Basics of Wireless Sensor Networks

Many companies and organizations rely heavily on data. Whether you're predicting growth of a company's revenue, or estimating how much inventory to order based on historic sales, data is essential to the management of all companies. As technology advances, the ability to collect outside data becomes easier via many computerized methods. One methods uses sensor nodes, tiny computers that collect data from their environment. The collection of sensor nodes that communicate together are known as a Wireless Sensor Network (WSN).

Sensor Nodes

Each sensor node is built with a sensor interface, a power source, storage, a microprocessor, and an antenna. Referring to Figure 1, we have a sensor chip which picks up various data from the environment, a power source which could be battery or solar powered, a radio transceiver with an antenna to communicate with other nodes, and a microprocessor to tie everything together [5].

Figure 1: Architecture of a sensor node. [5] — **Figure 1:** Architecture of a sensor node. [5]

Three components that are constantly being improved are the range of radio frequency, the lifetime of the power supply, and the size of the storage. Particularly with the size of the storage, we want a sensor node that is as small as possible while holding as much data as possible prior to transmission.

The small size of storage also gives rise to other problems in the cryptographic communication of nodes, as they hold a smaller number of keys. For more information, refer to my post on cryptographic pre-distribution in WSNs.

Sensors can pick up different types of data based on their type and configuration. Passive, omnidirection sensors require no trigger aside from the force acting upon it, and have no sense of direction - a wheel sensor on the road that tracks speed and traffic level fall into this category. Passive, narrow-beamed sensors send signal in a single direction and collect data based on objects that pass through this beam - store security sensors at the door fall into this category. Finally, active sensors are sensors which actively survey their environment - this can take the form of radio or sonar signals [6].

Figure 2: When nodes pick up data, they "hop" towards the sink. [4] — **Figure 2:** When nodes pick up data, they "hop" towards the sink. [4]

Sink, Gateway and Network Topology

In a WSN, the data collected from each node is forwarded to a sink, either directly or through multiple hops. This sink can then use the data locally, or it can transmit the data to various sources via a gateway [1]. In the following figures and examples, the sink node may or may not have a gateway attached.

Bus or In-Line Topology

A traditional bus topology in a wired network has all devices connected to a running wire that extends the whole network. In a WSN [2], one can think of the equivalent to an in-line topology, where each node can communicate to adjacent nodes only. For any node not adjacent to the sink node, several hops are necessary for the data to travel from the node to the sink. Note, this topology is a special case of a tree topology in which every node has at most one child.

Figure 3: An in-line topology of 5 sensor nodes and one sink, with each node only having adjacent nodes in its radius of communication. — **Figure 3**: An in-line topology of 5 sensor nodes and one sink, with each node only having adjacent nodes in its radius of communication.

Tree Topology

With the sink node at its root, a network in a tree topology has every node connected with at most one parent node and 0 or more child nodes. We can say the level of a node in such a network is the number of link separating itself and the root (sink). From this definition, we can clearly see that the sink node has a level of 0.

Star and Snowflake Topology

Similar to a wired star topology, there is a central device to which all other nodes connect; in this case it is the sink node. Although the sink may communicate to every node, each node may only communicate with the sink. A snowflake topology is an extension of the star topology, where each node connected directly to the source may branch out to further nodes. Similar to the in-line topology, these nodes would hop up the snowflake to the center sink node.

Figure 5: Five sensor nodes arranged in a star topology with a sink at the center. — **Figure 5:** Five sensor nodes arranged in a star topology with a sink at the center.

Figure 6: An extension of Figure 2 with four extra nodes branching out of the core star, creating a snowflake topology. — **Figure 6:** An extension of Figure 2 with four extra nodes branching out of the core star, creating a snowflake topology.

Mesh Topology

The benefit of a WSN is that a node doesn't need to rely on any cables to connect nodes; rather, it can communicate with any node that's within it's radius of communication. Simply, a mesh topology is any architecture that contains a cycles. One can also think of mesh topologies as being void of any dictated structure present in the topologies above. One major benefit of networks in mesh topologies is that losing one node due to computer failure or hacking does not disrupt the rest of the network. Furthermore, if a set of nodes in the middle slow down due to data traffic or maintenance, then alternate routes from the further nodes to the sink can be found.

Random and Arbitrary Networks

When dealing with wired networks, it is necessary to plan the network architecture in advance. Nodes are assembled, the architecture is planned, and the network with cables are deployed. The nature of wireless networks gives the user more freedom during the deployment process. No longer are we restricted by whether a cable pathway is feasible. This allows us to deploy networks with a large quantity of nodes. With these large scale networks, we deal with both random and arbitrary networks.

In a random network, nodes are randomly and uniformly distributed over a geographical area [3]. This uniform design allows us to collect data over approximately even intervals, and can be ideal for groups interested in full data coverage. The size of these networks are usually assumed to be large so that analysis of the network and its properties become well-define.

In arbitrary networks, we cannot guarantee the uniform distribution of the nodes. The are networks in which the deployment of the nodes may be ad hoc in nature. [3] For example, a company may be interested in deploying nodes over a forest, but only have access to a small number of nodes; these nodes are normally deployed by hand or with the use of a plane or helicopter to drop nodes over the most important areas. Similarly, a network may be required along a port in the water; in these cases, a boat may drop nodes at random, and the current may deposit the nodes off target. There are many examples where it may not be possible to guarantee or even attempt uniformity.

In all cases, multiple sinks are possible and may be useful as the number of nodes becomes large.

Figure 8: An example of a random grid network, where n nodes are deployed randomly and uniformly over a unit square region, further broken down — **Figure 8:** An example of a random grid network, where n nodes are deployed randomly and uniformly over a unit square region, further broken down

Figure 9: An example of an arbitrary network, where the nodes are spread throughout the forest to detect forest fires. — **Figure 9:** An example of an arbitrary network, where the nodes are spread throughout the forest to detect forest fires.

Wireless Sensor Networks benefit from being able to communicate with any node in it's RF radius, lending these networks to a variety of well connected architectures. Although the small storage of these sensors does create limiting issues in data collection and cryptography, advancements in technology are increasing the radius of communication and allowing more memory to fit into smaller sized storage.

With the prevalence of data science and machine learning in today's market, WSNs will become more popular as a cheap and efficient means of precise data collection.

References

[1] Buratti C., Conti A., Dardari D., Verdone R., An Overview on Wireless Sensor Networks Technology and Evolution Sensors,ISSN 1424-8220, August 2009

[2] Acharjya P. P., Santra S. A study and analysis on Computer Network Topology for Data Communication, International Journal of Emerging Technology and Advanced Engineering, Volume 3, Issue 1, January 2013

[3] El Emary, Ibrahiem M. M., Ramakrishnan, S., Wireless Sensor Networks: From Theory to Application, Boca Raton, Fla. : CRC Press, c2014 (Norwood, Mass. : Books24x7.com, Chapter 1.2, 1.3

[4] Chiara Buratti 1,* , Andrea Conti 2, Davide Dardari 1 and Roberto Verdone, An overview of Wireless Sensor Network Technology and Evolution, Sensor 2009, 9(9), 6869-6896

[5] Cao X., Hu F. Wireless Sensor Networks: Principles and Practice, Auerbach Publications Chapter 2

[6] https://en.wikipedia.org/wiki/Sensor_node#Sensors