Chapter 14
Wireless Sensor Networks
14.1 Wireless Sensor Networks: State of the Art
Wireless sensor networks are a class of short-range wireless networks composed of a number of sensor nodes and one or more base stations (also called sinks) communicating with each other through a wireless transceiver.
A sensor node is typically composed of several parts: a wireless transceiver for communication, one or more sensors (e.g., temperature, pressure, humidity, accelerometer, etc.), a microcontroller unit, and an energy source, typically a battery. In some cases, sensor nodes are also equipped with actuators allowing the node to undertake some action (e.g., reducing illumination) in response to specific environmental conditions (e.g., a too high illumination level). If a network is composed of nodes, some of which also include actuators, it is called a wireless sensor and actuator network. The size, complexity, and cost of a sensor node vary widely depending on the specific application scenario: there can be sensor nodes as small as a coin costing a few dollars, as well as sensor nodes as large as a shoebox costing hundreds of dollars.
Sink nodes are relatively more powerful nodes, which typically are not equipped with sensors/actuators, but they have a more powerful processing unit, a possibly more powerful wireless transceiver, and other network interfaces usually allowing connection to the Internet. Furthermore, sink nodes are typically power-plugged, so energy consumption, which is a primary constraint for sensor nodes, is not critical for sink nodes. A special class of sink nodes are the mobile sinks—typically realized by mounting the sink node on a mobile robot—whose purpose is to travel throughout the wireless sensor network deployment area and gather the data collected by the sensor nodes.
A typical wireless sensor network architecture is shown in Figure 14.1: several sensor nodes are located in a geographical region and use wireless transceivers to set up wireless links between themselves. One or more sink nodes are co-located in the monitored region; these nodes act as data collection points, and possibly also realize network coordination functionalities. Due to severe energy constraints on the battery-equipped sensor nodes, a wireless transceiver on the sensor nodes transmits at low power but, consequently, the transmission range is relatively short—in the order of few tens of meters typically. Then, multi-hop communication is commonly a prerequisite for transmitting data gathered at remote sensor nodes to the sink node(s). In most application scenarios, both the sensor nodes and the sink nodes are stationary—Figure 14.1, left. In other cases, though, sensor nodes (or, at least, a portion of them) are mobile, for instance, when sensor nodes are installed on animals to monitor their movement. Finally, in some cases the sensor nodes are stationary while the sink nodes are moving, with the purpose of visiting the sensor nodes and directly gathering their collected data—Figure 14.1, right.
Figure 14.1 Typical wireless sensor network architectures.
The wireless sensor network concept was introduced in the late 1990s, and has become more and more popular in recent years as technology has turned this concept into reality. Currently, several wireless sensor network hardware and software platforms are available on the market, and significant standardization activities have been going on in the past few years (see below). Wireless sensor network prototypes have been extensively deployed for research purposes, and wireless sensor network-based solutions are becoming available on the market. However, as we will see in the next section, the application scenarios are so different that wireless sensor network-based solutions must be almost completely redesigned for each different scenario, negatively impacting the profitable use of this technology.
14.1.1 Hardware and Software Platforms
A number of hardware and software platforms for wireless sensor networking have become quite popular in recent years, at least among the research community. Indeed, due to the fact that sensor nodes are embedded devices, these platforms typically are strongly integrated. Thus, buying a certain type of sensor node usually “forces” the designer to use a specific embedded operating system and software platform.
The most popular hardware platform for sensor nodes is that of Motes, initially designed at the University of California, Berkeley and nowadays produced by a number of companies including Intel, Crossbow, etc. Mote nodes are composed of a relatively small motherboard (the size of two AA batteries or even less) including a microcontroller unit and wireless transceiver, and one or more daughter boards hosting different types of sensors.
The embedded operating system which was developed for Mote nodes at UC Berkeley is called TinyOS (Team 2011b), and is currently the most popular operating system for wireless sensor networks. TinyOS is an event-based operating system designed to account for specific features of wireless sensor networks, and includes support for energy-saving techniques, in-network data aggregation, procedures for gathering data to sink nodes, etc. Since the memory size on sensor nodes is very limited, special care has been taken in reducing the TinyOS memory footprint as much as possible. Currently, TinyOS can be used in combination with a number of hardware platforms including the family of Motes (Mica, Mica2, Mica2Dot, MicaZ, TMote Sky, iMote), the Eyes platform developed at Technische Universität erlin, and so on.
Another relatively popular operating system for wireless sensor networks is Contiki (Team 2011a), developed at the Swedish Institute for Computer Science (SICS). One of the main features of Contiki is that a relatively advanced programming model including “proto”-threads and dynamic memory allocation is made available to the user on top of the event-driven operating system kernel. Currently, Contiki supports a number of hardware platforms including the Mote family.
14.1.2 Standardization Activities
Several standards for wireless sensor networks are currently either ratified or under development. The most important standard, first approved in 2003, is IEEE 802.15.4 (IEEE 2011), which defines the PHY and MAC layer specifications for low-cost, energy-efficient wireless personal area networks. Since the emphasis in IEEE 802.15.4 is on reducing energy consumption by as much as possible, data rates are relatively low and the communication range relatively short (between 10 and 75 m). Furthermore, support for transmit power control is included in the specifications. At the PHY layer, the standard operates in the ISM radio bands: 868 MHz in Europe, 915 MHz in the USA and Australia, and 2.4 GHz in most regions worldwide. The data rates are 20/40 kbps in the 868 and 915 MHz bands (increased to 250 kbps in the 2006 release of the standard), and 250 kbps in the 2.4 GHz band. At the MAC layer, the standard implements CSMA/CA, in a manner similar to that defined in the IEEE 802.11 standard.
The IEEE 802.15.4 standard defines two types of network nodes: the full-function device (FFD), which can serve as the coordinator of a personal area network (PAN) or as a common node, and includes advanced communication capabilities such as relaying messages; and the reduced-function device (RFD), which is a simplified device with limited communication capabilities, and can communicate only with a FFD.
IEEE 802.15.4 allows the formation of two types of network topology (see Figure 14.2): the star topology, in which all devices communicate directly with a central device called a PAN coordinator, and the peer-to-peer topology, in which devices can set up an ad hoc network with peer-to-peer wireless links.
Figure 14.2 IEEE 802.15.4 network topologies.
While IEEE 802.15.4 defines specifications for the lower layers of the network architecture, other standards have been defined for upper layer protocols. ZigBee is a specification (first released in 2005) for a suite of protocols encompassing the network layer and above, for networks based on 802.15.4 wireless links (Alliance 2011). As in IEEE 802.15.4, the main goal of ZigBee is to reduce the energy consumption of devices: just to give an idea of this emphasis on reduced power consumption, an individual device must have a battery lifetime of at least two years to pass ZigBee certification.
Starting from the PHY and MAC layers defined in IEEE 802.15.4, ZigBee introduces four main components into the standard: the network layer, the application layer, ZigBee Device Objects (ZDOs), and manufacturer-defined application objects.
At the network layer, ZigBee uses ad hoc on-demand distance vector (AODV) routing (originally designed for ad hoc networks) to route messages within the network. The application layer is the highest level layer defined by the specification, and it comprises the majority of components added by the ZigBee specification: ZDOs and relative management procedures, together with application objects defined by manufacturers. The ZDO is responsible for defining the role of a device as a coordinator, router, or end device. The coordinator is the most capable ZigBee device: it forms the root of a network tree, and it might act as a bridge to other ZigBee networks. There is exactly one coordinator in each ZigBee network, which is the device that started the network originally. The router is a device which can act as an intermediate router, relaying messages from other devices. Finally, the end device is the simplest device, and it encompasses functionality to communicate with the parent node (either the coordinator or a router). Several individual ZigBee networks can be connected to form a larger wireless sensor network with mesh topology (see Figure 14.3).
Figure 14.3 Example of ZigBee mesh network.
Similar to what happens in WLANs with the WiFi Alliance promoting usage of 802.11-based WLANs, the ZigBee Alliance is a group of companies that maintain and promote usage of the ZigBee standard and underlying 802.15.4 technology.
Another standard for higher layer protocols which is based on 802.15.4 PHY and MAC layer specifications is WirelessHART (HARTCommunicationFoundation 2011), first released in 2007 and the wireless evolution of the pre-existing HART standard for process measurement, control, and asset management applications. Given the considered application scenario, emphasis in the standard is put on reliability, security, and effective power management. The standard operates in the 2.4 GHz band only, and ensures full compatibility with legacy HART devices.
14.2 Wireless Sensor Networks: User Scenarios
As mentioned previously, wireless sensor networks find application in many diverse user scenarios. Roughly, we can divide these scenarios into the following categories: environmental monitoring, industrial monitoring, health and well-being monitoring, precision agriculture, seismic and structural monitoring, intrusion detection, and tracking of objects, people, or animals. In the following, we will briefly describe representative user scenarios in each of these categories.
14.2.1 Environmental Monitoring
Wireless sensor networks (WSNs) can be used to remotely monitor relevant environmental parameters in relatively large geographical areas. For example, a wireless sensor network can be deployed in an urban area to monitor pollution levels. In this case, sensor nodes are equipped with sensors able to detect polluting gases and particles. The advantage of using a WSN instead of traditional systems to monitor pollution in a city is that, due to the use of wireless technology, sensor deployment is much cheaper, and finer grain monitoring can be obtained at affordable costs. In other cases, WSNs can be used to monitor environmental parameters in dangerous regions, such as volcanic regions, where the use of wiring is virtually impossible. Another example of possible WSN usage is to monitor land movements in order to prevent or quickly notify about landslide events. Similarly, WSNs can be used to monitor glaciers and the status of snow in order to quickly alert about possible avalanches. Finally, another user scenario worth mentioning is forest fire detection.
In environmental monitoring scenarios, data collected by the WSN (possibly suitably aggregated by the sensor nodes themselves) is conveyed to the sink node(s), and from there sent to the user for remote monitoring (continuous data monitoring model). In some cases, an automatic alert system could be implemented in order to quickly inform the human operator when environmental parameters exceed certain thresholds, corresponding to the occurrence of possibly dangerous events (event-driven model).
A special class of WSN that can be used for monitoring purposes is the class of underwater WSNs. In this class, communication between sensor nodes is realized through either acoustic or optical communications, since radio waves do not propagate underwater. The challenges faced by a network designer, especially concerning communication between sensor nodes, are quite different in an underwater environment compared to those faced in terrestrial applications.
14.2.2 Industrial Monitoring
WSNs can be used to monitor complex industrial processes, such as in oil refineries, water or wastewater monitoring, power plants, nuclear plants, etc. In these scenarios, use of wireless technology offers unique cost reduction opportunities: it is in fact well known that a major cost in industrial process monitoring is related to wiring costs. However, the security and reliability of the WSN comes to the fore in this class of applications, considerably challenging the WSN designer.
In many cases, WSNs employed in industrial monitoring are wireless sensor and actuator networks: for instance, if a dangerous pressure level is detected in a pipe in an oil refinery, a certain actuator can be activated to turn a valve and reduce pressure. In this way, immediate action can be undertaken by the wireless sensor and actuator network itself in response to a potentially dangerous situation, without requiring the intervention of a human operator.
14.2.3 Health and Well-Being Monitoring
Another important class of WSN user scenarios is related to monitoring the health and well-being of people. For instance, a small WSN can be attached to a patient suffering from heart disease to continuously monitor blood pressure, heart rate, etc. The data collected from the WSN can be transmitted wirelessly to a sink node located in the home. If some abnormal condition is detected, the sink node issues an alert message to the nearest hospital asking for medical assistance. This way, the patient is free to move around at home without any physical restriction, while her/his health status is continuously monitored. Another application of WSNs is in monitoring the status of participants in large sporting events, such as a marathon: by equipping each participant with a wireless device endowed with pressure, temperature, etc., sensors, the health status of each athlete can be remotely monitored. An application of WSNs related to well-being is when sensor nodes are attached to different parts of the body (wrist, elbow, shoulder, knee, etc.) in order to monitor whether a certain physical exercise is performed correctly.
14.2.4 Precision Agriculture
WSNs can be used to provide fine-grained monitoring of crops in agriculture. For instance, WSNs can be used to monitor the sugar content of grapes, so as to judge the right level of ripening needed for a certain wine production. WSNs can also be used to monitor the level of humidity in the ground surrounding crops, so as to optimize use of the irrigation system. Note that these applications would be extremely expensive and difficult to realize without using a wireless technology.
14.2.5 Seismic, Structural, and Building Monitoring
Another important class of WSN applications is concerned with monitoring large structures or buildings. For instance, wireless sensor nodes can be installed at specific points in a large structure such as a bridge, in order to monitor movement of the various parts comprising the structure, and to promptly detect possible critical conditions. This way, engineers could monitor the “health” status of the structure remotely, avoiding costly and time-consuming site visits. Another advantage of using WSNs for the purpose of structural monitoring is the availability of a large amount of continuously collected data, as compared to weekly or monthly data collected during physical site visits. The availability of such fine-grained and massive amounts of data allows a much more detailed study of the events experienced by the structure, as well as better monitoring of its structural health.
Wireless sensor and actuator networks can also be used to monitor and optimize the management of a large building: sensors can be used to monitor temperature, illumination, humidity, etc. within the building, and actuators can be used to undertake appropriate actions in response to some detected conditions. For instance, if the temperature in a room is too high and illumination too intense, window shutters in that room could be automatically closed. The ultimate goal of a wireless sensor and actuator network used in building monitoring and management is to reduce energy consumption in the building, while at the same time providing optimal environmental conditions inside it.
14.2.6 Intrusion Detection
WSNs can be used also to detect intruders in a certain area or building. For instance, movement sensors, typically complemented with some camera-equipped sensor node, can be installed on a fence surrounding a restricted access area, in order to detect possible intrusions. Typically, WSNs used for intrusion detection obey the event-driven model, that is, an alert message is sent to a remotely located human operator to warn of a possible intrusion. On receipt of the message, the human operator can activate and operate wirelessly controlled cameras, to check whether an intruder is actually trying to access the restricted area.
14.2.7 Tracking of Objects, People, and Animals
Another class of WSN applications is related to tracking objects, people, or animals. For instance, WSNs can be used to track the movement of objects in a large warehouse. Tracking objects or people finds application in military scenarios, for instance, to track the movement of soldiers in the field or the movement of tanks and other military vehicles along roads. Finally, WSNs can be used to track the movement of both domestic (e.g., herds) and wild animals. As a matter of fact, one of the first examples of a working WSN prototype is ZebraNet (Martonosi et al. 2004), a WSN composed of a small number of radio-equipped zebras living in the Sweetwaters Reserve in Kenya.
14.3 WSNs: Perspectives
While WSN technology is relatively mature, due especially to the introduction and maturing of the IEEE 802.15.4 and related standards, its applications are still mostly at the prototype stage. Thus, a major development that we envision in the coming years is the evolution of WSNs into a commercially profitable technology. As briefly mentioned earlier in this chapter, a factor currently slowing down the wide adoption of WSN-based solutions on the market is the numerous and diverse applications of WSNs (see Section 14.2) such that the WSN designer is forced to almost completely redesign the network for each different application. Thus, this market, which is potentially very large given the very wide range of possible applications, is partitioned into several different “niche” markets, almost in a one-to-one relationship with possible applications. Clearly, this market fragmentation is a factor negatively impacting the development of profitable WSN-based solutions.
In terms of technology, two major challenges are yet to be fully addressed. The first refers to the reliability and security of WSN technology: while reliability and security are an issue in wireless technologies in general, they are even more severe in WSNs, due to the use of low-cost, low-power network devices in a potentially harsh environment. It is important to observe that reliability and security are prerequisites for introducing WSN-based solutions in one of the potentially more profitable niche markets, namely, that of industrial monitoring applications. Thus, significant research efforts must be undertaken in order to improve the degree of reliability and security level provided by WSNs.
The second major technological challenge relates to the use of energy harvesting techniques to prolong WSN lifetime. In fact, WSN lifetime is perceived as another major factor currently limiting the widespread use of WSN-based solutions, and energy harvesting techniques—using, for example, micro solar panels, extracting energy from vibrations, etc.—are increasingly being considered as a possible way of integrating the energy provided by batteries and significantly extending WSN lifetime. How to optimize the WSN design in order to fully exploit the harvested energy is currently a very active research field, and it is likely to remain so for several years to come. In fact, WSNs have so far been designed to reduce energy consumption by as much as possible, but under the assumption that the (limited) energy is continuously available over time. Instead, the amount of energy that can be harvested from the environment changes significantly over time, often very quickly (think about the impact on solar energy production of a cloud obscuring the Sun). Thus, the WSN should be optimized to operate in different conditions for the amount of energy available, and methods should be devised to quickly detect changes in the amount of this energy.
14.4 Further Reading
This chapter presented only a very short description of wireless sensor networking technology and related issues. The reader interested in gaining a better understanding of this topic is referred to the several survey papers and books available in the literature, such as the survey by Akyildiz et al. (2002) and the more recent survey by Yick et al. (2008). Suggested books include those by Karl and Willig (2007), Akyildiz and Vuran (2010), and Zhao and Guibas (2004).
Akyildiz I and Vuran M 2010 Wireless Sensor Networks (Advanced Texts in Communications and Networking). John Wiley & Sons, Chichester.
Akyildiz I, Su W, Sankarasubramanian Y and Cayirci E 2002 A survey on sensor networks. IEEE Communications Magazine 40(8), 104–112.
Alliance Z 2011 http://www.zigbee.org/.
HARTCommunicationFoundation 2011 Wireless hart technology http://www.hartcomm.org/protocol/wihart/wireless_technology.html.
IEEE 2011 http://www.ieee802.org/15/pub/TG4.html.
Karl H and Willig A 2007 Protocols and Architectures for Wireless Sensor Networks. John Wiley & Sons, Chichester.
Martonosi M, Lyon S, Peh LS, Poor V, Rubenstein D, Sadler C, Juang P, Liu T, Wang Y and Zhang P 2004 http://www.princeton.edu/mrm/zebranet.html.
Team C 2011a Contiki operating system. http://www.sics.se/contiki/about-contiki.html.
Team T 2011b Tiny operating system. http://www.tinyos.net.
Yick J, Mukherjee B and Ghosal D 2008 Wireless sensor network survey. Computer Networks 52(12), 2292–2330.
Zhao F and Guibas L 2004 Wireless Sensor Networks: An Information Processing Approach. Morgan Kaufmann, San Francisco.