ODIF: Optimum Duty Cycle Selection Using Indirect Feedback for Neighborhood-Discovery in Crowded Scenarios

A significant amount of research has focused on minimizing the time it takes for two nodes to discover each other when operating in on-off states to conserve energy, a process known as neighbor discovery. However, less attention has been paid to the related issue of neighborhood-discovery, which involves multiple nodes attempting to find each other. We consider the latter problem to be quite relevant, especially with the increasing popularity of networks made up of multiple IoT devices. This issue is more complex because the likelihood of collisions increases with a greater number of nodes and as the time nodes remain awake (duty cycle) extends. The method we propose, called ODIF, uses indirect feedback (collisions and no-encounters) to estimate the number of nodes and employs a dynamic duty cycle algorithm to minimize either latency or energy expenditure during neighborhood-discovery. We compare ODIF with a randomized discovery protocol (referred to as Plain) and two other similar methods found in the existing literature. It was found that ODIF significantly reduces the neighbor discovery latency (or the expected number of attempts) compared to the baseline randomized neighbor discovery protocol for most values of the duty cycle. Similarly, ODIF achieved neighborhood-discovery 1 to 10 times faster for high node densities compared to similar protocols.