«Carles Gómez Josep Paradells José E. Caballero Edita: Fundación Vodafone España Autores: Carles Gómez Montenegro* Universitat Politècnica de ...»
Pero el mercado y las oportunidades de negocio entorno a las redes de sensores (como habilitadoras de nuevos servicios) no serán una realidad si
no se consigue superar el problema de un ecosistema (tecnologías y protocolos) excesivamente fragmentado en el que no aparezcan estándares de referencia. Por tanto, la tercera línea de actuación debería ir en este sentido.
En cuarto lugar, las soluciones de inter-conexión con el mundo exterior para poner en valor esa información procedente de las redes de sensores precisan de una evolución de las tecnologías existentes hacia aproximaciones estándar, ampliamente aceptadas, típicamente basadas en el protocolo IP. IETF está trabajando en ello desde hace algún tiempo (p. ej. en los grupos de trabajo 6LoWPAN y ROLL), y ya se puede decir que existen soluciones de redes de sensores sobre IP que ofrecen un buen rendimiento. De hecho algunas organizaciones de referencia en el sector (que desarrollan sus propios estándares inicialmente no-IP), como por ejemplo ZigBee o Z-Wave, ya han anunciado que están evolucionando sus soluciones hacia una pila de protocolos que incluirá IP (IP stack). Este será un pilar fundamental sobre el que se pueda construir la “Internet de las Cosas”, cuya visión quedó establecida por la UIT hace algunos años, y que muy básicamente consiste en vislumbrar que a partir de un determinado momento empezará a haber muchas más máquinas que personas inter-conectadas (p. ej. a través de Internet), y gran parte de ellas podrían ser nodos de una red de sensores.
En quinto y último lugar, el desarrollo de una gran cantidad de aplicaciones entorno a la información obtenida desde las redes de sensores debería ser habilitado por la existencia de un marco de trabajo adecuado que facilitase el desarrollo de servicios innovadores, y que puede pasar por aproximaciones de middleware y plataformas “horizontales”, alineadas con el modelo exitoso de la web 2.0 actual, en contraposición a aproximaciones “verticales” que afrontan nichos de mercado concretos y que ofrecen soluciones propietarias.
A un plazo más largo, se puede pensar en las siguientes generaciones de las redes de sensores en las que los nodos sensores estén embebidos en las cosas y personas. En este punto se puede hablar de la redes de sensores en las superficies, en la materia y en las personas. En este contexto se necesitarán nuevos paradigmas de funcionamiento.
Introducción to Wireless Sensor Networks Chapter 1. Introduction to Wireless Sensor Networks
1. Introduction to Wireless Sensor Networks This chapter introduces the main concepts concerning Wireless Sensor Networks (WSNs). The chapter is organized as follows: Section 1.1 provides the definition of the main devices that can be found in a WSN ; Section 1.2 describes the main characteristics of WSNs; Section 1.3 introduces the architecture of a sensor node, and finally, section 1.4 presents the main WSN architectures, including network topologies and communication protocol solutions.
1.1. Elements of a WSN
A WSN is a wireless network of wireless sensor nodes (that may interact which each other) aimed at monitoring real world physical parameters (in many cases, covering a certain geographical area) and offering the sensed data to one or more data collection elements. In some cases, the same elements of a WSN can be used to activate actuators in order to perform certain tasks in response to an event or to exceed a predefined threshold for a certain parameter.
Sensor nodes are devices that account with at least one sensor and may include actuators as well as having processing and networking capabilities to process data and use the wireless access.
Sensors are devices included in the sensor nodes. Sensors measure the physical property and quantity of an observation, converting the measure
ment into a signal that may be electrical (e.g. current, voltage, power, resistance, etc.), mechanical (e.g. pressure, flow, liquid density, humidity, etc.), chemical (e.g. oxygen, carbon monoxide, etc.), acoustic (e.g. noise, ultrasounds, etc.), or any other signal type. Other parameters which can be sensed include the identification of a device or its location.
Actuator nodes are devices with wireless communications and processing capabilities which include at least one actuator.
Actuators are devices that can carry out an action (e.g. a physical response such as turn on a light, trigger an alarm, turn off an irrigation system, etc.) in response to a certain stimulus (caused by an input signal).
Typically, sensors are small, low-cost, and low-power consumption devices requiring a limited amount of information transfer. However, a wider concept may include microphones, GPS-receivers or cameras as sensors.
Furthermore, even a satellite system can be considered as a “sensor device” .
In data collection applications, there exist one or more sink nodes, which are special nodes in charge of gathering, processing and controlling the received data from a set of sensor nodes.
WSNs can be stand-alone networks, but it may be interesting to connect them to other networks (e.g. the Internet) for remote access and management. In this case, connectivity between the WSN and another network can be achieved by means of a Sensor Network Gateway, WSN technology opens the door to a wide range of control and monitoring Sensor Network Applications (or use cases) in a large spectrum of environments, such as home automation, building automation, industrial automation, environmental monitoring, urban monitoring, etc. (see Chapter 2).
1.2. General characteristics of WSNs A WSN can be as small as a two-node network or as large as a ten-million nodes network . The actual network size will depend on each particular
application and deployment, but it is assumed that, in general, a WSN is composed of a large number of nodes.
An important feature of WSNs is their unattended operation. Some reasons for this include the possibly large number of nodes and the fact that some WSNs may be placed in zones where access is difficult.
In many applications, the nodes may only obtain their power from a battery1. In this way, network deployment is easy and independent of power supply availability. However, battery replacement may not be practical or even possible. In consequence, energy conservation is a primary goal in WSN design, as the nodes will only operate before battery depletion.
In contrast with many other networking technologies, WSNs exhibit relaxed bandwidth requirements. In fact, in many applications, the data collected by a sensor (e.g. a temperature measurement) can be encoded using a reduced number of bytes and transmitted at a low rate (e.g. once per minute). However, the actual data transmission rate depends on each application, and some applications may exhibit low latency requirements (e.g. a light is turned on when the user presses a button in a remote control).
The nodes in a WSN use wireless communication technology, which offers greater flexibility and lower deployment costs than wired technology.
As the transmission range of a node is limited (see Chapter 3), a multi-hop communication solution is needed so that nodes can route data on behalf of other nodes and enable the transmission of data between two nodes that are not within the transmission range of each other. Typically, WSNs are multi-hop networks. On the other hand, there exist various radio propagation impairments (see Chapter 3) which make the quality of a wireless link uncertain and variable in comparison with that of a wired link. This is an important feature that WSN communication protocol design must take into account.
Wireless sensor nodes are typically constrained devices, which exhibit low processing power and low memory storage capacity. Some examples include Another option is to harvest energy from the environment-,( see Chapter 14), but this option does not provide an unconstrained energy supply.
8-bit, 16-bit and 32-bit processors running at tens of MHz and RAM sizes from 1 kB to 256 kB . Hence, the protocols running in WSNs must be lightweight.
This is challenging, as the large number of devices and unatten- ded operation of WSNs pose additional constraints to the networking solutions.
Some WSNs are mainly static (e.g. a set of nodes deployed in a forest).
However, the dynamics of wireless communications make the network conditions quite variable. On the other hand, in some scenarios, the nodes may exhibit a certain degree of mobility (e.g. a remote control, on-body sensor nodes, sensors attached to vehicles, etc.). Thus, the WSN protocols must be self-healing and capable of operating in the presence of topology changes.
1.3. Sensor node architecture This section overviews the architecture of a sensor node from the hardware (subsection 1.3.1) and software (subsection 1.3.2) points of view.
A sensor node is typically a small-sized, autonomous sensing and communication system. One of the seminal projects in the WSN field was the Smart Dust project , which started in the late nineties with the aim of developing a cubic millimeter sensor node. Most current sensor node hardware platforms have a larger size, but are still called motes, which is the name given to the devices developed by the Smart Dust project. In fact, the small size of sensor nodes is practical, cost-effective, and approaches WSNs to the concept of invisible, ambient intelligence. However, the node size poses constraints on the hardware capabilities of the devices, which affect communication protocol design on top of WSNs.
Fig. 1.1 shows the main components of a sensor node: processor, transceiver, memory, power source, transducers and, if necessary, analogue to digital converters (ADCs). A transducer2 refers here to a sensor device which A transducer may either be a sensor or an actuator.
Chapter 5 provides more detail on the hardware elements of a sensor node and the features of the main platforms available on the market.
1.3.2. Software Sensor nodes run special operating systems, communications protocols, management functionality and applications on top of them.
The operating systems have been specifically developed for the particular characteristics of sensor nodes. As already mentioned, the hardware platforms on top of which these operating systems run are constrained. Hence, these operating systems have been developed under the goals of simplicity and efficiency. Some of them are also adequate for event-detection applications, which are common among WSN applications. Chapter 10 further elaborates on the characteristics of the operating systems developed for constrained devices such as sensor nodes.
From a certain point of view, the applications running on a sensor node require a correct operation of the node and the network in terms of quality of service, data management and energy management. This software functionality as a whole is sometimes referred to as middleware (see Chapter 16).
1.4. WSN architectures This section presents the main network architecture paradigms of WSNs.
First, the main topologies used in WSNs are introduced in subsection 1.4.1.
Then, subsection 1.4.2 discusses the use of a layered protocol stack in WSNs. Finally, subsection 1.4.3 introduces the protocols needed in WSNs.
1.4.1. WSN topologies This subsection presents some of the main topologies that can be found in WSN solutions. These topologies are: star topology, tree topology and mesh topology. In some cases, combinations of these topologies are also supported by WSN solutions.
220.127.116.11. Star topology In the star topology, there exists a central node which has communication links with the other nodes (see Fig. 1.2). The central node plays an important role, as it may act as a controller scheduling the transmissions of the rest of nodes; it may route data between nodes if necessary (note that the largest paths will be composed of two hops), and it may also act as a sink node. As the central node has greater functionality and importance than the other nodes, it may not be as constrained as the others (e.g. in processing power, memory storage, power supply, etc.).
The star topology is suitable for small networks where the nodes have a direct communications link with the central node. A drawback of this topology is its lack of path redundancy. In other words, there exists a single path between any two nodes in the network. Given the dynamicity of wireless links, when data cannot reach the destination through a certain path, the communication cannot benefit from using other paths. Hence, the star topology should be used in very specific, well-dimensioned scenarios.
18.104.22.168. Tree topology
In a tree topology (see Fig. 1.3.a)), each node has a communications link with a parent node, which is the next hop in the path towards a special node called the root. (Note that the root does not have parents.). The nodes that have a certain parent are called the children of that parent. For example, in Fig. 1.3.a), nodes 4 and 5 are children of node 1, which is their parent. In turn, node 1 is a child of node 14, which is both the parent of nodes 1, 2 and 3 and the root of this tree topology.
The tree topology is adequate for data collection applications, whereby the data have to be transmitted to a sink node. The root of the tree can be the sink node. Communication between any two nodes is also possible in a tree topology, but without additional mechanisms it can be suboptimal. For example, in Fig. 1.3.a), the path followed by the data transmitted from node 10 to node 11 will be 10-5-1-14-3-8-11. This path is longer than the minimum possible path, which would be available in other possible network topologies (see Fig. 1.3.b), for instance).
A strict tree topology is not robust to link failures (e.g. due to radio impairments, battery depletion, node mobility, etc.) because there exists a single path between a node and the root. To increase path redundancy, some topology variants have been considered in WSNs, including the possibility of selecting more than one parent for a node, or to use siblings3 for forwarding data (see subsection 7.3.2, Chapter 7).