Creating small ad hoc networks: Swift presence notification strategies

Abstract

We consider the problem of k agents using a communication channel restricted to n transmission slots. Each agent aims to broadcast its message repetitively so that at least one copy of it avoids collision, where the agent cannot observe channel status, and there is no a priori information about other agents except their approximate number.Motivated by time-sensitive vehicle coordination use cases, we aim to indicate a better slot selection distributed strategy for achieving all-agent communication success. We analyze two strategies. With the default Bernoulli Trials (BT) strategy, an agent decides to transmit each slot independently with a fixed probability p. With the t-Slots strategy, an agent chooses a set of t slots to transmit.We derive formulas for the probability of all-agent communication success and show how to determine the optimal setting of p and t. We provide strong arguments that not only the t-Slots strategy significantly outperforms BT (in some practical setups, it is more than twice better), but is also more resilient to suboptimal configurations. Moreover, t-Slots reaches its peak performance for a lower average number of transmissions per agent, enabling even 40% energy consumption savings in the practical setup.