Network connectivity is paramount for multi-agent systems with limited sensing and communication capabilities, since agents need to coordinate and communicate to make appropriate decisions in formation control. The goal in this work is to steer a group of agents to a desired configuration from any given initially connected graph in a decentralized manner, without partitioning the underlying network, and avoiding collision with other agents and moving obstacles. To maintain network connectivity, an information flow is proposed to specify the communication among agents. The underlying network graph is connected as long as all the information flows are maintained. Based on the approach of information flow, each agent is able to choose a short path to reach the desired agent in the information graph by dynamically building new communication links or breaking existing links. A navigation function formulation is used to maintain the information flow among agents, and guarantee the convergence of the system to the desired configuration.