Article Preview
Top1. Introduction
Wireless Sensor Networks (WSNs) are based on sensor nodes that are battery-operated. According to the application for which the WSN is designed, sensor nodes report measurements of certain phenomena to a sink node. WSNs have different applications including RADAR detection, agriculture monitoring, smart cities and many more (Waghmare, Chatur, & Mathurkar, 2016). This paper focuses mainly on monitoring electromagnetic (EM) pollution (Viani, Donelli, Oliveri, Massa & Trinchero, 2011); however, the findings of the paper are applicable to similar event detection WSN applications.
One of the main challenges faced by the WSN technology is the energy consumption and the energy efficiency due to the use of battery-operated sensor nodes. It directly affects the network lifetime, which is typically defined as the time until the first WSN node fails due to battery outage (Mahfoudh & Minet, 2008; Heinzelman, Chandrakasan, & Balakrishnan, 2000; Mamun, Ramakrishnan, & Srinivasan, 2010). Such a definition of the WSN death implies that when one of the node’s energy falls below a specific threshold that allows it to send and receive data, then the whole network will be considered dead. This definition of the network lifetime had a huge disadvantage on the network’s energy efficiency as well as the network lifetime. This is because the death of one node within the network does not imply that the remaining nodes are also incapable of correctly performing their task of detecting the monitored event. In other words, the other nodes in the WSN have an amount of energy that is high enough to allow the network to perform the required functions.
In this paper, our goal is to evaluate the energy efficiency and reliability of the different definitions of the network lifetime. More specifically, we consider the cases in which the network lifetime is defined as at least one sensor is still functioning, at least half the sensors are functioning, and the legacy case which requires all the nodes to be functioning in order to consider the network alive. These three different death criteria cover the different WSN applications. Other related WSN lifetime definitions were discussed in (Kaur & Singh, 2016). However, they rely on mobile sensors, grid optimization and energy proficient clustering techniques. Moreover, several cluster heads exist in such networks, which are based on the LEACH algorithm (Heinzelman, Chandrakasan, & Balakrishnan, 2000; Heinzelman, Chandrakasan & Balakrishnan, 2002). In contrast, our work considers the whole network as one cluster which relies on a single network master per round. This has already been proven to result in a prolonged network lifetime in (Botros, Elsayed, Amer, & El-Soudani, 2009; Seoud, Nouh, Abbass, Ali, Daoud, Amer & Elsayed, 2010). In this paper, we evaluate the network lifetime criteria assuming a predefined number of cycles per network master as opposed to (Seoud, Nouh, Abbass, Ali, Daoud, Amer & Elsayed, 2010). We also study the impact of randomly choosing the sensor nodes that serve as network master during the operation cycles, versus (Seoud, Nouh, Abbass, Ali, Daoud, Amer & Elsayed, 2010; Nouh, Abbas, Seoud, Ali, Daoud, Amer, & Elsayed, 2010) which had an ordered circular selection of the network masters. The reason for that is to investigate whether the choice of the network master has a significant effect on the network behavior, and accordingly, on the sensors energy or not. Unlike (Nouh, Khattab, Soliman, Daoud, & Amer 2017), we analytically evaluate the impact of the WSN death criterion on the reliability of the networks against other types of hardware and software failures.
The remainder of the paper is organized as follows. In Section 2, we describe the network model. Section 3 presents the different evaluated death criteria. We evaluate their energy efficiency under different sensing cycle lengths and different network master selection approaches in Section 4 and Section 5, respectively. In Section 6, we analyze the impact of the death criteria on the network reliability against hardware and software failures. Section 7 concludes the paper.