3.3 Energy Efficient Network Layer
3.3.1 Energy Aware Routing
Limited battery capacity has brought energy aware routing protocols to research atten- tion for the last few decades. The traditional shortest path criterion might not be the best solution as some intermediate wireless hosts may not be operational anymore if they are frequently picked by routing protocols or have low battery capacity. Therefore route se- lection algorithms have taken into consideration the energy consumption of each possible route for packet delivery.
Minimum Total Transmission Power Routing (MTTPR) [146] calculates the total energy consumption of each path when selecting routes. The information of energy con- sumption at each host is obtained and added up in MTTPR, and the path that consumes the least energy is chosen for packet forwarding.
Besides total energy consumption, some other metrics are evaluated such as battery capacity and remaining energy level. To prolong the life of the network as a whole, Min- imum Battery Cost Routing (MBCR) [147] considers the remaining energy as the most important factor of routing algorithm. MBCR calculates the remaining energy of a whole path which is the sum of remaining energy of each hop and uses the path with device along it of most remaining energy. However, MBCR evaluates the energy of a path as a whole without considering individual nodes. Frequent choosing of a node as part of routing path may lead to quicker energy depletion and inactivity of part of or even the whole network death. MBCR avoids the hops with less energy to balance the remaining energy level of whole network and maximize the lifetime of whole network. Conditional Min-MBCR (CMMBC) [147] on the other hand combines these two solutions where both the remain- ing energy of a host and the total energy of a path are used for route selection.
Energy Aware Routing (EAR) [12] is a solution that normalizes both the value of energy consumption and residual energy level to achieve better energy distribution among nodes within a network. Each node obtains the processing energy value and remaining energy of its neighbours and a probability is given to each neighbour. Higher probability is assigned if the neighbour has more energy left and requires less energy for packet process-
ing and forwarding. Neighbouring hosts with higher priorities are chosen to forward data as part of route.
Some of the routing protocols are based on centralized algorithms where the energy in- formation is gathered at a host before the packet is forwarded, while other works distribute the decision of whether to participate in data transmission is made by the intermediate hops themselves. In Span [148], the available energy as well as the contribution of a host is evaluated by intermediate hosts to decide whether they should sleep or be active. The contribution of a host is a metric that refers to the number of pairs of hosts a node can help connect. The distributed feature of Span decreases routing overhead and increases scalability.
Hierarchical based routing protocols such as Low Energy Adaptive Clustering Hi- erarchy (LEACH) [13] [149] have been proposed to conserve energy. In this scenario, a host does not need to talk directly to the base station to exchange information about how far the node is from the base station. Instead, the network is divided into clusters where a host is elected as the head to communicate with the base station, and the remaining nodes only need to talk to the head which is closer than the central station which is more en- ergy efficient. Power-Efficient GAthering in Sensor Information Systems (PEGASIS) [150] [151] saves more energy by forming chains within a cluster so that each node only communicates with two neighbours which are its previous and next nodes in a chain. How- ever the cluster-based solutions have two obvious drawbacks. First, the battery of elected head tends to be drained as it is required to exchange information between all the remaining nodes within the cluster and the base station. Second, energy efficiency is achieved through short distance communication and the mechanism might not be effective when deployed in wide area. The problem is address in [169], where a multi-tier solution is proposed and the role of cluster head is dynamically chosen. In this solution, quick energy depletion is avoided as the role of head is rotated and short distance communication is guaranteed by introducing multiple tiers in clusters.
routing path as it is more energy efficient to communicate with hosts that are closer as in cluster-based solutions and also as better performance is achieved if a packet is forwarded in the direction towards its destination. In Geographical and Energy Aware Routing (GEAR) [153] a path is selected from source to the target area which consists of a small range of hosts including the destination device before the packet is transmitted within the area to the exact destination. GEAR first defines the target area which is the preliminary direction for the routing path to follow as a set of hosts including the destination host, and then a refined routing algorithm is applied to deliver the packet to its recipient. Minimum Energy Communication Network (MECN) [154] and Geographic Adaptive Fidelity (GAF) [155] make use of GPS to track the address of each hosts location. The address information in GAF is used to divide the area into grids and each host is fitted into a grid. The hierarchical routing method is also applied in GAF where a head node is elected within each grid as a central host to forward data to other hosts.
Traditional geographic-based routing protocols may lead to control overhead caused by dissemination of beacon data which is used to maintain address information within a network. In order to address this problem, Energy-efficient Beacon-less Geographic Routing protocol (EBGR) [156] adopts a localized solution where a localized routing decision is made only when the host has data to send. The choice of path is based on calculation of the ideal position of the next hop based on the direction of sink and energy efficiency of path, and then the neighbour that is closest to the ideal position is chosen as the next hop with respect to the upper bound of distance and energy consumption.
Protocols such as [152] [170] differentiate devices according to their energy features as part of their routing algorithms. As differently devices may join a network with different features and energy requirements, Energy-oriented Node Characteristics-Aware Routing Algorithm (ENCARA) [170] evaluates both the hardware and software specifications of each device to prolong the battery life of energy critical hosts. Energy Type Aware Rout- ing [152] categorized devices to senders and receivers. Devices that send packets most of the time are considered active senders and are not preferred as part of routing path as these hosts tend to generate traffic and require more energy transmitting and processing data.