Conventional fiber-Bragg-grating (FBG) sensor networks, in which sensors are simply arrayed in series or in parallel, are not designed for network survivability. This becomes especially worrisome when they are embedded in harsh environments such as dams, aircraft, or submarines, where replacing components such as faulty fibers, nodes, or sensors can be very costly if not impossible.
Recent efforts directed at designing new sensor-network models that incorporate a self-healing function may represent a sensible approach to address the problem.1–4 However, available models still have serious limitations: they are difficult to scale up and also require further enhancement since they only tolerate one or two failures in network grid links.
We designed a new FBG-based sensor network model, called the FBG Sector Sensor Network, with the goal of improving both network survivability and expansibility. Figure 1 illustrates our basic design. It features a main node (RN1), a series of remote nodes (RN2…RNn), and sensor and transmission sections. The main node may consist of a coupler or switches. Switches offer less loss and can multiplex more sensors. This is because sensors with the same peak wavelength can be placed in different sensor sections. Couplers are mainly used in systems that require real-time monitoring. With this option, the source light is first transmitted through the main node, then sent to different sensor parts by different transmission paths.
Figure 1. Basic sector sensor network model. RN1 is the main node. Other R boxes are remote nodes.
Figure 2. Four types of remote nodes, which reroute signals if links fail.
Figure 3. Sensor network functionality with different types of remote nodes.
To incorporate a self-healing function in a real-time measurement sensor network, we designed four different types of remote nodes for different network performance requirements (see Figure 2). The first type of remote node is an optical switch that simply provides another route when a link fails, as shown in Figure 3(a). The second type of remote node functions as the first when only one link fails. However, if more than one link fails, as shown in Figure 3(b), the performance of the network remains unaffected because the switch can select more than one sensor path, unlike the first type of remote node that can only select one. The third and fourth types of remote node designs can also be used in real-time measurement systems, with further survivability enhancement since they allow more network link failures, as shown in Figure 3(c) and (d).
To further improve network survivability, we arrayed the sensors in rings fitted with entrance nodes to route light into the rings (see Figure 4). We designed two types of entrance nodes: the first allowing light to be transmitted either clockwise or counterclockwise, and the second dividing the light into two parts, each transmitting half the sensor input.
Figure 4. Sensor network with a ring configuration.
Figure 5. Large-scale sensor network.
Our basic sector network model has the additional advantage of being easily expanded to a larger scale. Figure 5 shows an expanded sensor network divided into three tiers, each of which contains a basic sector sub-network. Light transmitted from RN1 passes node RN4, enters main node RN5 of the second level, and proceeds to RN8 before entering main node RN9 of the third level. The main nodes must be redesigned to balance the power reflected among the sensors of different levels.
Increasing the lifetime of a sensor network requires addressing the survivability problem. Our new model addresses both network survivability and expansibility issues. Our next step will be to design a medium-access control (MAC) protocol adapted to our model.