Ensembling Routing Algorithms

Energy and time consumption are the two most common challenges regarding communication optimization in the Wireless sensor network (WSN) routing algorithms. There are many different approaches for solving these two major issues in WSNs. These approaches optimize over time or energy separately and yet rarely take into account multiple criteria when choosing the next hop on the route. In order to address this open problem in WSNs, it is advantageous to combine multiple criteria during the next hop selection step. In this paper, we propose a new concept called Routing Ensembles. Our approach takes into account the individual selection criterion for the next hop using various multiple routing algorithms and applies a voting mechanism to decide on the next hop in the route collectively. We tested our routing ensemble with five well-known base routing methods, i.e., Greedy, NFP, MPoPR, MFR, and Compass and then selected the most voted node as the next hop. This process repeats until either the destination node is reached or the routing process fails. This concept was tested and validated using multiple simulated networks of varying sizes. The experimental analysis concluded that the routing ensembles allowed us to: (1) increase the delivery rate regardless of the network field size, (2) minimize both the energy and time consumption and (3) enhance the network performance only using traditional and directly routing algorithms.