Fuzzy-Based Multi-Layered Clustering and ACO-Based Multiple Mobile Sinks Path Planning for Optimal Coverage in WSNs

Wireless Sensor Network with Mobile Collectors (MCs) enable increased flexibility and convenience in data gathering for numerous large-scale applications. However, introducing MCs also brings a new set of challenges to overcome. To reduce the data delivery latency of the application, it is required to select the minimum number of Rendezvous Points (RPs) that allow most sensors to forward data in single-hop and best path must be planned for each MC to provide uniform path length and round-trip time for all MCs. In contrast to existing schemes, we propose a Fuzzy C-Means based multi-layered RP Clustering and ACO-based Route-Planning scheme (FCM-RP) which is a robust method to determine RP-positions and MC-assignments. More particularly, the existing works have given priority to either reducing network energy consumption or minimizing data gathering delay. In our paper, along with these priorities, we consider the robustness and adaptability of the algorithm; meaning that, our algorithm can modify the planned trajectories of deployed MCs in response to sensor node failures. This allows to adapt to any changes in network topology caused by node failures or external factors. Also, to maximize utilization of each RP and provide greater coverage for nodes, the ideal positioning of RPs with minimum coverage overlap with neighboring RPs is considered. Simulation analysis shows that the FCM-RP scheme defines better routes for MCs in terms of total path length and delay, and significantly outperforms the existing algorithms in terms of total energy consumption and network lifetime.

I. Introduction

A wireless sensor actuator network is made up of small battery-powered sensor nodes that monitor their physical environment and relatively resource-rich entities called actuators that act upon the received data from sensor nodes, supporting a wide range of applications in large scale networks via computation and control [1]–[3]. Traditionally, the sink node would be in a fixed location and all sensors would transmit their data via multi-hop routing. However, the use of multi-hop routing requires the nodes to expend much of their energy to relay information from other nodes that are far from the sink, leading to uneven energy distribution among the nodes. The nodes closer to the sink consume more energy in data forwarding and thus die out early. This can cause the network to become disconnected even though most nodes are still operational, which is sometimes referred to as the sink hole problem [4].