Research into fiber Bragg grating (FBG) sensor networks, in which sensors are arranged simply in series or in parallel, is traditionally concerned very little about network survivability. In practice, this is a problem because the fibers, nodes, and sensors can fail when these networks are placed in harsh environments such as dams, aircraft, and submarines: additionally, replacing broken elements can be costly or even impossible for these applications. Designing sensor network models that incorporate self-healing functions may be a good way to solve this problem.
Researchers1–4 have presented new network models that offer self-healing functions before. However, these are limited in that they are difficult to scale and because the survivability of the sensor network currently only tolerates one or two links failing in particular locations: this is not ideal for practical use. To enhance network performance, therefore, we have designed a new sensor-network model—called the FBG Sector Sensor Network5—the survivability and expansibility of which is carefully considered.
The basic FBG Sector Sensor Network, as shown in Figure 1, contains of the main node (RN1), remote nodes RN2∼RNn, and both sensing and transmission modules (parts). The main node can either consist of a coupler or switches, where switches suffer less loss and multiplex more sensors since sensors with the same peak wavelength can be placed in many different sensor modules. The coupler, on the other hand, is mainly used in systems that require real-time monitoring. The light from the source, transmitting first to the main node, is sent to different sensor modules via different transmitters.
Figure 1. The basic sector sensor-network model.
To add self-healing to the network, we designed four different kinds of remote nodes that provide different levels of network performance, as shown in Figure 2. The first remote node is an optical switch, which provides an alternative route when a link fails, as shown in Figure 3(a). Unfortunately, the performance of the network using this first design will be affected if two links fail, since the optical switch will have to choose one sensor module to communicate with at a time, which is unacceptable if the sensor network is required real-time monitoring, as shown in Figure 3(b). Using a second remote-node design, made of optical switch and coupler, can solve this problem. This design which acts in the same way as the first if one link fails, but the coupler can also divide the light into two and send each beam, simultaneously, to different sensor modules.
Figure 2. Remote nodes of four different kinds can enhance survivability.
Figure 3. Here, the sensor network is shown using different remote nodes for different survivability scenarios.
The third and the fourth remote nodes are also used for real-time measurement, but with further-enhanced survivability that works with more failed links, as shown in Figure 3 (c, d).
In order to enhance the survivability of the sensor network, we arranged the sensors in a ring, with light injected via an entrance node. We have designed two such nodes, which also determine the route through the ring: the light propagates either clockwise or anticlockwise using the first, while the second splits the light and sends half to each of the two neighboring sensor modules.
Figure 4. Sensor network using a ring as its sensor module.
The network can easily be extended to work on a large scale based on the basic sector model. As shown in Figure 5, the sensor network is divided into three levels, each of which contains a basic sector sub-network. The light transmitted by RN1 enters the main node of the second level (RN5) via RN4, and then enters the main node of the third level (RN9) via RN8. The main node has to be reconfigured to balance the power reflected between the different network levels.
Figure 5. A large-scale sensor network.
To increase the life of the sensor network, its survivability must be taken into account. Our model has been designed with careful consideration of both survivability and scalability. Our next step will be to design a medium-access control protocol (MAC) version.
Pu Wei, Xiaohan Sun
Electronic Science and Engineering Department
Pu Wei received his bachelors degree in automation from the Southeast University in 2004. He has been working towards his PhD in photonics and optical communication there since 2006. His study is focused on optical fiber sensor networks.
Xiaohan Sun is a professor of electronic engineering. She was a visiting professor at the Massachusetts Institute of Technology Research Lab of Electronics from 2002 to 2004. Her current research interests are high-speed optical communications systems and broadband optical networks including: optical pulse propagation in wavelength-division multiplexing systems influenced by nonlinear effects, polarisation mode dispersion, crosstalk and so on; management and control for optical networks; and design and measurement for GaAs/InP bas.