Análisis de Objetivo Específico N° 03

Advances in micro-electro-mechanical systems (MEMS) technology, wireless communications, and digital electronics have enabled the rapid development of the wireless sensor technology since the late twentieth century. We place this part here rather than in Section 3.2 because the wireless sensor is actually neither one kind of pure sensing technology nor just a new transmission method, but a new system that can carry out many tasks certainly including SHM. A wireless sensor network can comprise all of the components in a wire-based SHM system described previously such as SS, DATS, DPCS, DMS, and SES, whereas it has its unique characteristics as compared with the wire-based SHM systems.

3.7.1 Overview of Wireless Sensors

Development of wireless sensors is due to the fact that a robust system may require a dense network of sensors throughout the system. Traditional sensing systems usually attempt to develop more and more accurate sensors of limited quantity at optimized locations. A bio-system, however, usually comprises a huge number of distributed sensors each with limited functions. This philosophy inspires researchers to develop a network of low-cost, small size, large quantity sensors. In addition, the traditional sensing systems are usually wire-based which have high installation costs. Maintenance of such a monitoring system at a reliable operating level under the adverse environment for a long period of time is very difficult. Experience in monitoring civil structures shows that communication wires are more vulnerable to the environment than sensors themselves. The wireless transmission provides a more flexible communication manner and sensors can be deployed and scalable easily.

With the support of the US Defense Advanced Research Projects Agency, researchers at the University of California at Berkeley have developed the open platform, well known as Berkeley Mote, or Smart Dust, whose ultimate goal is to create a fully autonomous system within a cubic millimetre volume (Kahn et al., 1999). Such a system may comprise hundreds or thousands of sensor nodes each costing as little as about one US dollar.

Berkeley Mote is the first open hardware/software research platform, which allows users to customize hardware/software for a particular application. Its first generation is COTS Dust (Hollar, 2000), followed by Rene developed in 2000.

The third generation of Mote, the Mica, was released in 2001. Subsequent improvements to the Mica platform resulted in Mica2, Mica2Dot, and MicaZ. Another commonly used wireless sensor unit is the Intel Mote platform Imote (Kling, 2003) and Imote 2 (Kling, 2005).

Berkeley Mote, Intel Mote, and quite a few others have been used for general purposes in military, environment, health, home, and other commercial areas (Akyildiz et al., 2002; Yick et al., 2008). These systems have been customized for SHM applications (Kurata et al., 2003; Ruiz-Sandoval et al., 2006; Rice and Spencer, 2008). A wireless monitoring was implemented in the Jindo Bridge, Korea in June 2009, in which 110 wireless nodes were used. In the structural discipline, researchers from Stanford University have developed their own wireless sensor unit for SHM (Lynch et al., 2001; 2002).

3.7.2 Basic Architectures and Features of Wireless Sensors

A wireless sensor node usually consists of four basic components as shown in Figure 3.3: a sensing unit, a processing unit, a transceiver unit, and a power unit (Akyildiz et al., 2002). The components are carefully selected to meet the specified functions and keep a total low cost.

Figure 3.3 Structure of a wireless sensor node

The processing unit is a micro-processor (or micro-controller), which controls the sensing, data processing, computation, and communication with other sensor nodes or the central station. The on-board processor makes the wireless sensor node intelligent, which differs from a traditional sensor. The micro- processor has a small storage that stores internal programs and processed data.

The sensing unit is usually composed of a few sensors and the ADC. The analogue signals collected by the sensors are converted to digital signals by the ADC, and then sent to the processing unit. It is noted that the ADC in most general wireless sensor nodes is of only 8 or 10 bits. This is insufficient for vibration monitoring. In the customized Imote 2 (Rice and Spencer, 2008), a 16-bits ADC is embedded. The wireless sensor unit developed in Stanford University (Lynch et

al., 2001) has a 16-bits ADC as well.

A transceiver unit connects the node to the network. The transmission distance of most wireless nodes is about 50 m to 500 m in outdoor environment.

Micro-processor Transducers ADC Power unit Transceiver Processor core Storage

Consequently for large civil structures, this transmission range requires the sensor nodes communicate with the peers and send the data to the base station over the network. The wireless network has three kinds (Swartz and Lynch, 2009): star, peer-to-peer, and multi-tier, as shown in Figure 3.4. In the figure, the sensor nodes include generic nodes and gateway nodes. A gateway node, like the substation in the wired systems, gathers data from the adjacent generic nodes and transmits them to the base station. Most of the smart sensors to date adopt radio frequency for the wireless communication.

Figure 3.4 Wireless network topology

The power unit is an important component in wireless sensor node. Currently most of available smart sensors rely on the battery power supply, which has finite capacity and finite life. There have been several attempts to harvest energy at sensor nodes locally, for example, solar cell, wind turbine, mechanical vibration, fuel cells, and mobile supplier. Solar cell is the current mature technique and was used in wireless monitoring of the Jindo Bridge.

In wireless sensor nodes, communication consumes much more power than other operations including sensing and processing. Therefore, collected raw data are processed within the sensing unit to reduce the amount of the raw data transmitted wirelessly over the network. This also takes advantage of the computational characteristics of the processor board. Accordingly this distributed computation and monitoring make the wireless monitoring different from the tethered monitoring using the traditional wired system.

To facilitate this distributed monitoring, the micro-processor has two types of software: one is the operating system and the other is the engineering algorithms. The operating system controls the nodes and provides device drivers. One popular operating system is TinyOS (http://www.tinyos.net), an open-source operating system designed by the University of California at Berkeley. Both Berkeley Mote and Intel Mote run the TinyOS operating system.

Currently algorithms for distributed monitoring are relatively scarce and simple, mainly in modal analysis. The complicated monitoring algorithms used in the centralized monitoring usually need a large amount of memory, heavy computation, and data from multiple sensors. Consequently transplanting the available monitoring algorithms from the wired monitoring system directly is not feasible. It is imperative to develop appropriate algorithms for this distributed monitoring.

(a) Star (b) Peer-to-peer (c) Multi-pier

3.7.3 Challenges in Wireless Monitoring

Although wireless sensors and networks have been developed rapidly, at the moment the wireless monitoring is not mature for continuous health monitoring of large civil structures. Traditional wire-based systems still dominate practical SHM projects and wireless sensor nodes are mainly for research purposes or supplementary to the wired systems. Nevertheless, wireless sensors and networks might be a future direction for SHM. At present main challenges in wireless sensors are power supply, and communication bandwidth and range. As mentioned in the last section, lack of mature distributed monitoring algorithms is another big issue.

For many civil structures, AC power outlets are not available adjacent to the sensor nodes. (Even if AC power outlets are available, practitioners prefer to adopt the wired DAQ system adjacent to the power outlets and transmit the collected data to the base station via the wireless communication. In any case, wireless communication is just one communication method rather than wireless sensor networks.) For a battery-powered monitoring system, power consumption is a critical issue to maintain the operation of the sensor nodes in the long term. Currently used battery-powered sensor nodes can only operate for hours in full working state and weeks in standby state (or sleep mode). The power restraint requires the sensor components should be energy efficient. However, lower power consumption often comes with reduced functions such as lower resolution, shortened communication range, and reduced speed (Swartz and Lynch, 2009).

The majority of wireless sensor nodes operate with the unlicensed industrial, scientific, and medical radio band, in which the output power is limited, to 1 W, for example in the United States. The limited radio band limits the amount of data that can be reliably transmitted within the network during a given time period. In addition, the limitation of the output power restricts the effective communication range of the sensor nodes. For a large civil structure, a few sensor nodes may be insufficient and the sensor network should be designed carefully.

3.8 THE STRUCTURAL HEALTH MONITORING SYSTEM OF THE