The power-law distributions of contact durations and waiting intervals of all pairs of agents together, dubbed as bursts of aggregate pairs, have been found in many realistic opportunistic networks. Many models for such opportunistic networks are under a homogeneous assumption that the distribution from every pair of agents performs the same format as that from all pairs of agents. But our empirical study shows an opposite scenario, e.g., each agent performs Poissonian dynamics while all pairs perform burst dynamics. Therefore, we propose a bottom-up human mobility model to replicate this opportunistic network, and prove that bursts of aggregate pairs can be rooted from individual independent Poissonian self-activity and preferential mobility. Finally we validate our findings with numerical simulations.