Efficient Load Balancing and Data Aggregation Multipath Routing in Wireless Sensor Networks

Abstract

Wireless Sensor Networks (WSNs) comprise spatially distributed wireless sensors, small battery powered devices, which are deployed among geographical areas to collect the data continuously. The routing plays an important role in a network for deciding the path for transmitting the data between source and destination based on certain Quality of Service (QoS) parameters. In order to transmit the data in an energy efficient and reliable manner, multipath routing is the most viable routing mechanism. The energy of nodes near to sink drains quickly and thereby giving rise to network hole and hotspot problem and resulting into partitioning of network. If the traffic or load can be distributed over the multiple routes, then network hole problem can be avoided and energy dissipation of a network can be balanced. To ensure reliability and load balancing, the trade-off is needed between the level of redundancy and energy efficiency. As the energy dissipation for communication is significantly higher than the energy required for computation, the energy can be conserved by performing in-network data aggregation at intermediate nodes instead of sending the whole data to the Base Station individually. Therefore, to optimize the QoS parameters such as energy, delay and network lifetime etc., there is a need of multipath routing framework for energy efficient data aggregation and load balancing for WSNs. Currently, most of the data aggregation techniques maximize the energy efficiency without paying considerable attention to the data accuracy and varying conditions of underlying network topology. The level of data aggregation and frequency of data reporting should be adaptive with the varying density of the network and traffic pattern. The energy level of a node also gets affected due to collisions happening during the transmissions. These collisions should be prevented by applying protocols that are free of contention. While splitting and sending the data over the multiple paths, reliability and security should be ensured. Whenever the data is lost in between the source and the destination due to dead nodes, it should be recovered at the destination and data if captured by the malicious nodes, only a part of it iv Abstract should be revealed. Consequently, there is a need of multipath load balancing routing technique, which can assure the required level of Quality of Service (QoS) parameters and avoids hotspot problem. This work is carried out to develop load balancing and data aggregation multipath techniques to efficiently communicate the data based on energy efficient usage of network resources. A comprehensive investigation has been conducted to study various existing multipath routing techniques in WSNs for data aggregation and load balancing techniques. The load balancing through bio-inspired, nature inspired and other optimization techniques have been explored for WSNs. The investigations have been extended for multi-objective optimization in WSNs. A novel Dynamic Adaptive Hierarchical Data Aggregation (DAHDA) technique has been presented for uniform and non-uniform networks while maintaining the data accuracy. The aim of this proposed technique is to reduce the energy consumption of sensors and therefore, increase the network lifetime without critically affecting the data accuracy. In addition, the algorithm is able to handle sudden bursts in the underlying data by recording the data in the area of interest for the whole event duration. It introduces the concept of weighted sensors and density based clustering to decide nodes to be selected as CHs and nodes responsible for sending data at certain rounds. The assignment of weights to nodes is based on the residual energy and density. It is a newly developed measure to determine the percentage of its neighbor nodes to the total number of nodes. The algorithm includes the adaptivity feature, which can handle any sudden bursts in the underlying data values to continuously make sure that the data in the areas of interest is captured. Three variants namely DAHDA, Extended DAHDA (EDAHDA) and Modified EDAHDA are presented using the level of underlying functionalities. Proposed algorithms are compared with variants of LEACH. A Cross-layer Energy-efficient Clustering (CEC) technique is proposed for efficient contention free data aggregation in heterogeneous networks. In this technique, clusters of sensor nodes are formed in hexagonal shape. The cluster head is opted from the members of cluster itself on the basis of the ideal cluster head distance and remaining energy of sensor nodes having value greater than threshold value. In order to make a balance between consumption of energy and the network traffic, the rotation of cluster heads is performed. The energy level of a node gets affected due to collisions happening during the transmissions, which can be prevented by applying protocols that are free of contention. Slots are allocated Abstract v by cluster head to all member nodes within a cluster on the basis of their remaining energy so that nodes can switch to sleep mode. In the proposed technique, cluster head selection probability changes dynamically and data aggregation is performed by Cluster Heads based on the cost metric i.e. residual energy of the cluster head. The performance of proposed CEC technique is evaluated and compared with SOEECP, LCM and EEPCA. A Particle Swarm Optimization Energy Efficient Load Balancing (PSO-EELB) technique is proposed. Based on maximum residual energy, paths are selected and load balancing is performed by splitting and sending the data over the selected paths using network based coding. Deterministic PSO is utilized for faster convergence of the algorithm. The underlying network is based on the clustered architecture in which a Cluster Head is elected on the basis of fitness function within each cluster. The fitness function is defined in terms of distance and residual energy of a node. The proposed technique performs better than existing techniques in terms of various QoS parameters. In case of WSNs, there are scenarios when conflicting objectives are to be dealt. Based on the type of the application, the network scenario and required input/output of the problem, the type of optimization problem changes. Multi-objective Load Balancing Clustering (MLBC) technique has been proposed. Multi-Objective Particle Swarm Optimization (MOPSO), an evolutionary optimization approach, is utilized for accounting multiple objectives at a time. Two objective functions namely Energy Efficiency and Reliability have been considered simultaneously. Energy efficiency is measured in terms of residual energy of cluster head and reliability is measured in terms of packet delivery ratio. Weight is assigned to each node on the basis of residual energy and distance. The node having highest weight is selected as CH. A healing function is utilized in order to avoid loops in the generated path. The objective functions are evaluated for each individual. After, the completion of the execution, a set of non-dominating solutions called Pareto set is obtained. In order to choose the best compromised solution, fuzzy based approach is applied. The performance of the proposed MLBC technique is compared with techniques namely JPSO, MOPSO-DE and IMOWCA in terms of residual energy, packet delivery ratio and number of active nodes.