**1. Introduction**

Presently, houses, buildings, parks and, cities in general, involve several electronic devices working with different wireless technologies. The type of application determines the appropriate combination of standards and protocols to be used. One of the main characteristics of these wireless devices is their power requirements.

Wireless Sensor Networks (WSNs) involve devices characterized by small nodes, low energy consumption, limited battery life, low task processing, and low storage capacity. These self-configuring networks are easy to implement and to deploy. In these networks, communications through channels with multiple interferences and computing capabilities to operate at low energy are assessed. Sensor networks should exhibit an optimal performance with reduced delays and provide reliable information with a minimum energy consumption in order to provide valuable information for long periods [1]. However, energy consumption may become a major issue because of the low-battery power. The life span of the nodes should be as long as possible to avoid constant human intervention due to the harsh environment of some of their applications, such as in the study of natural behavior, risk areas, medical industry, domotics, agriculture, battlefields, and home networks [2].

Considerations on energy consumption are critical in sensor networks because their implementation must be simple, enduring, and resilient to topology or configuration changes. All these factors significantly influence the energy expenditure of a network and are represented by its performance parameters.

In this study, an efficient energy model for sensor networks is implemented with the aim of quantifying energy consumption during the execution of the main tasks of a node within a network [3]. The types of energy considered by this model correspond to the following phases of operation: turning on, channel auditing, receiving packets, sending packets, switching activities, microcontroller processing, and turning off. The novelty of this model is that it features a simple design scheme capable of characterizing energetic behavior against possible network anomalies in which consumption levels exceptionally increase, as it can be seen in [4]. In the proposed model, energy is considered to be an indicator of a typical behavior for a network exposed to interference attacks. In addition, it is an easy scheme to implement in a node, which provides reliable responses to modifications in node behavior. In this way, it is also possible to optimize energy consumption for specific node activities as well as for the overall network performance. The model allows for scalability and demonstrates the main reaction modes of a network node.

This paper intends to test the proposed energy model and observe its repercussions on proactive and reactive sensor network protocols. The analysis is described quantitatively by observing the performance metrics that would positively or negatively affect such model. This is where the contribution of this work becomes relevant: the proposed model quickly shows changes in the network performance, its implementation is simple, and it does not represent higher processing consumption. The proposed model is then compared against the performance of network sensors under some widely known protocols: Ad hoc On demand Distance Vector (AODV) [5], Dynamic Source Routing (DSR) [6], ZigBee Tree Routing (ZTR) [7], Low Energy Adaptive Clustering Hierarchy (LEACH) [8] and, Power Efficient Gathering in Sensor Information Systems (PEGASIS) [9]. These protocols will also be compared against the Multi-Parent Hierarchical (MPH) routing protocol proposed, designed, and implemented by the authors in a previous work [10].

There are also other protocols analyzed in the literature, such as Cluster-based Energy Efficient Location Routing Protocol (CELRP), which is a hierarchical protocol with nodes distributed in clusters and arranged in quadrants. Each quadrant contains two clustering, which would be like the master nodes, and other nodes transmit data with two hops data transmission. Another similar protocol is Position Responsive Routing Protocol (PRRP), which is more energy efficient. This protocol makes a choice of the cluster head based on distance from the sink, energy level, and the average distance of neighboring nodes from the candidate master node. PRRP is similar to the LEACH protocol in which any node can communicate with the sump and the data transmission mechanism is the time-based schedule. In PRRP, the number of nodes of the branches of the hierarchical tree and the distance from the non-leaf node is smaller compared to LEACH and CELRP. This makes energy conservation candidate for optimization [11]. The PRRP protocol dramatically increases data transfer and provides a better solution to the routing problem focused on energy efficiency, due to the efficient selection and distribution of gateways. Another important protocol to mention is the Energy-Efficient data Routing Protocol (EERP) for WSN [12], which selects a set of good roads, and chooses the one based on the node state and the road cost function. In EERP, each node has several neighbors through which packets can be routed to the base station. A node bases its routing decision on two metrics: status and cost function.

In this study, AODV, DSR, ZTR, LEACH, and PEGASIS are quantitatively compared and assessed based on several efficiency metrics that analyze how these routing protocols optimize energy through various schemes in order to find the best routes in the shortest possible time. As the hierarchy algorithms, such as the ZTR, denote simple and fast routing that reduce network overloads, they are reliable and have a distributed addressing scheme that only permits neighbor tables, not long, and elaborated routing tables. The performance of WSN is closely related to that of the routing protocol, because routes can vary dynamically over time. Energy-aware protocols such as LEACH and PEGASIS seek to increase the lifetime of the network. They propose to find sub-optimal paths to allow a more equitable distribution of the network's energy consumption. Hierarchical protocols such as ZTR and MPH have advantages in terms of scalability and efficiency in communications. Particularly for WSN, nodes with higher energy can be used to process and send information, while those with lower energy are used to monitor the environment and send the information to the node with greater energy capacity. Finally, proactive type protocols, which establish routes before there is a real traffic demand, are suitable for real-time traffic, since they have low latency; however, they waste bandwidth due to periodic updates and they are not energy efficient. The MPH protocol is a hybrid protocol, i.e., it is a combination of reactive and proactive nature protocols. The AODV and DSR protocols are two protocols widely recognized in WSN for their rationality and the ZTR protocol is a proactive protocol par excellence. Therefore, we believe that the comparison between protocols of natures of the same type is relevant. In addition, we wanted to complement our study with energy-aware protocols because the main objective of this work is to demonstrate and analyze the higher energy costs in the sensors, according to the type of tasks they perform on the network.

This paper proposes a simple energy model, which quickly shows changes in the network performance, its implementation is simple, and it does not represent higher processing consumption. In the WSN literature, there are few energy models [13,14] and some energy-aware protocols that seek to optimize the energy of networks. The need for an energy model that impacts the performance metrics of a network is an advantage that not all models exhibit. This indicates that we can know, according to each type of task that the node performs, what is the major and minor impact on parameters such as: resilience, overhead, packet retransmissions, listening retries to the communication channel, delay, and many others.

The energy consumption problem is not the same for all network nodes [15]. This is due to the fact that there are several collector nodes that cluster information around them. This coordinating collector node is robust, with its own energy supply and with greater processing resources than other nodes in the network. Consequently, this node has the capacity to process all the information gathered from the nodes of the network and subsequently, obtains results when assessing the information received. When there are one or more collector nodes and there are nodes nearby that forward all network traffic, they are more likely to exhaust faster their energy. This problem is known as the energy hole problem [16] and generates a high amount of packet losses, which will be represented by collisions. Unequal energy depletion causes the expiration time to unexpectedly generate loss of information from the network.

A possible solution to this problem, as per the literature, is the creation of groups to promote network scalability and the problem of zoning [17]. The analysis of the network areas is distributed in concentric rings to stimulate traffic between nodes as they approach their destination. The authors in [18] study variables such as constant bit rate, where the nodes are uniformly and randomly distributed. Performance parameters are analyzed to establish relationships between the different rings located near and far away from the collector node. In this study, the concept of node zones is used to observe energy repercussions based on accurate and reliable performance metrics that

directly influence energy expenditure and are presented as reliable evidence of network changes or anomalies. The link load imbalance problem is addressed through tree-type topology protocols aimed at minimizing packet delays and the number of jumps in the path to the root node or collector node. This imbalance is reflected as a non-uniform energy expenditure on the nodes. This failure is compensated in reactive protocols by adding more links, but this is prone to greater delays or outdated routes and their maintenance may become a challenge [19]. The work in [20] shows a method for balancing network traffic and ensuring uniform use of destination node routes. Nezhad et al. [21] propose a protocol in which the collector node has a global view of the network topology to maximize the life span of each node and the use of a load balancing algorithm to select the best routes.

Herein, some energy consumption performance metrics are proposed and used to compare the MPH routing protocol against other known sensor network protocols: AODV, DSR, ZTR, LEACH, and PEGASIS, all as per the IEEE standard 802.15.4/ZigBee. Node shutdown tests are performed to study how the network behaves according to the routing protocol implemented and its response capabilities [22]. The results of the simulation show how performance, reliability, and energy consumption are affected within the communications network. In this light, MPH is shown to be an efficient protocol as it presents the best performance among the protocols under evaluation. Compared to AODV, MPH routing exhibits a 26.9% decrease in overall energy consumption and a 41.2% increase in the protocol's ability to recover the topology after a failure event. In addition, an energy model for the CC2530 chip is proposed and used in the simulations of the four protocols aforementioned, resulting in a 16%, 13%, and 5% reduction in energy consumption for the MPH routing when compared to AODV, DSR, and ZTR, respectively. The proposed model allows us to determine that between MPH, LEACH and PEGASIS there is only a difference of 3% and 2% energy savings for the last two protocols. Thus, the model analyzes the energy impact of each type of energy for optimization of the algorithm in various protocols of the literature. These protocols and the energy model are implemented in an event simulator programmed in C++. The proposed energy model is implemented in the simulator for each routing protocol to observe the impact of the performance parameters and how they influence energy consumption in each protocol. This proposal resembles the MPH routing protocol [10], which is a hybrid that combines proactive and reactive protocols and energy conservation properties at the same time. It is used in this work as a protocol with hierarchical topology for optimizing the sending of information, maintaining a smaller number of routes, and monitoring them to determine whether they remain valid or have become obsolete [23].

Figure 1 shows a conceptual scheme for a hierarchical topology with a sink node or information collector. This scheme is widely used in WSN for low power consumption [24]. The problem described here is that energy is not distributed equally throughout the topology and there are different energy levels called crowns [25]. Energy increases in the nodes that form information bottlenecks. Therefore, in this work we classify the types of energy to establish compensations in the busiest nodes of the network. This scheme overviews the energy consumption of a many-to-one network. The coordinator or sink node is the sole destination of the network nodes (there may be more than one coordinator node), which makes it a single failure point. The nodes send traffic to the coordinator node. As the nodes are closer to the coordinator, they have to forward traffic from other network nodes. Therefore, with this scheme, we want to represent an ideal scenario of energy behavior of the nodes from the coordinator node to the nodes farthest from it. Being the darkest color the higher energy consumption nodes and the lightest, the lower energy consumption ones.
