Memory Matters: Bumblebee Behavioral Models for Vehicle-to-Vehicle Communications

Vehicles forming connected communication networks are routinely challenged with the complex decision problem of either staying with the same wireless channel or moving to a different wireless channel when experiencing highly variable channel quality conditions. In order to obtain a practical solution to this problem, we refer to bumblebee behavioral models, which possess evolved decision-making mechanisms to adaptively solve similar problems while foraging in environments containing multiple floral resources (channels). In order to enable vehicles to adapt to these time-varying channel conditions, we propose in this paper a bumblebee-inspired decision-making algorithm in which channel quality information is stored and updated in vehicle memory. This information is used to estimate qualities of channel options and then weighed against switch costs to determine optimal channel selection. We incorporated our algorithm into a VDSA-based VANET model, while the GEMV2 Vehicle-to-Vehicle (V2V) propagation simulator was used to test its performance under different memory parameters and against existing models. Our results show that a memory system based on the averaging of stored channel information dramatically increased channel selection performance over a memoryless system in both urban and highway scenarios by 52% and 20%, respectively.