sensors can be fixed or mobile. The devices are autonomous units that consist of a microcontroller, power energy (usually a battery), a radio transceiver (RF) and a sensor element. Due to the limitations of battery life, the nodes are built keeping in mind the energy conservation, and generally spend much time at all 'sleeping' (sleep) of low power consumption. The WSN are capable of self-restoration, ie failure of a node, the Web will new ways to route data packets. In this way, the network will survive as a whole even if individual nodes lose power or be destroyed. The capacity of auto diagnosis, self-configuration, self-organization, self-restoration and repair, are properties that have been developed for these networks to solve problems that were not possible with that they were not possible with the other devices. Elements of a wireless sensor network (WSN) Two approaches have been taken. The first to integrate all components (sensor, radios and microcontrollers) on a single plate initiated by Moteiv Corporation. They have a lower production cost and are more robust in harsh environments or effects. The second approach started by Crossbow Technology Inc. is to develop plate with transceivers that can be connected to the microcontroller board. This approach is more flexible. The nodes typically consist of a plate of sensor or data acquisition and a “mote or speck” (plate processor and transmit/ receive radio). These sensors can be communicated with a gateway, which has Fco. Javier Osta Wireless Sensors Networks 5 capacity to communicate with other computers and other networks (LAN, WLAN, WPAN…) and Internet. In connection with the software they need, there are specific operating systems like TinyOS for embedded systems. Routing systems and security are fundamental structure of a wireless sensor network. Research data system The sensors are different in nature and technology. They take environmental information and converted into electrical signals. In the market-sensing plates measure many different, such as barometric pressure sensors, GPS, light, measure solar radiation, soil moisture, air humidity, temperature, sound, wind speed and many more. Examples: MTS300/310, sensor capable of detecting acceleration, light, microphone, sound, magnetometer, temperature, and the MTS420 sensor can detect temperature, humidity, luminosity, is sensitive to the light, contains a barometer. Fig.1. MTS300/310 To control environments in real time systems, smart transceivers meet the IEEE 1451. Sensor standardization: A family of proposed standards. To define an interface for sensors and actuators, which is independent of protocols communication network used. The transceivers are equipped with sensors and micro actuators controller that provides them with “local intelligence” and communication skills. Designed as interface between 802.11 (WiFi), 802.15.4 (Bluetooth) and 802.15.5 (ZigBee). The figure shows the architecture of the IEEE 1451 standars. - The point-to-point interface that meets the IEEE Standard 1451.2 - The distributed multi-drop interface that meets the IEEE standard 1451.3 - The Analogic+digital interface that meets the IEEE standard 1451.4 - The wireless interface that meets the IEEE standard 1451.5 (WIFI, Bluetooth, ZigBee) - The CanOpen interface that meets the IEEE standard 1451.6 Fco. Javier Osta Wireless Sensors Networks 6 Motes Motes equipped with processing and communication to the sensor node. Processors radio take sensor data through its data ports, and send the information to the station base. Typical components are: Batteries A CPU. Flash Memory Separate data memory programs A plate of sensors: light, humidity, pressure, etc. Radio to communicate with other motes. ADC: analog-digital converter They are resistant to weather and inhospitable terrain and able to run an application. Gateway They permit the interconnection between the sensor network and a TCP/IP. Example: MIB600Ethernet. (TCP/IP) Network gateway which in turn serves as a programmer with Ethernet connection that we can connect from a PC. Base Station Data collector based on a common PC or embedded system. Fco. Javier Osta Wireless Sensors Networks 7 Parameters of a WSN The core values that characterize a wireless sensor network are: • Lifetime • Network Coverage • Cost and ease of installation • Response time • Accuracy and frequency of measurements • Security • The core values that characterize the sensor node are: • Flexibility • Robustness • Security • Communication skills • Computing Capacity • Synchronization Facility • Size and cost • Energy cost Architectures The modular design is necessary in order to reuse the items. However, being involves modular design limitations and has to be careful to ensure that the interfaces between modules, hardware and software requirements are general enough to allow portability expected. There are two architectures: Centralized architecture in which nodes communicate only with the gateway and the others it is distributed architecture in which sensor nodes communicate only with other sensors in the same range. Another aspect is the distributed computing where nodes cooperate and implement algorithms distributed to obtain a single global measure that is responsible for coordinating node communicate to the base station. The nodes not only capture the information but also use their computing capacity to developed measures. Standard and proprietary wireless technologies for sensors Wireless The most popular wireless standards are for the IEE 802.11b LAN (“WiFi”) for PAN, the IEEE 802.15.1 (Bluetooth IEEE 2002) and IEEE 802.15.4(ZigBee IEEE 2003). Fco. Javier Osta Wireless Sensors Networks 8 They use the ISM bands (Instrumentation, Scientific and Medical radiobands), 902-928 MHz (USA), 868 to 870 MHz (Europe), 433.05-434.79 MHz (USA and Europe) and from 314 to 316 MHz (Japan) and band of 2,400 GHz – 2.4835 GHz (universally accepted). The current WSN are based on IEE 802.15.4 ZigBee 802.15.4, that which is more general WISA. Including multi-hop, which implies in what the messages uses different hops wave. Nodes are not assigned specific time intervals, but must compete to access the channel. This allows access more users to the wireless medium, but introduces uncertainty in the system, as the delay and consumption energy increase when a node is waiting for turn. Furthermore, intermediate nodes know the when they can be applied to route packets for others. It is therefore advisable to have intermediate nodes. ZigBee is ideal for active monitoring. When the number of nodes to interconnect is very high, the networks solution is more than one level with different technologies (hybrid networks). Also it developed proprietary technologies (Crossbow Technology and Freescale Semiconductor). We understand that future developments should be based on standards. Topologies Besides the classic mesh network topology of WSN, there are others topologies. I.e. Topology star networks, wireless nodes communicate with a gateway device a bridge of communication with a wired network. An emerging compromise is to have common WSN router devices that communicate with the gateway. The sensors only need to communicate point to point routers and can therefore be kept simple and power while improving the range and redundancy of the network itself. Fco. Javier Osta Wireless Sensors Networks 9 Routing The nodes have no knowledge of the topology of the network must discover it. The basic idea is that when a new node to appear on a network, announces its presence and listen broadcast of its neighbors. The node is informed about the new nodes to its scope and way of moving through them, in turn, can announce to the other nodes that can be accessed from a time, each node knows which nodes have around and one or more ways to achieve them. Routing algorithms in wireless sensor networks must meet the following standards: • Maintain a routing table reasonably small. • Choose the best route to a given destination (either the faster, more reliable better capacity or the route of least cost). • Keep the table regularly to update the fall of nodes, their change of position or appearance. • Require a small number of messages and time to converge Routings Models There are several types of routing protocols. Direct Broadcast Protocol (one-hop): This is the most simple and direct communication is. All nodes in the network transmitted to the base station. It is an expensive model in terms of energy consumption and infeasible because nodes have limited transmission range. Its transmissions can’t always catch the base station, have a maximum distance of radio, so the communication directs is not good solution for wireless networks. Multi-hop model (multi-hops): In this model, a node transmits to the base station to resend the data to one of its neighbors Wireless Sensor Network which is closer to the base station, while he sent to another node closer to you get to the pot base. Then the data travels from source to destination hop by hop from one node to another until it reaches the destination. Given the limitations of the sensors is a viable approach. A large number of protocols use this model, including all Multihop Tmote Sky and Telos: multihop LQI, Miniroute. Schematic model based on clusters: Some protocols use optimization techniques to improve the effectiveness of the previous model. One of them is the aggregation of data used in all routing Fco. Javier Osta Wireless Sensors Networks 16 Logical data layer Postgre SQL Database (DB) Server collects the data from the distributed monitoring stations. The central data collection (CDC) Server acts as an intermediate component of different physical layer devices and support data validation required by the DB Server. A Sensor Web Enablement (SWE) compliant data repository is installed to enable data exchange, accepting data from both internal DB Server and external sources trough the Open Geographic Consortium (OGC) web services. Web presentation layer and user layer The web presentation layer consists of a web portal where you can find: K-12 Education Outreach, Real time environmental monitoring, historical data download and modeling analysis synthesis. This can be implemented with a Sensor Observation Services (SOS) at this layer to facilitate data exchange. The SOS web service will be published to a catalog service in the OGC SWE framework to make it publicly accessible on Ineternet. Fig. 1. System architecture of integrated environmental monitoring system. System architecture of sensor nodes The WSN hardware platform used in this current design is the IRIS mote from Crossbow Technology. The Base Station (BS) node transmits aggregated Fco. Javier Osta Wireless Sensors Networks 17 data to the remote field gateway (RFG) Server through serial port (RS232). To accomplish this without human intervention, these devices are equipped with a solar cells and rechargeable batteries. In figure 2 we can see the functional block diagram of sensor node. In general, in environmental monitoring applications, every sensor node periodically carries out three main tasks, including data generation through sensing, data processing, and data reporting through multihop wireless communications. To perform the data generation tasks, sensor reading are collected periodically, which necessitates global time synchronization in the network. After in the data processing task, sensor nodes calibrate, aggregate, summarize, and compress the data. Finally, in the data reporting task data are transmitted to the BS node trough multihop wireless communications. Fig.2. Functional block diagram of sensor node. Networking protocols for multihop data collection Energy efficiency is one of the major design considerations in environmental monitoring sensor networks. Therefore is important to choose a good design of networking protocols. In wireless sensor networks, Media Access Control (MAC) layer are broadly categorized in 2 groups, Schedule-based and contention-based. A contention-based protocol is highly autonomous but relatively energyinefficient while schedule-based protocol may eliminate overhearing and collision among neighboring nodes to store high energy efficiency, but it may suffer co-channel interference from other types of devices operating in the same frequency band. In this system they developed a hybrid MAC layer protocol that integrates Carrier Sense Multiple Access (CSMA) and duty-cycle scheduling to achieve high energy efficiency to support long-term, low-rate and large-scale sensor Fco. Javier Osta Wireless Sensors Networks 18 networks applications. It uses a distributed duty-cycle scheduling algorithm to coordinate sensor nodes’ sleeping. Our protocol is divided in superframes and these in time slots. Fig.3. Time slot structure of a super frame in the hybrid MAC protocol. The super frame starts with a signaling slot where the nodes actively broadcast and received packets. Sensor nodes compete for Time Division Multiplexing Access (TDMA) slots and exchange control information with neighbors during this period. In many-to-one sensor data collection networks, all of the data packets are routed from child nodes to their parents. With the TDMA slots we can guarantee that, there will be only one parentchild pair active in any two-hop neighborhood. This system is mainly designed to provide a collision-free channel forwarding such data packet. However other neighboring nodes within the vicinity of the parent-child pair can save energy by avoiding overhearing unnecessary packets. This is the biggest difference between our duty cycle scheduling protocol and most of the existing protocols. Other protocol like B-MAC can save energy because it does not need time synchronization, however, this time synchronization can’t be deleted in the system without requiring other components as well. The flooding time synchronization protocol (FTSP) proposed in time-stamps synchronization messages at the MAC layer, with removes the nondeterministic delay at both sender and receiver caused by uncertain processing time in the operating system for context switches. This protocol can synchronize multiple receivers with a single broadcast message. In a tree-structuctured network, if every node synchronizes to its parents, ultimately all of the nodes in the network could synchronize to the root to achieve global time synchronization. In this experimentally study, it is observed that FTSP can achieve less than 1 ms timing errors in a three-hop network when the power management functionality is turned off. To save energy in low-power-modes, existing timer drivers pull down clock frequency at which repeat timers request when there are no active one-shot timers. Unfortunately, switching between high and low frequencies results in inconsistent time stamps. Therefore, had been developed a new two-layer time driver to replace the original drivers which employed two individual hardware clock to tackle the two types of timer separately. A high speed clock is used to drive one-shot timers. On the contrary, it allows speed and thus low power clocks runs continuously to support the repeat timers. Fco. Javier Osta Wireless Sensors Networks 19 Although the data flow is unidirectional in our tree structure, the nature of the wireless structure makes difficult to maintain a reliable multihop routing hierarchy. Here we implemented a link quality estimator based on exponentially weighted moving average (EWMA) estimation method. The multihop routing protocol makes use of the link quality estimator to maintain a reliable routing topology. Wireless Telemetry system Wireless telemetry system hardware design. To integrate a variety of devices in field is implement a RFG Server using a compact, rugged, ultra-low-power single-board computer SBC. The SBC provides a standard set of on-board peripherals and a software power consumption control for on-board peripherals. This saves energy by controlling between sleep and actives modes. The devices deployed in the field are commonly equipped with a RS232 serial port, including data loggers, wireless modem, and the WSN BS node. Thus, with five serial ports onboard, SBC is well suited to served as a gateway server Wireless communication from field to CDC Server is implemented by using a General packet radio service (GPRS) modem. It is a packet-oriented mobile data service. These GPRS use Point-to-point (PPP) protocol. To be energyefficient, the wireless modem is powered off during the system’s sleep period. The target supporting the system with a fully charged battery for at least a week without recharging, for this; has been implemented a power budget analysis system. Thus it can use solar energy with a large solar panel and a lead-acid rechargeable battery. Remote data collection services Our SBC deployed the full-feature Debian GNU/Linux. It is convenient to develop remote data collection services by taking advantage of the software packages that Debian provides. The RFG Server wakes up periodically to carry out data collection services. The wireless modem is powered on at the same time as RFG server. Then, several independent data collection processes are started to poll data from the WSN BS node and dataloger trough RS232 ports. The data collected by the RFG Server is inserted into a local file system or directly CDC server through the wireless modem. The CDC Server then synchronizes its database to the RFG database. Fco. Javier Osta Wireless Sensors Networks 20 The time of synchronization depends on the amount of data and traffic load condition in the network. We implement a simple duty cycle negotiation protocol between the RFG Server and the CDC Server to enhance energy efficiency of solar-powered remote monitoring system. As fig 4. shows, the RFG server and the CDC Server are protected from potential networks failures by timers T1 and T2, respectively; that is data collection process is terminated when the timers expired. Fig.4. Duty cycle negotiation protocol between the RFG Server and the CDC Server. Various system status data are also collected in the same way as sensor data to enable remote monitoring and management of the monitoring system deployed in the field. Sensor data management There are two ways to bring data to CDC server, or the CDC; it periodically connects to the data source and pulls the data, or opens the port and wait for the rod to be pushed from the data source. Or received data are stored in local file system. Later are tested according to a set of predefined validation rules, and sent to the PostgreSQL DB Server. Data handler may also require the RFG server to recollected and retransmit missing data packet. Sensor data base design is driven by the emphasis on system extensibility because of the need to handle a large volume of data collected from heterogeneous sources in long-term operations. All sensor information is contained in one relation or table, whereas each observation is stored in a separate relation. An example of such database schema is shown in Fig 5 Fco. Javier Osta Wireless Sensors Networks 21 Fig.5. Illustration of the database schema. Sensor data visualization and dissemination The research was developed a dedicated web portal, Texas Environmental Observatory (TEO) Online, for the sensor data visualization and dissemination. To take the most of the flexible Google Maps APIs to associate sensor with their geographical locations intuitively as shown in Fig 6. Such an interfaces provides direct visualization of special distribution of sensors data. Fig.6. A snapshot of the TEO web interface for data visualization TEO Online portal provides a variety of ways to explore sensor observation data. For example, data can be browsed under different overlapping categories, such as type of observations (UV, soil moisture, etc).Several predefined functions allow the analysis of data statistics such as average and maximum values. Fco. Javier Osta Wireless Sensors Networks 22 Data interoperability rises as an important issue with the wide application of sensor networks and different field. Data exchange can be performed through Extensible Markup Language (XML) data exchange, Really Simple Syndication (RSS) feed, and Sensor Observation Services (SOS), as show in fig 7 Fig 7. Sensor data dissemination and exchange frame work. The sensor data interoperability is achieved by a dedicated backend SWE data repository, as well as front-end RSS feed and web services building upon the repository. RSS feed items and links are stored in a RSS table in the repository, while the live data is encapsulated in the RSS page by the web layer RSS class functions. Thus the general users can subscribe and watch the data through his browser or can request to receive the most recent sensor observation data automatically. The SWE is a new standard that specifies interoperable interfaces and metadata encodings to enable real-time integration of heterogeneous sensor data. Major encoding standards include SensorML that describes sensor system information and O&M (Observation and Measurement) that encodes actual live data. The interoperability interface standards include SOS (sensor Observation Services), SAS (Sensor Alert Services), SPS (Sensor Planning Services), and WNS (Web Notification Services). These modules collaborate together and can be used to customize various web-based or desktop-based applications for environmental scientists. The database is syndicated with a sensor table, which allows the formation of the SWE standard. Thus SWE is synchronized with the DB server. During this synchronization, SWE synchronizer can retrieve external data. The data in the SWE repository is converted to the SensorML and O&M Fco. Javier Osta Wireless Sensors Networks 23 format by an SWE handler, which then feed the information to the upper-layer web services. Deployment and field testing results Deployment of sensor networks in the field The weather station in Denton, Texas has nine years to control the weather, with a wired system, but in Match 2008 was extended with a modem wireless system of 8 motes as shown in Fig.1. It way to 16 motes along a cross-sectional transect as shown in fig.8. Support a long-term, each sensor node collects data every 10 minutes from soil moisture sensor and onboard temperature and relative humidity sensor. This cross network topology is perfect to check the moisture and a vegetation changes as a function of the height above the river and soil type. Fig. 8. Sensor deployment topology in the field as shown on the Google Mapbased Teo Online web portal. The motes are installed in weatherproof boxes and on top of metal poles to avoid flooding water. Before we put the motes we’ll have to check to measure the one-hop radio communication range. We observed that with a maximal transmission power of 3dBm, IRIS motes are able to transmit on overage 30 m with 95% packet reception rate (PRR) and 50 m with 80% PRR. Thus, we deployed motes with a maximum one-hop distance of about 30 m. The TEO Online web portal has been operational since March 2008 with most of the basic web services implemented. System performance characterization and field testing results Duty cycling provides an effective way to achieve energy efficiency. The RFG wakes up for 90 s every 10 min. The Table 1 shows the current draw and duty cycle devices deployed inside the GBC station, powered by solar panel with a peak current of 960 mA and a 12 Ah lead-acid rechargeable battery. The battery supports 7 days without recharding. In general, the capacity of a solar panel should be at least 10 times the average power consumption. Fco. Javier Osta Wireless Sensors Networks 24 Tab.1 The current draw and duty cycle of the devices deployed inside the GBC station With the duty cycle scheduling algorithm, motes are only active during a few TDMA slots to report and relay sensor data and a signalind slot to synchronize time, manage neighbor list, and update parent information. Motes remain in the sleep mode and consume much less power than in the active mode. Fig 9 shows a measurement result of current draw and duty cycle, capture with an oscilloscope. Fig 9 Measurement of the current draw and duty cycle of a mote. The spender an average current of 24 mA for 16 us in active mode and 2.73 mA for 90 us in sleep mode. Thus, two fully charged 2500 mAh NiMh batteries with a self-discharging rate of30% can sustain a mote for about 4 weeks without recharging. The solar cell used at the power motes is able to provide 100 mA peak current at 3 V, about 36 times the average current draw of the load. The variation of weather condition was also captured by sensor on motes as shown in Fig 10. The difference in data among the three soil moisture sensors reveals spatial variation characteristics of the soil moisture condition n that area. Mote 3 is installed in a transparent box to put a solar cell inside the box, Mote 4 is installed in a non-transparent box with the solar panel. Fco. Javier Osta Wireless Sensors Networks 25 Fig 10. Sample sensor data collected in field testing. Table 2 shows a few statistics of the transect sensor network status data that has been collected from field tests over a one-month period. From the hop count measurements (between each node and BS) we can clearly observe to the tree structure of the multihop sensor network experiences dynamic variations and the sensor network is able to reorganize autonomously in the face of environmental and network changes. Tab 2. Statistics of sensor network status data collected from field tests. Mobile sensor network programming system (MSNPS) In traditional sensor networks, the sensor are deployed manually the manual installation and upgrading of physical infrastructures will not scale if we want to build the sensor networks in a larger scale. Fco. Javier Osta Wireless Sensors Networks 32 In WSN, this software will run either on the powerful laptop acting as sink deployed on the incident commander’s car or on the machines located in the fire department. Web-enabled service and integration. Not only does the incident commander sitting near the fire field need the information collected by the WSN, but also the officers sitting in the fire department, which is located far away from the fire field. In a big city like New York, there maybe several fires happened at the same time, so the officers in the fire department need to make schedule on how to control these fires effectively and concurrently. Thus the real-time information from different fire fields is needed by the fire department, and the optimized schedule will be made based on this global information. Webbased service is one of the most convenient ways to provide these information to these officers. By doing so, the real-time information from each fire field is wrapped as a web-enabled service, accessible through regular Web browsers. Because the fire department is located far away to the fire field, the traditional Internet will act as the bridge to connect the fire field and fire department. First, the webenabled service should provide the information that the fire department interested via the network, e.g., it continuously reports the live situation of each fire field. Second, it will automatically generate some events to the fire department to ask aid when more firefighters or vehicles are needed. Moreover, the collected data can be stored and analyzed later to find some good rescue models to support the future fire fight. 2 FireNet Architecture Having known the requirements of the fire rescue application, we are now in a position to propose FireNet, a wireless sensor network architecture for the fire rescue application. Compared with traditional wired sensor systems, the novelty of WSN lies in that the sensors deployed in the sensor field can self-organize into a connected ad-hoc network via wireless communication. We call the specific wireless sensor networks deployed for the fire rescue application FireNet. The architecture of the FireNet is shown in Figure 1. In the FireNet, the vehicles and firefighters are equipped with sensors which form a self-organized heterogenous wireless sensor network. In the incident commander’s vehicle, a powerful laptop connected with a powerful sensor acts as the gateway of WSN. The ladder vehicle and the two engine vehicles are loaded with sensors having GPS equipped. These vehicles can act as the landmark for the whole WSN because they will have relatively stable location, i.e., other sensors can calculate their location based on the location of these vehicles. Each firefighter in the sensor field carries a sensor, such as MICA2 or MICAz from Crossbow attached with available sensor board which can sense interested parameters, acting as the active badge for each firefighter. The active badge records all the information expected by the incident commander and fire department (for later Fco. Javier Osta Wireless Sensors Networks 33 analysis), such as the firefighter information, the fire field environment information and emergent events, as listed in the Figure 1. The role of each active badge (i.e., sensor) has two-fold: sensing the data and forwarding the packets. We can also install different program on the sensors for the firefighters with different specialty, thus these sensors can have different functionality. Furthermore, the sink of FireNet, located in the incident commander’s vehicle, is connected to the fire department headquarter via traditional Internet so that the WSN and the fire department can communicate each other and keep contact. After the fire fight team starts their work, the sensors attached to vehicles and firefighters are self-organized into a WSN via wireless communication. Then, the sensors start to operate according to their pre-installed program. For example, the sensors attached to firefighters will collect the information of firefighters, sample the environment parameters, and generate the vital events happened in the fire field, as shown in the right part of Figure 1. These data will be reported to the sink by the multi-hop routing protocol and further delivered to the fire department via Internet. Then, both the incident commander and the officers in the fire department have the accountability and real-time information from the fire field, which is abstracted and presented by the powerful preinstalled software in the sink or fire department. By doing so, the whole fire field is monitored and the status of each firefighter is clear to the incident commander and fire department. Based on this, the incident commander and fire department will make optimized fire schedule according to the suggestion of the intelligent software. For instance, in the case of fire rescue in September 11, 2001, some firefighters were covered by the dust or buried in some part of fire field of ground zero, which is a very dangerous situation. If WSN is used, it will be much easy to detect such a situation and the location of firefighters in anger, a rescue can be arranged immediately to help them out of the disaster. Moreover, the schedule commands from the incident commander can be sent to firefighters via FireNet to instruct the firefighters to move or take some other ctions as well. In addition to the communication ability, each sensor has its CPU and local memory so that it can do some calculation such as data aggregation and store the data for a period of time. With the development of the semiconductor technology, the computational capability and the memory size have been extended a lot in the last several years. We believe that wireless sensors will be more powerful in the near future so that more information can be stored and more efficient realtime decision making algorithms can be employed. In summary, as described above, wireless sensor network is a very promising technology in the application of fire rescue. It is very useful in terms of not only monitoring the whole fire field including the firefighters and environment Fco. Javier Osta Wireless Sensors Networks 34 information, but also sending schedule information and commands to the firefighters in a more intelligent way. Fig 1. The architecture of FireNET 3 Research Challenges As described above, the fire rescue application has its special characteristics, such as high mobility and real-time, and specific requirements as described in Section 1.2. Previous protocols are not satisfactory and need to be revisited for this application. New protocols and support from hardware as well as software are needed to build the sensor system. Next, we narrate them in detail. Protocol Challenges First, four protocol challenges, namely real-time selforganization, faulttolerant routing, service differentiation, and real-time and mobile localization, need to be addressed for this application. Real-time self-organization. FireNet is a pure ad-hoc wireless network, thus the self-organization of FireNet is very crucial. Unlike other WSN applications like environment monitoring, the fire rescue application of WSN has two specific characteristics, the high dynamics of sensors due to the high mobility of firefighters in the fire field, and the realtime requirements of data collection. These two inherent features make the self-organization of FireNet become more challenge. Moreover, we expect that the wireless communication between sensors in a fire field has extremely high loss rate due to the harsh environment. Thus, it is not easy to achieve the goal of real-time self- Fco. Javier Osta Wireless Sensors Networks 35 organization. A good selforganization protocol should be proposed to support the automatic re-configuration of the sensor network, which should have the capability to make all the sensors attached to fire-fighters and other utilities in the sensor field connect or reconnect to the sensor network in a limited period of time after they newly join the network, lose the connection and reconnect, or change their location. In such a protocol, several functions are highly wanted. First, the relationship between each firefighter and his/her associated vehicle should be captured automatically; Second, in a short period of time after firefighters start the fire fight process, these firefighters and the utilities are all self-organized into a connected adhoc WSN; Third, during the procedure of firefighter moving, the sensor network should keep connection in spite of the absence of several firefighters from their original area, and the moving sensors should re-connect to WSN in a limited time at the new location. The challenge of this protocol lies in how to satisfy the real-time requirement, and always keep the connectivity and coverage of the whole WSN. Fault tolerant routing. As argued above, one significant characteristic of the fire rescue application is the high dynamics of network resulting from the mobility of firefighters or failure of sensors. Although the real-time selforganization protocol takes care of the re-configuration of WSN, it cannot assure the successful delivery of messages to the sink. However, the fire rescue application has high requirement on accuracy and real-time to the collected data. A fault tolerant 4routing protocol is essential to guarantee the successful delivery of the data and events in the case that the mobility and absence of some sensors as well as the unreliable wireless communication will cause a high package lost rate. The fault tolerant routing protocol should deal with both the mobility of sensors and the failure of sensors happened in the sensor field in a timely fashion. Previous research on the routing protocol in WNS usually assume that sensors are located in fixed position, neglecting the high dynamics of the network. Thus, novel routing protocols are expected to handle the problem of sensor mobility and real-time delivery. One possible solution is to build multi-path routing protocols in the real time, which can dramatically improve the probability of successful delivery in a high dynamic environment. Service differentiation. The main function of the WSN in this application is to collect different information from the fired field, e.g., the environment parameters such as temperature, humidity, wind speed, and chemical and biological leak, the emergent events such as the dramatic changing of the environment parameters and the death of firefighters, and the status of the firefighters such as location, and the workload of firefighter. Although all these data need to be monitored and collected, their importance is different and then these parameters should be treated in a different way in the system. Thus, we argue that an efficient service differentiation scheme, including timeliness, Fco. Javier Osta Wireless Sensors Networks 36 storage requirement, and processing priority, should be provided to implement multiple quality-ofservice (QoS) in such an environment. As a matter of fact, how to arrange the collection and delivery of these data in an effective and efficient way is a challenge because sensors have limited low bandwidth and share the wireless communication medium. A good service differentiation scheme is needed to schedule the data collection and transmission through the FireNet architecture. One possible direction is exploiting the inherent consistency requirements of those parameters. For example, an event of firefighter death is much more important than any other parameters in the network, so it will be set the highest priority and reported to the sink as soon as possible. In addition, the tradeoff between the energy efficient and data consistency including data accuracy and timeliness should also be examined in the protocol. Real-time and mobile localization. As we described in Section 1.2, the location, especially the real-time location, of firefighters in a fire scene is a very important and valuable piece of information. Given this piece of information, the incident commander could have a clear view of the distribution of deployed firefighters, and make real-time decisions. Moreover, location information is very useful for other protocols in WSN as well. Most research topics in WSN, e.g., fault tolerant routing, aggregation, event detection and tracking, and so on, directly or indirectly lend on accurate location information provided by the underlying localization service. Admittedly, location in the fire rescue application is not a trivial task given the fact that firefighters are moving very fast and randomly in a real rescue operation as well as the inherent ad-hoc feature of FireNet. Localization in WSN has been extensively studied in the literature [11]; however, as a reality check, few of practical localization algorithms are deployed in the real applications, and practical localization, especially mobile localization, is still a challenge from the perspective of real deployment. Intuitively, Global Positioning System (GPS) is a pretty good positioning system at outdoors, however, it is not accurate enough for indoor tracking. Moreover, most of existing localization solutions did not take the mobility into consideration, i.e., they always assume the location of sensors is static, which is obviously not the case in FireNet. The few mobile localization algorithms such as do not consider the moving speed and the dynamics of the system. Therefore, we argue that the localization protocol for a WSN in fire rescue application needs to address the following issues: mobility, heterogeneity, locality, robustness, feasibility, and accuracy, each of which is described as follows. • Mobility The fast movement of firefighters makes the localization a big challenge in a timely fashion. Fco. Javier Osta Wireless Sensors Networks 37 • Heterogeneity Due to the heterogeneity of the sensors used in FireNet, the localization algorithm should take these diverse platforms into consideration, e.g., the computing devices on some vehicles/equipments, such as laptops or tablets, could be integrated with GPS support which provides some reference points for further location resolving, while on the other side, the sensors carried by firefighters would be very simple and possess only limited computing resource and energy support. • Locality Each sensor has limited computing power, memory, and communication range, thus only a completely localized algorithm is applicable in FireNet, where each sensor interacts with its neighbors only. • Robustness Sensors in FireNet are working in a very harsh and highly failure-prone environment. Robustness is a key requirement of the localization algorithm, i.e., the failure of some sensors should not affect the calculation of other sensors location. • Feasibility The localization algorithm has to be practical enough so that its computing cost could be affordable by the limited hardware/software supporting of those tiny sensors. • Accuracy The real scheduling by the incident commander and fire department is based on the accurate location of firefighters. Accurate positioning in a static environment is already nontrivial, it becomes more challenge in such a highly dynamic environment. Software Challenges As analyzed in Section 1.2, software components are required in the sink and fire department. These software components are used to analyze the collected data and make good schedule suggestion. Because the sink and fire department have different requirements to the software, two types of software should be designed for them respectively. The software for the sink should has the following functions. First, it needs to analyze the collected data and abstract the useful information from the data such as the accountability information of each type of firefighters. Second, it should make some schedule suggestions based on the collected data and some preset rules. Third, it needs to present the abstracted information and the schedule suggestion to the user of this software. The design of this software includes two parts, the graphic user interface (GUI) part and the intelligent decision making and scheduling part. All the information needed by the incident commander and fire department will be presented in the GUI interface. For instance, the accurate real-time location of firefighters will be shown in a map of the fire field. And the accountability information as well as the status of firefighters will be provided by the GUI. The scheduling part has the function of Fco. Javier Osta Wireless Sensors Networks 38 analyzing data and generating schedule suggestion. A well designed rule set is essential to provide good scheduling schemes. Some technologies from artificial intelligence community may be useful to design the rule set through self-learning. Besides, to support the remote access of these real-time information, we need to provide a web-enabled interface, which needs to take the security into consideration because of the confidential information of firefighters. Another software component is needed at the fire department side. We can use the similar software as in the sink. However, the officer in the fire department may not care such detailed information as a incident commander. For them probably the general information is enough. Meanwhile, the decision rules at the fire department are different from those of a incident commander. Comparing with the incident commanders, the officers in the fire department headquarter need to re-schedule the firefighter squads as a whole across multiple fire fields. Hardware Challenges The fire rescue application also posts some specific requirements on the hardware support for sensors. Normal commodity-of-the-shelf wireless sensors, e.g., motes from Crossbow, are applicable in this application in terms of functionality. However, to our knowledge, they haven’t considered the extremely harsh environment like fire field, which is a very important issue to the success of the WSN deployment. The FireNet architecture proposed for the fire rescue application is running in a dangerous environment which normally has fire, water, dust, and extremely high temperature. Moreover, the sensors are shaking and moving dramatically with the moving of firefighters. If the sensors are exposed to such an environment without well protection, it will stop working immediately. So we argue that two issues should be considered to address the hardware challenge. One is packaging, and we need to come out an ideal packaging scheme for sensors to make sure they are water-proof, fireproof, and vibration-proof. The other is the internal design of wireless sensors. Some fault tolerance features should be taken into consideration during the design procedure. Fco. Javier Osta Wireless Sensors Networks 39 Monitoring structures An application of WSN is monitoring of structures has been mainly conducted in the United States and Canada. Here are estimated to have about 25 trillion of dollars invested in civil structures and investment so they want to take control of the structures built. The technology used is called SHM Structural Health Monitoring of English and works with the identification and monitoring of bizarre behavior. These may be a flaw in a structure such as bridges, buildings or other structures. Vibration control of bridges Thanks to new microelectromechanical systems (MEMS) and we can have acceleration sensors that can measure wireless media; we can control the vibrations in buildings. The University of Berkeley, California, conducted a study on a pedestrian bridge over the highway I-80 in Berkeley. This application gives information on the state of life of the structure as well as events that occur during monitoring it, such as an earthquake. Fig 1. Location of sensor on the bridge. Health monitoring of civil infrastructure (Golden Gate Bridge) Another example of simple harmonic motion (SHM) application is held in the Golden Gate Bridge in Sa Francisco, Where the nodes are designed to monitor in various parts of the structure vibrations that have occurred, whether by the passage of vehicles or or by atmospheric conditions. Fco. Javier Osta Wireless Sensors Networks 40 Sixty-four nodes were implemented in a system of 46 jumps measured the vibration environments with an accuracy of 30 g. Vibration environments were sampled at 1kHz with an exposure time less than 10 s. Fig.2 Location of sensor on the bridge Golden Gate Automotive With the characteristics of WSN, cars may soon be available to talk to each other and with infrastructure in roads and highways. The sensors can be applied to the wheels of the vehicle to assist the driver and warn of possible warning messages. For example, during an emergency braking, emergency message from the car that stops may be sent to all cars nearby so that they take action with regard to this event. Another interesting application is the collection of traffic data in real time. The information that a car may have come in the opposite direction can be valuable. A vehicle can receive information from other fixed sensor information. All this information can be passed from vehicle to avoid congestion and plan alternate routes. Ford Capstone Project. With 70% of Ford motor company vehicles being remodeled is important to keep current and learn the new features and releases. Ford is looking at new ways to innovate when it comes to collect measurements from their vehicles. Sensor networks are also increasing their intelligence and data can be obtained not only limited to light, temperature, humidity and movement. A team from the University of Michigan, along with Ford, is designed and developed a way to identify how many times a vehicle is inspected by a potential buyer. This could include how many times a door is opened, the hood was raised or luggage was inspected. The events studied could be extended to where a person took a seat and how much time remains in it. Fco. Javier Osta Wireless Sensors Networks 41 The environment, by definition, cam be very dynamic, making the ability to connect in a mesh network critical to long term. Agriculture and ranching. Farming and ranching are two areas where this technology can be important since a situation where monitoring outdoor conditions that help to improve production and quality in agricultural production or control of cattle on the move can be very difficult with traditional technology. Camalie vineyards Camalie vineyards, in the United States have one of the most advanced systems for measuring soil moisture. They use wireless technology developed by Berkeley University in collaboration with Intel and marketed by Crossbow. The application is optimized irrigation, reducing water consumption, energy used in pumping and improving the quality of the grapes. It provides monitoring of the irrigation system, showing faults that may have a substantial impact in the long-term. Fig 1Devices implanted in vineyards. Once implemented the system found a significant increase in production and a decrease in energy consumption when using the facilities. Routine monitoring in pigs Dr. Philippe Bonnet University of Copenhagen developed an application designed to facilitate the work of veterinarians, controlling several variables in the daily routine of pigs on a farm.