A Population Protocol Model with Interaction Probability Considering Speeds of Agents

Abstract

This paper proposes a new extension of the population protocol (PP) model, the linearly-weighted interaction population protocol (LIPP) model, which introduces weights of agents (or mobile devices) as abstract speeds of agents. The model assumes that the interaction probability between agents is relatively proportional to the weights of the agents, which is almost validated from preliminary simulation results. Each agent can control its weight to adjust its interaction probability. This implies that mobility of agents is semi-passive (not completely passive) since they can change only their abstract speeds. This paper considers how the expected convergence time (measured by the number of interactions) of naive PP protocols for information dissemination, leader election and majority can be improved in the new model by assigning appropriate weighs to agents. The presented results show potential possibility and limitation of the LIPP model.