An adaptive path selection model for WSN multipath routing inspired by metabolism behaviors

In multipath routing of wireless sensor network (WSN), greedy path selection is always prone to cause path oscillation (frequent path changes) between each couple of sensor and sink nodes. To alleviate the side effect, we propose an adaptive path selection model (called a WSN path selection model based on the adaptive response by attractor selection (ARAS) model (WARAS)) inspired by metabolism behaviors of Escherichia Coli . The model consists of two main features. The first one is a new formula for a parameter called path-activity used to indicate adaptation goodness of multipath traffic transmission in dynamic network environments, which is inversely proportional to absolute value of difference between current path quality and best path quality. The second one is a novel attractor expression for attractors of multi-attractor equations to concretely specify stochastic effect of noise items in the equations on the path selection. Then, in an experimental WSN scenario composed of many source nodes and their shared neighbor nodes, we validate a dynamic-adaptive selection characteristic of the WARAS on distributing loads of the neighbor nodes. Subsequently, we design a path quality probe scheme in a multipath ad hoc on-demand distance vector routing (AODV) protocol. Compared with the greedy path selection through the path quality probe scheme, simulation results show the WARAS can perform better on reducing network delay and the path oscillation.