Undersea Sensor Networks for Intrusion Detection:
Foundations and Practice

Challenges and design issues

We consider sensor dispersal patterns that result in a random distribution in a 3-dimensional (3D) ocean space of interest. For example, sensors may be dispersed from an aircraft or artillery ordinance, as shown in Figure 1. Previous work on sensor networks for intrusion detection has primarily focused on 2-dimensional (2D) terrestrial strips. In 2D strip sensor networks, a chain of sensors from one end of the strip to the other end with overlapping sensing areas of adjacent sensors can detect any intruders regardless which paths they choose to follow (see Figure 1(c)). Constructing a 3D undersea sensor network to detect undersea intrusions, however, is much more complicated. Even if an unbroken chain of overlapping sensor clusters from one side to the other side exists in a cuboid, intruders may still evade detection by passing through such a chain. Thus, an overlapping sensor chain no longer meets the requirement of detection. Instead, we require that a certain curved

Figure 1: (a) Sensors are dispersed from artillery ordinance on a ship. (b) Sensors are dispersed from an aircraft. (c) A 2D barrier on a plain surface.

surface that cuts across the space of interest (i.e., a 3D cross section) be fully covered by sensors. Figure 2 illustrates this requirement, where a sensor is denoted by the solid red dot and its sensing range by the cloud-like red sphere with the sensor in the center. In Figure 2(a), the coverage of sensors does not contain 3D cross section, where holes exist to allow intruders to pass through undetected. In Figure 2(b), the coverage of sensors contains a 3D cross section, which can detect any intruders.

Figure 2: (a) The sensor coverage does not contain a continuous cross section and intruders may pass through undetected. (b) The sensor coverage contains a continuous cross section that can detect any intruders regardless their moving paths.

The unique characteristics of underwater acoustic channels make the 3D intrusion detection even more challenging. Due to the quick absorption of radio in water, acoustic communication has been the most viable method for underwater environments. Such a transmission medium change, however,
brings grand challenges to the entire communication and networking stack. The propagation speed of acoustic signals in water is about 1.5 × 103 m/s, which is five orders of magnitude slower than that of radio signals in the air (3 × 108 m/s). In addition, the bandwidth capacity of underwater acoustic channels is limited and depends on both transmission range and frequency. Moreover, underwater acoustic channels are affected by path loss, noise, multi-path fading, Doppler spread, and a number of other factors, resulting in high error probability. In short, underwater acoustic channels feature long propagation delays, low available bandwidth, and high error probability. Underwater sensor nodes are typically powered by battery. It is very hard, if not impossible, to recharge or replace battery in practice. Thus, high energy efficiency thus becomes one of the most important design considerations for undersea sensor networks. To construct effective 3D undersea intrusion detection networks, all the aforementioned new issues and requirements should be considered and addressed.