A Multi-objective Ant Colony Optimisation-based Routing Approach for Wireless Sensor Networks Incorporating Trust

In the near future, Wireless Sensor Networks (WSNs) are expected to play an important role for sensing applications, in the civilian as well as in the military sector. WSNs are autonomous, distributed, self-organised networks consisting of multiple sensor nodes. Usually, the limited radio range of the nodes, arising from energy constrains, is overcome by the cooperation of nodes. As the Combinatorial Optimisation Problem (COP) of routing is computationally hard, often approximation algorithms are preferred, which are capable of finding near optimal solutions within polynomial time. A simple but robust way of solving the routing COP is the application of Ant Colony Optimisation (ACO)-based routing algorithms. When multiple (conflicting) objectives should be considered, ACO algorithms can be extended to Multi-objective Ant Colony Optimisation (MOACO) algorithms that are capable of considering multiple objectives at the same time within the optimisation process. Normally, the routing in WSNs is susceptible to adversaries due to their deployment in unattended or in hostile environments. Particularly, attacks from compromised nodes (insider attacks) are a severe problem in WSNs. As insider attacks cannot be alleviated by classical security measures, often soft security measures (trust and reputation) are applied to mitigate the impact of these attacks. In this thesis, the idea of using trust as security measure against insider attacks is seized and interweaved with an MOACO-based routing approach. The Multi-objective Ant Colony Optimisation Routing Framework for WSNs (MARFWSN) is developed, a routing framework for WSNs that provides an interface for the docking of MOACObased algorithms that can be used for the routing. Different MOACO-based algorithms – including ASMOACO, MMASMOACO, ACSMOACO and SRMOACO – are implemented and tested, considering multiple objectives incorporating trust. In the first step, several pre-experiments are conducted to find the best input parameters for the MOACO-based WSN routing algorithms. Subsequently, further experiments aim for comparing the MOACO-based routing algorithms to the well-known Dynamic Source Routing (DSR) protocol regarding several performance criteria. Different topologies and traffic patterns are taken into account to obtain heterogeneous results, depending on the specific application area of the tested routing algorithms. The simulation results show that the MOACO-based routing algorithms perform similarly compared to the DSR-based protocol, implicating the possibility for their utilisation in WSNs. The additional benefit of the mitigation of insider attacks for the MOACO-based algorithms, though result in a slightly higher routing overhead and end-to-end delay.